DARPA, at least, has a clue
Robert Myers
https://www.fbo.gov/download/63c/63cbc206740fea482de76e6bf5e9414c/SN_09-46_UHPC_RFI_24_June_2009.pdf
<quote>
-New system-wide technology approaches specifically including hardware
and software co-design to minimize energy dissipation per operation
and maximize energy efficiency, with a 50GFLOPS per watt goal, without
sacrificing scalability to ultra-high performance DoD applications -
efficiency.
-New technologies and execution models that do not require application
programmers to explicitly manage system complexity, in terms of
architectural attributes with respect to data locality and
concurrency, to achieve their performance and time to solution goals -
programmability.
-Technology that will manage hardware and software concurrency,
minimizing overhead for thousand- to billion-way parallelism for the
system-level programmer.
-A system-wide approach to achieve reliability and security through
fault management techniques enabling an application to execute
correctly through both failures and attacks.
Current processing systems are grossly power-inefficient and typically
deliver only a small fraction of peak performance. Until recently,
advances in Commercial Off-The-Shelf systems performances were enabled
by increases in clock speed, decreases in supply voltage, and growth
in transistor count. These technology trends have reached a
performance wall where increasing clock speed results in unacceptably
large power increases, and decreasing voltage causes increasing
susceptibility to transient and permanent errors, DARPA stated.
</quote>
Three-dimensional chips, microfluidic cooling, wired for IBM.
Anything else? Any other potential players? Give all the money to
LLNL and LANL? Not a chance. DARPA too smart for that, I hope.
Robert.
MitchAlsup
A link from your link one layer indirect:
http://users.ece.gatech.edu/~mrichard/ExascaleComputingStudyReports/exascale_final_report_100208.pdf
Their best current building block is on the order of 0.2-0.3 GF/Watt
(Blue Gene).
In addition concurrency scaling is limiting the systems to the 1.5 GF/
Watt level in 2023.
Overall, it might be easier to up the data center power limits from 20
MW and just build the thing besides a power plant.
Mitch
Robert Myers
> Looks like they have a loooonnnngggg way to go.
>
> A link from your link one layer indirect:
>
>
> Their best current building block is on the order of 0.2-0.3 GF/Watt
> (Blue Gene).
> In addition concurrency scaling is limiting the systems to the 1.5 GF/
> Watt level in 2023.
>
> Overall, it might be easier to up the data center power limits from 20
> MW and just build the thing besides a power plant.
>
lunacy of government stewardship of research in computing; viz., that
passively waiting for the wizardry inherent in the industry was not
only good enough but probably better than anything that aggressively
funded research could come up with.
This kind of timidity has been creeping into public management of
RDT&E ever since 1969, when we landed a man on the moon.
I have to admit that, for the last two decades, the reliance on COTS
hardware has proven to be an amazingly good bet. With the limits that
are now upon us and that, as far as I know, first started appearing at
90nm, we are either at the end of the road or can see it clearly.
If there is to be a DARPA, then it has to do something better than
building bigger power plants and buying more real estate, which is
what big HPC projects have largely been relying on. If nothing else,
speed of light considerations limit what you can accomplish simply by
scaling up. In addition, if the cost per gigaflop can't be driven
down further, then what we will ever accomplish through computation is
going to run into a brick wall.
The bisection bandwidth per flop of the current generation of most
computational monsters is unacceptable for some important
calculations--a reality that reliance on linpack benchmarks
conveniently ignores.
I don't know where DARPA got its goals. I'd prefer that they be
unrealistic. I don't care what they reach for so long as it is risky
and ambitious. If you can make a business case for your research
program (goals are attainable with reasonable ROI) there is no reason
for DARPA to be funding it.
Robert.
nm...@cam.ac.uk
Robert Myers <rbmye...@gmail.com> wrote:
>On Jun 28, 7:30=A0pm, MitchAlsup <MitchAl...@aol.com> wrote:
>>
>> Overall, it might be easier to up the data center power limits from 20
>> MW and just build the thing besides a power plant.
Which brings in another problem, about which people may have read :-(
>The good news here, for me, is that someone has woken up from the core
>lunacy of government stewardship of research in computing; viz., that
>passively waiting for the wizardry inherent in the industry was not
>only good enough but probably better than anything that aggressively
>funded research could come up with.
Well, maybe. I agree that was and is lunatic.
>This kind of timidity has been creeping into public management of
>RDT&E ever since 1969, when we landed a man on the moon.
Actually, it started before, but it became dogma after about 1970.
Both in the USA and UK.
>I have to admit that, for the last two decades, the reliance on COTS
>hardware has proven to be an amazingly good bet. With the limits that
>are now upon us and that, as far as I know, first started appearing at
>90nm, we are either at the end of the road or can see it clearly.
I disagree. It has proved an amazingly BAD bet! What it has done is
to lead to incredible advances in some aspects, leading to other
critical aspects being completely ignored, and heading towards an
entirely predictable dead end.
The prime purpose of government is social engineering, and what
should have been done (here as in so many other areas) is to guide
the direction and ensure that alternatives were not eliminated.
And, please, Oh Mad Monetarists, I said "guide" not "constrain";
the latter is normally justified only to avoid overall harm or
unacceptable risk.
A huge amount could have been done, for negligible cost, and much
could still be done, cheaply. Unfortunately, changes can be any two
of radical, cheap and quick - but never all three.
Regards,
Nick Maclaren.
Paul Wallich
> Looks like they have a loooonnnngggg way to go.
>
> A link from your link one layer indirect:
>
> http://users.ece.gatech.edu/~mrichard/ExascaleComputingStudyReports/exascale_final_report_100208.pdf
>
> Their best current building block is on the order of 0.2-0.3 GF/Watt
> (Blue Gene).
> In addition concurrency scaling is limiting the systems to the 1.5 GF/
> Watt level in 2023.
Hence the HW/SW codesign blather from DARPA, I guess. Does anyone have
specs for the kinds of algorithms that can be profitably burned into
real silicon? That would provide a pretty good upper bound.
paul
nm...@cam.ac.uk
Paul Wallich <p...@panix.com> wrote:
>MitchAlsup wrote:
>> Looks like they have a loooonnnngggg way to go.
>
>specs for the kinds of algorithms that can be profitably burned into
>real silicon? That would provide a pretty good upper bound.
No, but we have a pretty good idea of why that won't help.
To a first approximation, the major gains by burning into silicon
rather than emulating come when you are doing bit-munging and
similar things that can be described simply, but which are very
hard to map into 'RISC' instructions. Consider decoding an IEEE
754 format floating-point number using integer operations, for
example.
The gains from all such areas are constant - i.e. they are a simple
multiplicative factor and do not scale with the number of processors.
No such gains will solve DARPA-scale problems.
The only changes that might help are at the design level, and those
can be implemented as well in software as in hardware. All right,
some need hardware support, either for the reasons given above or
because the existing primitives are just plain inadequate. But
that doesn't change the fundamental point.
Regards,
Nick Maclaren.
Robert Myers
> In article <h2ambv$5u...@reader1.panix.com>,
> Paul Wallich <p...@panix.com> wrote:
>
> >MitchAlsup wrote:
> >> Looks like they have a loooonnnngggg way to go.
>
> >Hence the HW/SW codesign blather from DARPA, I guess.
Why do you see it as blather?
Once you manage to shut entrenched software high priests up long
enough that anyone else can say anything, two things become apparent:
1. There is little point in the DoD, whose needs change overnight,
building hardware that can be programmed only with great difficulty by
hard-to-find software specialists with a very high skill level, with
the resulting software almost unmaintainable, except possibly by
similar elite cadres.
2. Software design methodologies and languages currently in use make
most software highly resistant to extreme optimization and
artificially limit the usefulness of energy-efficient approaches like
stream computation. Adapting to the installed base makes the problem
hopeless. Time to throw the installed base overboard and start with a
clean sheet of paper.
Robert.
Robert Myers
> In article <b4e4b41c-6aff-4cd4-93f9-e655a6ecc...@j32g2000yqh.googlegroups.com>,
> Robert Myers <rbmyers...@gmail.com> wrote:
> >This kind of timidity has been creeping into public management of
> >RDT&E ever since 1969, when we landed a man on the moon.
> Actually, it started before, but it became dogma after about 1970.
> Both in the USA and UK.
That the USA and the UK should follow similar paths is hardly
surprising, given the historically close coordination in matters of
national security. In the USA, there was a confluence of events: we
actually did put a man on the moon and it was apparent that we held a
huge technological lead over any Soviet threat, the relationship
between universities and the defense establishment became hostile as a
result of the Viet Nam war experience, and the boom of the sixties was
replaced by the stagflation of the seventies. The urgency was gone,
the DoD wanted to cut back and pull what work remained into captive
labs, and the Federal government was rubbing the buffaloes off the
nickels as it counted what it was spending.
My memory of history aligns with your comment about the timidity
having deep roots. The worry about money had started almost as soon
as Kennedy had put the man on the moon proposition out there, and
there was an effort to wiggle out of that commitment without a loss of
face by trying to arrange something dramatic with the Soviets. Once
we had ICBM's that worked reliably, there was nothing much left to be
urgent about (except for SSBN's as a plausible and invulnerable MAD
response, which is another story, entirely).
> >I have to admit that, for the last two decades, the reliance on COTS
> >hardware has proven to be an amazingly good bet. With the limits that
> >are now upon us and that, as far as I know, first started appearing at
> >90nm, we are either at the end of the road or can see it clearly.
> I disagree. It has proved an amazingly BAD bet! What it has done is
> to lead to incredible advances in some aspects, leading to other
> critical aspects being completely ignored, and heading towards an
> entirely predictable dead end.
It's been a good bet from my perspective in that everything I ever
could have dreamed of can now be had for a song. That just proves
that my imagination was limited. On the other hand, government
investments don't seem to have paid off (Cray MTA, Thinking Machines,
for example). Of course, the fact that government research hasn't
paid off just might be related to the fact that you have to prove the
idea will work before doing the "research." That is to say, the
government now funds proof-of-principle work, but not genuine
research.
> A huge amount could have been done, for negligible cost, and much
> could still be done, cheaply.
Like what, for example?
Robert.
nm...@cam.ac.uk
Robert Myers <rbmye...@gmail.com> wrote:
>
>> >I have to admit that, for the last two decades, the reliance on COTS
>> >are now upon us and that, as far as I know, first started appearing at
>> >90nm, we are either at the end of the road or can see it clearly.
>
>> to lead to incredible advances in some aspects, leading to other
>> critical aspects being completely ignored, and heading towards an
>> entirely predictable dead end.
>
>It's been a good bet from my perspective in that everything I ever
Not everything that I have dreamed of - nor everything that I have
actually used! Provably correct systems (albeit only against gross
memory errors and security breaches)? Nope. Massively parallel
desktops (256-way and up)? Nope. Soft desktops, to allow amateurs
to experiment with architectures? Nope. 'Sea of threads' parallel
designs? Nope.
Note that all of those have been delivered, though not on desktops
and not affordably. None are unrealistic.
> That just proves
>that my imagination was limited. On the other hand, government
>investments don't seem to have paid off (Cray MTA, Thinking Machines,
>for example). Of course, the fact that government research hasn't
>paid off just might be related to the fact that you have to prove the
>idea will work before doing the "research." That is to say, the
>government now funds proof-of-principle work, but not genuine
>research.
It could also be because governments have limited imaginations, and
are controlled by demagogues and bureaucrats, not engineers. But
your last point is definitely right, too.
>> A huge amount could have been done, for negligible cost, and much
>> could still be done, cheaply.
>
>Like what, for example?
Like the USA (or even UK) government stating that central procurements
will specify a 10% reduction on power (watts) year-on-year, until we
get down to 50 watts per user.
Regards,
Nick Maclaren.
Rick Jones
> Like the USA (or even UK) government stating that central
> procurements will specify a 10% reduction on power (watts)
> year-on-year, until we get down to 50 watts per user.
That sounds more like an arbitrary mandate than a research investment.
Given there are < 50 Watt CPUs now, all it takes is a willingness to
mandate a ban > 50 Watt processors for personal use, rather like
banning incandescents or plasma TVs.
rick jones
--
No need to believe in either side, or any side. There is no cause.
There's only yourself. The belief is in your own precision. - Joubert
these opinions are mine, all mine; HP might not want them anyway... :)
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...
nm...@cam.ac.uk
Rick Jones <rick....@hp.com> wrote:
>nm...@cam.ac.uk wrote:
>> Like the USA (or even UK) government stating that central
>> procurements will specify a 10% reduction on power (watts)
>> year-on-year, until we get down to 50 watts per user.
>
>That sounds more like an arbitrary mandate than a research investment.
Actually, I was referring just to public procurements.
>Given there are < 50 Watt CPUs now, all it takes is a willingness to
>mandate a ban > 50 Watt processors for personal use, rather like
>banning incandescents or plasma TVs.
Total bans are much harder than constraints - and, paradoxically,
less effective.
Also, you say that there are low-powered CPUs. Have you TRIED to
buy a low-powered desktop recently? I have. I couldn't find one,
at an even remotely affordable price - the point of my condition
is to ensure that they become available, and development is slanted
towards that direction.
Regards,
Nick Maclaren.
Rick Jones
> Also, you say that there are low-powered CPUs. Have you TRIED to
> buy a low-powered desktop recently?
Nope, I prefer that my desktop actually perform :) Rather ironic in
today's performance context at least given what if anything will
appear in the headers of this post as to the conditions under which my
character mode news reader were running.
> I have. I couldn't find one, at an even remotely affordable price -
> the point of my condition is to ensure that they become available,
> and development is slanted towards that direction.
Is it too great a stretch to call a laptop with a docking station a
desktop these days?
rick jones
--
oxymoron n, commuter in a gas-guzzling luxury SUV with an American flag
Robert Myers
> In article <h2bcpg$de...@usenet01.boi.hp.com>,
> Rick Jones <rick.jon...@hp.com> wrote:
>
> >n...@cam.ac.uk wrote:
> >> Like the USA (or even UK) government stating that central
> >> procurements will specify a 10% reduction on power (watts)
> >> year-on-year, until we get down to 50 watts per user.
>
> >That sounds more like an arbitrary mandate than a research investment.
>
> Actually, I was referring just to public procurements.
>
> >Given there are < 50 Watt CPUs now, all it takes is a willingness to
> >mandate a ban > 50 Watt processors for personal use, rather like
> >banning incandescents or plasma TVs.
>
> Total bans are much harder than constraints - and, paradoxically,
> less effective.
>
> Also, you say that there are low-powered CPUs. Have you TRIED to
> buy a low-powered desktop recently? I have. I couldn't find one,
> at an even remotely affordable price - the point of my condition
> is to ensure that they become available, and development is slanted
> towards that direction.
>
not the least of which is that it would punish some of the last
funders of cutting edge research; e.g., hard core gamers.
Something like the fleet fuel efficiency averages of US CAFE standards
would work pretty well. Only foot-dragging by Detroit (labor and
management) kept Federal CAFE standards from working even better than
they have.
It's a happy accident that energy policy and climate policy both push
in the direction the industry needs to go, but I'd hate to see public
policy mandates used as a substitute for more courageous and more
aggressively-managed publicly-funded research, much of which needs to
be wrested out of the hands of bureaucrats with a hotline to the PR
department. I don't think the Manhattan project needed a single press
release--unlike the acres of dead trees produced by every new "super"-
computer.
Robert.
Andrew Reilly
> 1. There is little point in the DoD, whose needs change overnight,
> building hardware that can be programmed only with great difficulty by
> hard-to-find software specialists with a very high skill level, with the
> resulting software almost unmaintainable, except possibly by similar
> elite cadres.
I don't get that. Why should the DoD require their systems to be built
by monkeys? NASA uses elite specialists to design their space systems,
large-scale civil works are designed by elite specialists. That's how
good stuff works. Of course, there's nothing wrong with having programs
to develop tools to help these guys, that's just common sense that
applies in all engineering fields. Also no point calling for tenders for
systems that rely on luck or magic. Maybe that's what you were saying.
> 2. Software design methodologies and languages currently in use make
> most software highly resistant to extreme optimization and artificially
> limit the usefulness of energy-efficient approaches like stream
> computation. Adapting to the installed base makes the problem hopeless.
> Time to throw the installed base overboard and start with a clean sheet
> of paper.
Doesn't that happen anyway? With the possible exception of a few dusty-
deck numeric libraries, it seems to me that most code (perhaps outside of
banking) is rewritten every ten years or so anyway, just because that's
easier for the new hires than talking to the old guys who wrote the last
version. How much code are we talking about in this installed base, and
who does it belong to?
A large, related problem, I think, is that quite a lot of research now
seems to finish with code and graphs: the actual workings never find
their way into published papers in english and maths. That's going to be
a big problem for those who would otherwise write new versions using new
tools, later.
I.e., I think that the installed base argument is mostly a red herring.
The new and interesting (research) problems are already being solved with
the new and interesting tools. The new tools don't solve all of the
problems, of course, but useful progress seems to be being made. (Check
out http://blip.tv/file/812787 if you have a spare 152-odd minutes to get
an overview of how immutable-by-default data structures, and STM are
useful for concurrency in Clojure. There are plenty of dissenting views
out there, of course...)
--
Andrew
Paul Wallich
> On Jun 29, 12:20 pm, n...@cam.ac.uk wrote:
>> In article <h2ambv$5u...@reader1.panix.com>,
>> Paul Wallich <p...@panix.com> wrote:
>>
>>> MitchAlsup wrote:
>>>> Looks like they have a loooonnnngggg way to go.
>>> Hence the HW/SW codesign blather from DARPA, I guess.
>
> Why do you see it as blather?
I see it as blather because people have been talking hardware/software
codesign for 30 years or so that I know of, and probably before then as
well. It's not really a falsifiable thing to say.
> 2. Software design methodologies and languages currently in use make
> most software highly resistant to extreme optimization and
> artificially limit the usefulness of energy-efficient approaches like
> stream computation. Adapting to the installed base makes the problem
> hopeless. Time to throw the installed base overboard and start with a
> clean sheet of paper.
Now there are the beginnings of some ideas that could actually be funded
and tested.
One of the things I'd love to see attempted in this area would be pretty
much the opposite of what DARPA is apparently asking for, namely analog
systems. Lots of those flops are being burned calculating things that
circuits do already (and that digital designers have spent hundreds of
thousands of engineer-years getting them to stop doing). Could be
impossible, but much of what I used to see suggested that the
impossibility was on the cognitive rather than the hardware side.
paul
Stephen Fuld
snip
> Something like the fleet fuel efficiency averages of US CAFE standards
> would work pretty well.
No, they don't, or at least didn't. We have experience with these and
they don't reduce gasoline consumption as much as you would think. The
reason is that given higher MPG cars, the price of making a trip
declines and, surprise, surprise, people drive more. They move farther
away from work, make more trips, etc.
The most reliable way to decrease the usage of something is to raise its
price. When gasoline was $4.00 a gallon, you couldn't buy a Toyota
Prius. When it dropped to about $2.30, you couldn't sell one.
BTW, my proposal is to create a revenue neutral tax swap. Increase the
tax on gasoline say 30 cents per gallon each year for the next 10 years.
Take 100% of that money and refund it to the people via say a per
capita refundable tax credit. This means that people will know that gas
prices will go up which allows them to plan when they buy their next
car, etc. It also taxes those people who use more than the average
amount of gas, but these are the people who would gain the most from
taking actions such as buying a more fuel efficient car. The policy
rewards those who use less than the average amount of gas as they
actually make money on the deal.
A slight modification, which may be useful is to make the tax on only
the petroleum portion of the gas. Thus if we used say 10% bio-materials
in the gas, the tax would be reduced by 10%. By doing this, we would
incent the usage of things like ethanol, and we could phase out the
ethanol subsidy we currently pay to farmers, thus actually reducing
government expenditures.
> It's a happy accident that energy policy and climate policy both push
> in the direction the industry needs to go,
Not necessarily. Reducing petroleum usage favors things like electric
cars and plug in hybrids, but this increases electricity demand. In the
US, most electricity is generated from coal, so we have to be careful
not to increase emissions due to burning more coal. (yes, I know that
overall, it is still a win, but it isn't totally simplistic.) Also,
people are talking about hydrogen cars, but the most common way to make
hydrogen is from natural gas. This stuff is very tricky.
--
- Stephen Fuld
(e-mail address disguised to prevent spam)
Robert Myers
users.org> wrote:
> On Mon, 29 Jun 2009 13:08:47 -0700, Robert Myers wrote:
> > 1. There is little point in the DoD, whose needs change overnight,
> > building hardware that can be programmed only with great difficulty by
> > hard-to-find software specialists with a very high skill level, with the
> > resulting software almost unmaintainable, except possibly by similar
> > elite cadres.
>
> I don't get that. Why should the DoD require their systems to be built
> by monkeys? NASA uses elite specialists to design their space systems,
> large-scale civil works are designed by elite specialists. That's how
> good stuff works. Of course, there's nothing wrong with having programs
> to develop tools to help these guys, that's just common sense that
> applies in all engineering fields. Also no point calling for tenders for
> systems that rely on luck or magic. Maybe that's what you were saying.
>
people with all kinds of skill levels, almost no matter what kinds of
controls are put into place. If you're doing cutting edge science or
engineering (that's what the code is for), then you need a cutting-
edge scientist or engineer. If, on top of that, you need a software
wizard to make the most trivial of changes safely, perhaps to respond
to some real-time event (a damaged shuttle circling the earth?), you
may be out of luck.
Software shouldn't be written by monkeys, but it should be written in
a way that monkeys would have to work hard to do serious damage. That
describes almost no serious code that I've come across, and anyone in
the business has to have seen a software train wreck brought on by
someone with good intentions who doesn't understand what they're
doing.
Even wizards have a hard time coding concurrent processes safely,
witness some of the scary discussions in this forum.
As to luck, almost all software depends on it, so far as I can tell,
since the only software I know of that is routinely formally verified
is flight control software.
> > 2. Software design methodologies and languages currently in use make
> > most software highly resistant to extreme optimization and artificially
> > limit the usefulness of energy-efficient approaches like stream
> > computation. Adapting to the installed base makes the problem hopeless.
> > Time to throw the installed base overboard and start with a clean sheet
> > of paper.
>
> Doesn't that happen anyway? With the possible exception of a few dusty-
> deck numeric libraries, it seems to me that most code (perhaps outside of
> banking) is rewritten every ten years or so anyway, just because that's
> easier for the new hires than talking to the old guys who wrote the last
> version. How much code are we talking about in this installed base, and
> who does it belong to?
>
Eugene Miya would be the one to give an answer to that. "Installed
base" is his phrase, at least in this forum.
> A large, related problem, I think, is that quite a lot of research now
> seems to finish with code and graphs: the actual workings never find
> their way into published papers in english and maths. That's going to be
> a big problem for those who would otherwise write new versions using new
> tools, later.
The problem is money. The pressure to get out a colored graph with
the right answer is overwhelming. Failure to do so can be hazardous
to funding. Advancing the common pool of knowledge, creating code
that can be understood by future generations, or creating a solid
foundation for future work all have low priorities because no one gets
a pay raise for such contributions.
> I.e., I think that the installed base argument is mostly a red herring.
> The new and interesting (research) problems are already being solved with
> the new and interesting tools. The new tools don't solve all of the
> problems, of course, but useful progress seems to be being made. (Check
> outhttp://blip.tv/file/812787if you have a spare 152-odd minutes to get
> an overview of how immutable-by-default data structures, and STM are
> useful for concurrency in Clojure. There are plenty of dissenting views
> out there, of course...)
I will find the spare minutes.
Robert.
Rick Jones
asserting that DARPA has no clue :)
rick jones
--
a wide gulf separates "what if" from "if only"
MitchAlsup
> In article <h2bcpg$de...@usenet01.boi.hp.com>,
> Rick Jones <rick.jon...@hp.com> wrote:
>
> >> Like the USA (or even UK) government stating that central
> >> procurements will specify a 10% reduction on power (watts)
> >> year-on-year, until we get down to 50 watts per user.
>
> >That sounds more like an arbitrary mandate than a research investment.
>
> Actually, I was referring just to public procurements.
>
> >Given there are < 50 Watt CPUs now, all it takes is a willingness to
> >mandate a ban > 50 Watt processors for personal use, rather like
> >banning incandescents or plasma TVs.
>
> Total bans are much harder than constraints - and, paradoxically,
> less effective.
For example: instead of banning >50 W CPUs, if everyone simply turned
lights off in rooms they just exited, you would save more electricity
(per unit time) than by eliminating all >50 Watt processors (with a
wave of your hand) for all time hense. It is currently thought that
unplugging wall warts (those 9V and 12V plastic boxes that plug into
the AC and powed portable devices) when not in use would eliminate the
need for 25-50 coal fired power plants over here in the USA.
Mitch
Andrew Reilly
> It is currently thought that
> unplugging wall warts (those 9V and 12V plastic boxes that plug into the
> AC and powed portable devices) when not in use would eliminate the need
> for 25-50 coal fired power plants over here in the USA.
I wonder whether that holds for the newer, switch-mode ones, or just the
old iron-cored transformers?
Cheers,
--
Andrew
Chris M. Thomasson
news:h2bn1g$nvq$1...@news.eternal-september.org...
> Robert Myers wrote:
>
> snip
>
>> Something like the fleet fuel efficiency averages of US CAFE standards
>> would work pretty well.
>
> No, they don't, or at least didn't. We have experience with these and
> they don't reduce gasoline consumption as much as you would think. The
> reason is that given higher MPG cars, the price of making a trip declines
> and, surprise, surprise, people drive more. They move farther away from
> work, make more trips, etc.
>
> The most reliable way to decrease the usage of something is to raise its
> price. When gasoline was $4.00 a gallon, you couldn't buy a Toyota Prius.
> When it dropped to about $2.30, you couldn't sell one.
>
> BTW, my proposal is to create a revenue neutral tax swap. Increase the
> tax on gasoline say 30 cents per gallon each year for the next 10 years.
> Take 100% of that money and refund it to the people via say a per capita
> refundable tax credit. This means that people will know that gas prices
> will go up which allows them to plan when they buy their next car, etc.
> It also taxes those people who use more than the average amount of gas,
> but these are the people who would gain the most from taking actions such
> as buying a more fuel efficient car.
What about all of the people who have to drive a lot, and also need a
non-fuel efficient vehicle (e.g., a big truck)? Well, I guess they are
screwed anyway, so, lets raise their taxes! Humm, why not have a special tax
specifically for everybody who drives a vehicle that gets less than 30 MPG?
Say, they have to pay 3-4 dollars more per-gallon. That will teach those
ignorant bastards a lesson!
;^(
Robert Myers
> Robert Myers wrote:
>
> snip
>
> > Something like the fleet fuel efficiency averages of US CAFE standards
> > would work pretty well.
>
> No, they don't, or at least didn't. We have experience with these and
> they don't reduce gasoline consumption as much as you would think. The
> reason is that given higher MPG cars, the price of making a trip
> declines and, surprise, surprise, people drive more. They move farther
> away from work, make more trips, etc.
>
> The most reliable way to decrease the usage of something is to raise its
> price. When gasoline was $4.00 a gallon, you couldn't buy a Toyota
> Prius. When it dropped to about $2.30, you couldn't sell one.
No problem there. The price of electricity is going up. And cars
*are* more fuel-efficient and would be even more fuel efficient were
it not for Detroit obstructionism (which was ultimately cleared away
by a bankruptcy bulldozer).
I have no interest in another debate in this forum about either energy
or climate policy.
My concerns are computers and the people who use them. Most
consumers, and especially 24/7 users like me, would be better off if
there were pressure to decrease power consumption while maintaining
performance.
If computers are more efficient and therefore less expensive to run,
people will be able to do more computation for the same money, and
they will probably do so if they can afford it. Unlike people making
more and longer car trips, though, it seems likely that some of that
computation will go to a good purpose, perhaps even to the benefit of
science or technological progress.
There's an issue even beyond cost, which is the ability to plug
whatever it is into a standard socket of some kind or other. The more
gigaflops you can plug into a wall outlet, the better off many
technical professionals will be.
None of this addresses the hard-core issues of the DARPA program, but
it's all movement in the right direction.
Robert.
Stephen Fuld
Most big trucks run on diesel and are used for commercial purposes.
There are at least a couple of alternatives. One is to tax only
gasoline, not diesel. This would work if it didn't cause a significant
increase in diesel use for personal (as opposed to commercial)
transportation. The other is to give a refundable tax credit for the
additional taxes for businesses, but you have to make sure it wasn't abused.
If there are people who need big trucks to drive a lot for personal, non
commercial use, please give some examples. They may exist, but I am not
aware of them. Note that I did propose that the refundable credit for
the average gas tax was per capita. This was done purposely to give a
break to large families who may need say a minivan to carry lots of kids.
> Well, I guess they are
> screwed anyway, so, lets raise their taxes!
No, that was not my intent, and I have thought about the issues and have
some answers such as I mentioned earlier in this post, but the original
post was getting long and I didn't want to go into all the gory details.
> Humm, why not have a special
> tax specifically for everybody who drives a vehicle that gets less than
> 30 MPG? Say, they have to pay 3-4 dollars more per-gallon. That will
> teach those ignorant bastards a lesson!
Too hard to administer.
Chris M. Thomasson
I failed to clarify. When I wrote "big truck", I was referring to something
more along the lines of:
http://www.rivervalleytruckoutfitters.com/images/big%20chevy.jpg
Sorry about that.
> There are at least a couple of alternatives. One is to tax only gasoline,
> not diesel. This would work if it didn't cause a significant increase in
> diesel use for personal (as opposed to commercial) transportation. The
> other is to give a refundable tax credit for the additional taxes for
> businesses, but you have to make sure it wasn't abused.
I think that Audi created an efficient diesel powered car:
http://www.audiusanews.com/newsroom.do;jsessionid=44D6987DE43FBDB919E9AE17446B8997?id=55&mid=83
> If there are people who need big trucks to drive a lot for personal, non
> commercial use, please give some examples. They may exist, but I am not
> aware of them.
non-commercial... Well, not really. However, I did know a couple of general
contractors who were building some huge mansions back in the late 80's that
were located around 40-50 miles away from where they lived. They basically
had to frequently commute back and fourth to the worksite, roundtrip almost
100 miles. They had to use non-fuel efficient trucks to carry all their
equipment. A hike in gas prices would of hit them fairly hard. I know some
home inspectors that had to frequently commute long distances during the
housing boom in the 90's. They needed trucks to carry all their equipment
(e.g., big ladders) and a high gas price would hurt them as well.
Also, basically anybody who has to commute long distances to work in any
non-fuel efficient car would get hurt. Perhaps they cannot afford to buy a
brand new fuel efficient car. They would get screwed.
> Note that I did propose that the refundable credit for the average gas tax
> was per capita. This was done purposely to give a break to large families
> who may need say a minivan to carry lots of kids.
>
>
>> Well, I guess they are screwed anyway, so, lets raise their taxes!
>
> No, that was not my intent, and I have thought about the issues and have
> some answers such as I mentioned earlier in this post, but the original
> post was getting long and I didn't want to go into all the gory details.
>
>> Humm, why not have a special tax specifically for everybody who drives a
>> vehicle that gets less than 30 MPG? Say, they have to pay 3-4 dollars
>> more per-gallon. That will teach those ignorant bastards a lesson!
>
> Too hard to administer.
Yeah. However, it would rapidly get rid of all the gas hogs. Also, it might
cause mass riots in the streets... ;^o
One way would be to make it illegal for somebody to sell any car that got
less than, say, 40-50MPG after 5 or so years from now. That would not rule
out SUV/trucks, it would just force the car companies to give that class of
vehicles powerful and highly efficient engines. I think they are smart
enough to come up with something. If not, well, good bye to their company.
nm...@cam.ac.uk
Rick Jones <rick....@hp.com> wrote:
>
>> Also, you say that there are low-powered CPUs. Have you TRIED to
>> buy a low-powered desktop recently?
>
>Nope, I prefer that my desktop actually perform :) Rather ironic in
>today's performance context at least given what if anything will
>appear in the headers of this post as to the conditions under which my
>character mode news reader were running.
As you know (or, at least, ought to!) there is no fundamental conflict
between delivering decent performance and low power. The reasons that
modern systems are so power-hungry, hot and noisy are benchmarketing,
bloatware and fanatical gamers - and why do you want to encourage the
first two (or even let the third dominate the industry)?
>> I have. I couldn't find one, at an even remotely affordable price -
>> the point of my condition is to ensure that they become available,
>> and development is slanted towards that direction.
>
>Is it too great a stretch to call a laptop with a docking station a
>desktop these days?
Yes. They come with appalling specifications - not even parity on
memory, fer chrissake! Secondly, once one has attached a separate
keyboard, mouse and monitor to restore some decent ergonomics, they
don't leave much of a desk left ....
But you may be realising the gist of my point. If any major vendor
made a "green line" of desktops and blades, using laptop power targets
but otherwise with normal desktop/server specifications, they could
easily reduce then power by a quarter while reducing the performance
by only a factor of two. ALL that a major government has to do is to
encourage their production by some procurement rules for enough public
contracts, and vendors WILL make them.
I know a LOT of people and companies who would jump at those - and
some would buy four times as many (i.e. they are limited by room
power/cooling). I think that there would be a (partial) modal switch.
Regards,
Nick Maclaren.
Terje Mathisen
> But you may be realising the gist of my point. If any major vendor
> made a "green line" of desktops and blades, using laptop power targets
> but otherwise with normal desktop/server specifications, they could
> easily reduce then power by a quarter while reducing the performance
> by only a factor of two. ALL that a major government has to do is to
> encourage their production by some procurement rules for enough public
> contracts, and vendors WILL make them.
Isn't this exactly what the Asus "Eee Box" tiny desktop boxes do?
Pretty much the same hardware as the Netbook machines, but in a tiny
standalone box with external mouse/kbd/screen.
>
> I know a LOT of people and companies who would jump at those - and
> some would buy four times as many (i.e. they are limited by room
> power/cooling). I think that there would be a (partial) modal switch.
OK, this is the server/cluster environment, in which case you're looking
at something like Google's custom motherboards, or maybe the units used
for the commercially available PetaBox?
Terje
--
- <Terje.Mathisen at tmsw.no>
"almost all programming can be viewed as an exercise in caching"
nm...@cam.ac.uk
Terje Mathisen <"terje.mathisen at tmsw.no"> wrote:
>
>> But you may be realising the gist of my point. If any major vendor
>> made a "green line" of desktops and blades, using laptop power targets
>> but otherwise with normal desktop/server specifications, they could
>> easily reduce then power by a quarter while reducing the performance
>> by only a factor of two. ALL that a major government has to do is to
>> encourage their production by some procurement rules for enough public
>> contracts, and vendors WILL make them.
>
>Isn't this exactly what the Asus "Eee Box" tiny desktop boxes do?
No. I said "normal desktop/server specifications". They have bottom
end laptop specifications.
>Pretty much the same hardware as the Netbook machines, but in a tiny
>standalone box with external mouse/kbd/screen.
Precisely.
Regards,
Nick Maclaren.
Ken Hagan
I think it quite specifically doesn't, just as most TVs on stand-by now
consume about 1W rather than the received wisdom of "almost as much as
they do when they are on".
I hear these two claims (about chargers and appliances left on stand-by)
quite often, but as far as I know they are both *now* entirely bogus,
whatever their historical truth. Repeating them is merely handing
ammunition to the anti-green lobby, which presumably isn't the intention
of the sort of people who habitually repeat them. :)
Kim Enkovaara
> screwed anyway, so, lets raise their taxes! Humm, why not have a special
> tax specifically for everybody who drives a vehicle that gets less than
> 30 MPG? Say, they have to pay 3-4 dollars more per-gallon. That will
> teach those ignorant bastards a lesson!
That would be then about the same price as in europe. Here in europe we
pay already over $7/gallon and that does not depend on the car. Just
rise the taxes in gasoline enough, and you start to see energy efficient
cars. And maybe that tax-money will fix the budget deficit also.
--Kim
Kim Enkovaara
> Most big trucks run on diesel and are used for commercial purposes.
> There are at least a couple of alternatives. One is to tax only
> gasoline, not diesel. This would work if it didn't cause a significant
> increase in diesel use for personal (as opposed to commercial)
> transportation. The other is to give a refundable tax credit for the
In western europe ~50% of the sold new cars are already diesel. So the
jump to diesel would be quite quick if taxation would be like that.
--Kim
Morten Reistad
Stephen Fuld <SF...@Alumni.cmu.edu.invalid> wrote:
>Robert Myers wrote:
>
>snip
>
>> Something like the fleet fuel efficiency averages of US CAFE standards
>> would work pretty well.
>
>No, they don't, or at least didn't. We have experience with these and
>they don't reduce gasoline consumption as much as you would think. The
>reason is that given higher MPG cars, the price of making a trip
>declines and, surprise, surprise, people drive more. They move farther
>away from work, make more trips, etc.
>
>The most reliable way to decrease the usage of something is to raise its
>price. When gasoline was $4.00 a gallon, you couldn't buy a Toyota
>Prius. When it dropped to about $2.30, you couldn't sell one.
As I keep harping about, the generation of base energy in sufficient
quantities at reasonable prices is the #1 challenge for western
civilisation. Note that South America, Russia, China and India is
rapidly joining the western civilisation as we speak. We threfore
need to plan for an energy consumption at around 3 times the current
level. In oil equivalents we are using close to 180 bbls/day, of
which a tad more than 80 comes from oil and oil equivalents.
We do not have the option of tripling oil production. We probably
had peak oil in late 2006, but the peak seems to be a slowly sliding
plateau with some jagged terrain, not the immediate fall doomssaysers
have predicted.
In the short to medium term I see no other alternative than nuclear
power plants. We simply cannot spew out more coal. Most of east
Asia is already in a coal-induced haze.
>BTW, my proposal is to create a revenue neutral tax swap. Increase the
>tax on gasoline say 30 cents per gallon each year for the next 10 years.
> Take 100% of that money and refund it to the people via say a per
>capita refundable tax credit. This means that people will know that gas
>prices will go up which allows them to plan when they buy their next
>car, etc. It also taxes those people who use more than the average
>amount of gas, but these are the people who would gain the most from
>taking actions such as buying a more fuel efficient car. The policy
>rewards those who use less than the average amount of gas as they
>actually make money on the deal.
You are too late. Carter, or even Bush sr could have done this. Now
the oil market will do this for us, except the tax is a few dollars a
gallon and the recipients live in Saudi Arabia.
>A slight modification, which may be useful is to make the tax on only
>the petroleum portion of the gas. Thus if we used say 10% bio-materials
>in the gas, the tax would be reduced by 10%. By doing this, we would
>incent the usage of things like ethanol, and we could phase out the
>ethanol subsidy we currently pay to farmers, thus actually reducing
>government expenditures.
>
>
>> It's a happy accident that energy policy and climate policy both push
>> in the direction the industry needs to go,
>
>Not necessarily. Reducing petroleum usage favors things like electric
>cars and plug in hybrids, but this increases electricity demand. In the
>US, most electricity is generated from coal, so we have to be careful
>not to increase emissions due to burning more coal. (yes, I know that
>overall, it is still a win, but it isn't totally simplistic.) Also,
>people are talking about hydrogen cars, but the most common way to make
>hydrogen is from natural gas. This stuff is very tricky.
If we designed things right this could be a slow, silent migration
instead of the huge shifts we probably will see now. And we need to
get the nuclear industry moving again.
The current hybrids still think petrol, or they go all overboard with
all-electric. We don't have the batteries for this, yet. But one of
those all-electric drive trains, with added integrated combustion
engine and generator, plus some battery capacity could revolutionise
automotion. After all, we use a significant part of time in our cars
near home, work or other places where we could charge a car; even if
that radius is down to 20 km. Having plug-ins that could handle this
driving would make a real dent in petrol consumption.
-- mrr
Morten Reistad
Most processors readily step down to below 50W when they are idling.
Unfortunatly lots of software is such a bloat you never get the
processor to idle.
-- mrr
nm...@cam.ac.uk
Morten Reistad <fi...@last.name> wrote:
>
>Most processors readily step down to below 50W when they are idling.
>Unfortunatly lots of software is such a bloat you never get the
>processor to idle.
Yeah, but the objective should be ones that consume under 50 watts
at full load.
Regards,
Nick Maclaren.
Stephen Sprunk
> Robert Myers wrote:
>> Something like the fleet fuel efficiency averages of US CAFE standards
>> would work pretty well.
>
> No, they don't, or at least didn't. We have experience with these and
> they don't reduce gasoline consumption as much as you would think. The
> reason is that given higher MPG cars, the price of making a trip
> declines and, surprise, surprise, people drive more. They move farther
> away from work, make more trips, etc.
>
> The most reliable way to decrease the usage of something is to raise its
> price. When gasoline was $4.00 a gallon, you couldn't buy a Toyota
> Prius. When it dropped to about $2.30, you couldn't sell one.
Logically, though, increasing the price of gas will lead to customers
buying higher-mileage cars, with the same result. It just takes longer
to see, since buying gasoline is a weekly cycle while buying cars is a
multi-year cycle.
> BTW, my proposal is to create a revenue neutral tax swap. Increase the
> tax on gasoline say 30 cents per gallon each year for the next 10 years.
> Take 100% of that money and refund it to the people via say a per
> capita refundable tax credit. This means that people will know that gas
> prices will go up which allows them to plan when they buy their next
> car, etc. It also taxes those people who use more than the average
> amount of gas, but these are the people who would gain the most from
> taking actions such as buying a more fuel efficient car. The policy
> rewards those who use less than the average amount of gas as they
> actually make money on the deal.
>
> A slight modification, which may be useful is to make the tax on only
> the petroleum portion of the gas. Thus if we used say 10% bio-materials
> in the gas, the tax would be reduced by 10%. By doing this, we would
> incent the usage of things like ethanol, and we could phase out the
> ethanol subsidy we currently pay to farmers, thus actually reducing
> government expenditures.
This makes perfect sense and is the automotive fuels part of a "carbon
tax" plan: tax all production or import (not consumption, which is easy
to cheat) of fossil fuels while not taxing carbon-neutral fuels. This
would alter the economics in favor of biofuels, which are already nearly
cost-competitive with fossil fuels, in turn driving up the prices of
fuel crops like corn, sugar cane, switchgrass, etc. and reducing the
need for agricultural subsidies.
The problem with that, or with your original plan, is that any US
politician who voted for it would find himself out of a job after the
next election, and they know it. The American public has grown
dependent on cheap oil and, like any addict, will lash out at anyone who
gets in the way of their fix -- even if they know, intellectually, that
getting sober is better for them long-term.
The other issue, which I've never seen fully discussed (usually because
debates are so heated that they never get there) is what to do with all
that "carbon tax" revenue...
>> It's a happy accident that energy policy and climate policy both push
>> in the direction the industry needs to go,
>
> Not necessarily. Reducing petroleum usage favors things like electric
> cars and plug in hybrids, but this increases electricity demand. In the
> US, most electricity is generated from coal, so we have to be careful
> not to increase emissions due to burning more coal. (yes, I know that
> overall, it is still a win, but it isn't totally simplistic.)
Coal is dominant in some parts of the country; in others, it's
relatively minor and hydro, fuel oil, natural gas, and/or nuclear dominate.
A carbon tax would cover all of those fossil fuels as well; that should
be enough to make greener sources of electricity (e.g. hydro, solar,
wind, tidal, geothermal, nuclear) dominant.
Nuclear power is our best choice in the short term; it's cheap, it's
reliable, it's well understood, there's plenty of fuel, and it doesn't
have any net emissions. The tree-huggers are against it because of the
"waste" storage problem, but the only reason we _have_ a "waste" problem
is Carter's ban on breeder reactors which can use that "waste" as fuel.
> Also, people are talking about hydrogen cars, but the most common way
> to make hydrogen is from natural gas. This stuff is very tricky.
... or from electrolysis, which again increases demand for electricity.
S
--
Stephen Sprunk "Stupid people surround themselves with smart
CCIE #3723 people. Smart people surround themselves with
K5SSS smart people who disagree with them." --Isaac Jaffe
Stephen Sprunk
> What about all of the people who have to drive a lot, and also need a
> they are screwed anyway, so, lets raise their taxes!
First of all, they can switch to biofuels, which are only marginally
more expensive than fossil fuels, or plug-in hybrids (cf. discussion
about greener electricity).
Second, their desire (not need) to drive long distances in inefficient
vehicles does not override the rest of the population's need (not
desire) for a clean environment.
Third, it's just another cost of doing business (or of commuting).
Perhaps, if the true cost of their consumption was passed on to them,
they would move closer to their destination or someone who is already
close to the destination would get the business instead, reducing total
consumption.
(For real trucks, the answer is that more of that traffic would shift to
railroads, which are far more energy-efficient than trucks and can more
easily electrify. Trucks would shift to taking mostly short-haul loads
between an intermodal center and the origin/destination, rather than
mostly long-haul trips as they do today. Shippers would also relocate
near rail lines, eliminating many of the origin-to-intermodal trucks and
again reducing overall consumption.)
> Humm, why not have a special tax specifically for everybody who drives
> a vehicle that gets less than 30 MPG? Say, they have to pay 3-4 dollars
> more per-gallon. That will teach those ignorant bastards a lesson!
Good luck enforcing that. The only place you can reliably implement
taxes on commodities is in the supply chain close to the producer (or
importer), long before you know who's using any individual unit.
MitchAlsup
Why not have an adjustible tax. Those vehicles that are driven in such
a way that the vehicle achieves greater than EPA milage are imune from
gasoline (or diesel) taxes, while those that do not achieve their EPA
milage pay double? The vehicle engine computer calculates the actual
milage and then "talks to the "pump" durring the fill.
This encourages driving the vehicles for milage. BTW, every vehicle I
have owned in the past 19 years has regularly gotten 15% more than EPA
rated milage.
Mitch
Stephen Fuld
snip
> If any major vendor
> made a "green line" of desktops and blades, using laptop power targets
> but otherwise with normal desktop/server specifications, they could
> easily reduce then power by a quarter while reducing the performance
> by only a factor of two.
Is that really true? I agree that you would save on CPU, but RAM is the
same, and presumably you want the increased power of a large screen and
the capacity of a 3.5 in disk (rather than the laptop's 2.5). Low end
desktops already use integrated graphics so you don't save much there.
You might save a little by eliminating the extra ports, etc. but you
might want to preserve expandability. I'm not sure just going to a
laptop type CPU would save 1/4 of the total system power.
Thomas Womack
>In article <h2bcpg$de5$1...@usenet01.boi.hp.com>,
>Rick Jones <rick....@hp.com> wrote:
>>nm...@cam.ac.uk wrote:
>>> Like the USA (or even UK) government stating that central
>>> procurements will specify a 10% reduction on power (watts)
>>> year-on-year, until we get down to 50 watts per user.
>>
>>That sounds more like an arbitrary mandate than a research investment.
>
>Actually, I was referring just to public procurements.
>
>>Given there are < 50 Watt CPUs now, all it takes is a willingness to
>>mandate a ban > 50 Watt processors for personal use, rather like
>>banning incandescents or plasma TVs.
>Also, you say that there are low-powered CPUs. Have you TRIED to
>buy a low-powered desktop recently? I have. I couldn't find one,
>at an even remotely affordable price
To what extent does the current Mac Mini, which is a 2.4GHz dual-core
64-bit processor with 2G memory running BSD unix in a box the size of
quite a small box, using about 50W, costing about 600 pounds and
purchasable just by turning up to the shop in the Grand Arcade, not
meet your requirements?
Tom
Stephen Fuld
Oh! I thought you meant things like 18 wheelers. As for things like
the one pictured in the link above, I believe most of them are used for
"fun" and for people who want a "big" image, and I don't have a problem
making them pay. But see below for the exceptions you discuss below.
>> There are at least a couple of alternatives. One is to tax only
>> gasoline, not diesel. This would work if it didn't cause a
>> significant increase in diesel use for personal (as opposed to
>> commercial) transportation. The other is to give a refundable tax
>> credit for the additional taxes for businesses, but you have to make
>> sure it wasn't abused.
>
> I think that Audi created an efficient diesel powered car:
>
> http://www.audiusanews.com/newsroom.do;jsessionid=44D6987DE43FBDB919E9AE17446B8997?id=55&mid=83
Yes, I know. But if you try to fix the problem with 18 wheeler type
trucks by not taxing diesel, you would create an incentive for personal
use of diesel and no larger incentive to conserve it. This is the
situation we have now with gasoline and is the problem we are trying to
fix. That is, I am not opposed to cars such as the Audi you reference,
but to the potential appearance of "diesel hog" personal use vehicles.
>> If there are people who need big trucks to drive a lot for personal,
>> non commercial use, please give some examples. They may exist, but I
>> am not aware of them.
>
> non-commercial... Well, not really. However, I did know a couple of
> general contractors who were building some huge mansions back in the
> late 80's that were located around 40-50 miles away from where they
> lived. They basically had to frequently commute back and fourth to the
> worksite, roundtrip almost 100 miles. They had to use non-fuel efficient
> trucks to carry all their equipment.
Two comments. first, these are temporary situations - once the
"mansion" is built, they don't have to do that anymore. Second, I have
no problem with the extra taxes on their gas causing them to raise their
price and making the mansion more expensive. I suspect the additional
cost of the larger gas tax is a trivial percentage of the mansion's cost.
> A hike in gas prices would of hit
> them fairly hard. I know some home inspectors that had to frequently
> commute long distances during the housing boom in the 90's. They needed
> trucks to carry all their equipment (e.g., big ladders) and a high gas
> price would hurt them as well.
Well, that is a commercial use, and temporary. And adding high gas
costs to the one or two trips for an inspection isn't that much. Say
the inspector had to travel 100 miles and that his vehicle got only 10
MPG (a very low estimate) If the gas tax was up to $3.00 per gallon,
that would increase the cost by %60, total. Not a huge amount of money
in the context of a probably $100,000 house.
> Also, basically anybody who has to commute long distances to work in any
> non-fuel efficient car would get hurt. Perhaps they cannot afford to buy
> a brand new fuel efficient car. They would get screwed.
The issues here are "has to commute" and "non-fuel efficient vehicle".
I guess I am maintaining that people in this category are few and far
between. The examples you cited are not really "commutes" in they don't
happen every day. And remember, the tax is phased in over 10 years. So
most people are going to buy a new vehicle in that time frame anyway.
So this proposal just gives them a "push" for that vehicle to be more
fuel efficient.
snip
> One way would be to make it illegal for somebody to sell any car that
> got less than, say, 40-50MPG after 5 or so years from now. That would
> not rule out SUV/trucks, it would just force the car companies to give
> that class of vehicles powerful and highly efficient engines. I think
> they are smart enough to come up with something. If not, well, good bye
> to their company.
Several problems with that. We know from past experience that simply
increasing fuel efficiency doesn't reduce fuel consumption as much as
one would think (that was the reason for my first post in this thread).
Second, we have the problem of distinguishing "cars" from SUV's, that
allowed a glaring exception to the CAFE standards the last time. It's
amazing how many customers suddenly determined that they needed an SUV
when they were exempted from the CAFE standards. Third, you state you
"think" the car companies are capable of doing as you suggest and if you
are wrong, "screw them". And if no company is, then good bye auto
industry. I'd hate to risk a whole industry and the tens of thousands
of jobs on your "thoughts".
Stephen Fuld
> Chris M. Thomasson wrote:
>> screwed anyway, so, lets raise their taxes! Humm, why not have a
>> special tax specifically for everybody who drives a vehicle that gets
>> less than 30 MPG? Say, they have to pay 3-4 dollars more per-gallon.
>> That will teach those ignorant bastards a lesson!
>
> That would be then about the same price as in europe.
Yup. That is why I picked the numbers I did.
> Here in europe we
> pay already over $7/gallon and that does not depend on the car. Just
> rise the taxes in gasoline enough, and you start to see energy efficient
> cars.
Yes, but remember, we have somewhat different situation in the US that
we have to take into account. We have a much larger (physically)
country and it is more spread out than most of Europe. We have to take
this into account.
And maybe that tax-money will fix the budget deficit also.
NO! The only way this could possible pas is to make it totally budget
neutral by the refund mechanism I proposed. Simply raising the gas tax
and giving the money to the government is politically impossible - i.e.
it would never pass our Congress. The reason is that no one thinks the
money would go to reducing the deficit - it would just get spent on
other things.
Stephen Fuld
> In article <h2bn1g$nvq$1...@news.eternal-september.org>,
> Stephen Fuld <SF...@Alumni.cmu.edu.invalid> wrote:
>> Robert Myers wrote:
>>
>> snip
>>
>>> Something like the fleet fuel efficiency averages of US CAFE standards
>>> would work pretty well.
>> No, they don't, or at least didn't. We have experience with these and
>> they don't reduce gasoline consumption as much as you would think. The
>> reason is that given higher MPG cars, the price of making a trip
>> declines and, surprise, surprise, people drive more. They move farther
>> away from work, make more trips, etc.
>>
>> The most reliable way to decrease the usage of something is to raise its
>> price. When gasoline was $4.00 a gallon, you couldn't buy a Toyota
>> Prius. When it dropped to about $2.30, you couldn't sell one.
>
> As I keep harping about, the generation of base energy in sufficient
> quantities at reasonable prices is the #1 challenge for western
> civilisation. Note that South America, Russia, China and India is
> rapidly joining the western civilisation as we speak. We threfore
> need to plan for an energy consumption at around 3 times the current
> level. In oil equivalents we are using close to 180 bbls/day, of
> which a tad more than 80 comes from oil and oil equivalents.
We have to be careful here not to conflate oil use, which is primarily
for transportation, and other things which are primarily used for heat
and electricity generation. While there is some ability to inter
convert between the two it is better to talk about them separately. My
proposal strictly addressed petroleum usage for transportation.
> We do not have the option of tripling oil production. We probably
> had peak oil in late 2006, but the peak seems to be a slowly sliding
> plateau with some jagged terrain, not the immediate fall doomssaysers
> have predicted.
Agreed, According tho the then relevant experts, we have been running
out of oil every few years for about a century. :-( But we will
certainly run out sometime.
> In the short to medium term I see no other alternative than nuclear
> power plants. We simply cannot spew out more coal. Most of east
> Asia is already in a coal-induced haze.
I agree we need more nuclear power generation. And we need to eliminate
President Carter's ban on the breeder reactors we need to substantially
reduce nuclear waste. (What the French do.)
>> BTW, my proposal is to create a revenue neutral tax swap. Increase the
>> tax on gasoline say 30 cents per gallon each year for the next 10 years.
>> Take 100% of that money and refund it to the people via say a per
>> capita refundable tax credit. This means that people will know that gas
>> prices will go up which allows them to plan when they buy their next
>> car, etc. It also taxes those people who use more than the average
>> amount of gas, but these are the people who would gain the most from
>> taking actions such as buying a more fuel efficient car. The policy
>> rewards those who use less than the average amount of gas as they
>> actually make money on the deal.
>
> You are too late. Carter, or even Bush sr could have done this. Now
> the oil market will do this for us, except the tax is a few dollars a
> gallon and the recipients live in Saudi Arabia.
I don't think it is to late, but I agree the increased money will go to
the Saudis. Note that my proposal is for a steadily increasing tax, so
people will anticipate and start to reduce consumption immediately.
>> A slight modification, which may be useful is to make the tax on only
>> the petroleum portion of the gas. Thus if we used say 10% bio-materials
>> in the gas, the tax would be reduced by 10%. By doing this, we would
>> incent the usage of things like ethanol, and we could phase out the
>> ethanol subsidy we currently pay to farmers, thus actually reducing
>> government expenditures.
>>
>>
>>> It's a happy accident that energy policy and climate policy both push
>>> in the direction the industry needs to go,
>> Not necessarily. Reducing petroleum usage favors things like electric
>> cars and plug in hybrids, but this increases electricity demand. In the
>> US, most electricity is generated from coal, so we have to be careful
>> not to increase emissions due to burning more coal. (yes, I know that
>> overall, it is still a win, but it isn't totally simplistic.) Also,
>> people are talking about hydrogen cars, but the most common way to make
>> hydrogen is from natural gas. This stuff is very tricky.
>
> If we designed things right this could be a slow, silent migration
> instead of the huge shifts we probably will see now. And we need to
> get the nuclear industry moving again.
Agreed.
> The current hybrids still think petrol, or they go all overboard with
> all-electric. We don't have the batteries for this, yet. But one of
> those all-electric drive trains, with added integrated combustion
> engine and generator, plus some battery capacity could revolutionise
> automotion.
That is called the Chevy Volt. It is called an "extended range
electric". All electric drive train with plug in rechargeable batteries
and the 1.4 liter engine from their European division connected to a
generator. Range on batteries is projected to be 200 miles with
automatic start-up of the gas engine for longer trips. Will be
available, IIRC in late 2010 or 2011. But it will be more expensive
than a pure gas or electric due both electric and gas engines and more
expensive than a hybrid due to the need for more batteries than a hybrid.
> After all, we use a significant part of time in our cars
> near home, work or other places where we could charge a car; even if
> that radius is down to 20 km. Having plug-ins that could handle this
> driving would make a real dent in petrol consumption.
That's GM's plan.
Robert Swindells
As a data point, I have a MiniITX motherboard with an Atom 230 CPU, 1GB
DDR2 and a 250GB 3.5 in SATA disk. Using a cheap ATX PSU it draws 45W flat
out.
The screen isn't included in this power reading.
Robert Swindells
Stephen Fuld
> Stephen Fuld wrote:
>> Robert Myers wrote:
>>> Something like the fleet fuel efficiency averages of US CAFE standards
>>> would work pretty well.
>> No, they don't, or at least didn't. We have experience with these and
>> they don't reduce gasoline consumption as much as you would think. The
>> reason is that given higher MPG cars, the price of making a trip
>> declines and, surprise, surprise, people drive more. They move farther
>> away from work, make more trips, etc.
>>
>> The most reliable way to decrease the usage of something is to raise its
>> price. When gasoline was $4.00 a gallon, you couldn't buy a Toyota
>> Prius. When it dropped to about $2.30, you couldn't sell one.
>
> Logically, though, increasing the price of gas will lead to customers
> buying higher-mileage cars, with the same result.
Not quite. Higher mileage cars plus higher cost gas is, for sake of
argument cost neutral for a given trip. Just higher mileage cars with
no increased gas cost means lower cost for a longer trip.
snip
> The problem with that, or with your original plan, is that any US
> politician who voted for it would find himself out of a job after the
> next election, and they know it. The American public has grown
> dependent on cheap oil and, like any addict, will lash out at anyone who
> gets in the way of their fix -- even if they know, intellectually, that
> getting sober is better for them long-term.
>
> The other issue, which I've never seen fully discussed (usually because
> debates are so heated that they never get there) is what to do with all
> that "carbon tax" revenue...
You missed one of my big points. The 100% of the revenue from the
increased gas tax going back to the consumer in the form of a per capita
creditable tax credit. Thus, it is cost neutral to the person using the
average amount of gas. And it is a tax that can be avoided by buying a
more fuel efficient car, etc. The hope is that this would allow the
Republican's to vote for it, as it isn't a net tax increase and the
Democrats would accept is even though they don't get any more revenue
out of it because they want to see reduction in gas usage. This is
designed explicitly to solve the political problems you mention above.
snip
> Nuclear power is our best choice in the short term; it's cheap, it's
> reliable, it's well understood, there's plenty of fuel, and it doesn't
> have any net emissions. The tree-huggers are against it because of the
> "waste" storage problem, but the only reason we _have_ a "waste" problem
> is Carter's ban on breeder reactors which can use that "waste" as fuel.
Agreed, but that is a different issue from petroleum usage, which was
the subject of my first post.
nm...@cam.ac.uk
Stephen Fuld <SF...@Alumni.cmu.edu.invalid> wrote:
>
>> If any major vendor
>> made a "green line" of desktops and blades, using laptop power targets
>> but otherwise with normal desktop/server specifications, they could
>> easily reduce then power by a quarter while reducing the performance
>> by only a factor of two.
>
>Is that really true? I agree that you would save on CPU, but RAM is the
>same, and presumably you want the increased power of a large screen and
>the capacity of a 3.5 in disk (rather than the laptop's 2.5). Low end
>desktops already use integrated graphics so you don't save much there.
>You might save a little by eliminating the extra ports, etc. but you
>might want to preserve expandability. I'm not sure just going to a
>laptop type CPU would save 1/4 of the total system power.
That's not what I said, and not what I meant. The point is that you
would have to engineer the WHOLE system for economy, and not just
the CPU. Yes, the RAM would be slower, but remember that factor of
two. And similarly for disks - some performance would be sacrificed
for lower power.
And I was talking about reducing the power TO a quarter, not BY a
quarter.
Regards,
Nick Maclaren.
nm...@cam.ac.uk
Thomas Womack <two...@chiark.greenend.org.uk> wrote:
>
>>Also, you say that there are low-powered CPUs. Have you TRIED to
>>buy a low-powered desktop recently? I have. I couldn't find one,
>>at an even remotely affordable price
>
>To what extent does the current Mac Mini, which is a 2.4GHz dual-core
>64-bit processor with 2G memory running BSD unix in a box the size of
>quite a small box, using about 50W, costing about 600 pounds and
>purchasable just by turning up to the shop in the Grand Arcade, not
>meet your requirements?
Well, you can start with no parity on memory and only one disk, and
then carry on with the fact that Apple quote it at 110 watts, even
EXCLUDING the monitor.
http://www.apple.com/macmini/specs.html
I should have to do more research to see if Mini-DVI is acceptable;
I can witness that DVI itself can be a major problem.
Regards,
Nick Maclaren.
Thomas Womack
>In article <hCk*BZ...@news.chiark.greenend.org.uk>,
>Thomas Womack <two...@chiark.greenend.org.uk> wrote:
>>
>>>Also, you say that there are low-powered CPUs. Have you TRIED to
>>>buy a low-powered desktop recently? I have. I couldn't find one,
>>>at an even remotely affordable price
>>
>>To what extent does the current Mac Mini, which is a 2.4GHz dual-core
>>64-bit processor with 2G memory running BSD unix in a box the size of
>>quite a small box, using about 50W, costing about 600 pounds and
>>purchasable just by turning up to the shop in the Grand Arcade, not
>>meet your requirements?
>
>Well, you can start with no parity on memory and only one disk
To a good approximation nothing has parity on memory, because it's not
needed. I've run a lot of GF(2) sparse-matrix calculations -
iterations v->Mv on a 10^7-square matrix of weight 10^9 - which are
perfect for picking up bit-flips - for ten or twenty core-weeks on
modern customer machines filled up with memory, with successful
results except on one machine whose memory I've not bothered to send
back since it's worked fine as file-server and general workstation.
I am very unconvinced by the need for parity memory; it only lets you
tell if the memory has become flaky, whilst with a modern OS, machines
are reliable enough that random crashes imply flaky memory, and
something like prime95 or msieve tests memory hard enough to convince
you.
If you want another disk, the Mac Mini has Firewire 800 which is
enough to support a bus-powered drive.
110 watts seems a bit high for what is a high-spec laptop, though I
suppose nVidia's chipsets are not terribly good at powering down. I
don't know if desktop OS X will let you pretend that it's in a laptop
in low-battery state and clock the CPU at 1.6GHz even when you have
work to do, if you prefer low peak power consumption to getting things
done quickly - full-speed and then idle seems the sensible
average-power answer, and the Mac Mini is 13W idle.
Tom
Bill Todd
...
> I am very unconvinced by the need for parity memory; it only lets you
> tell if the memory has become flaky, whilst with a modern OS, machines
> are reliable enough that random crashes imply flaky memory, and
> something like prime95 or msieve tests memory hard enough to convince
> you.
A fair number of people see value in catching flaky memory *before* it
has the opportunity to corrupt other information in the system (of
particular importance in a file server, for example, where trashing
random portions of the data is bad enough but trashing random metadata
can lose everything).
Or did I not understand your contention above?
- bill
Rick Jones
> Thomas Womack wrote:
> > I am very unconvinced by the need for parity memory; it only lets
> > you tell if the memory has become flaky, whilst with a modern OS,
> > machines are reliable enough that random crashes imply flaky
> > memory, and something like prime95 or msieve tests memory hard
> > enough to convince you.
> A fair number of people see value in catching flaky memory *before*
> it has the opportunity to corrupt other information in the system
> (of particular importance in a file server, for example, where
> trashing random portions of the data is bad enough but trashing
> random metadata can lose everything).
My wetware dimms have too many parity errors of their own to have the
specifics down pat, but I seem to recall around the turn of the
century/millenium there was a big dust-up over parity errors in the
caches on systems like Enterprise 10000's causing folks like eBay to
hit the front page of the newspaper. Finding specifics via web
searches is a bit difficult - perhaps the Anne & Lynn Wheeler bot will
kick-in with something?-) but until then I did find this:
<http://www.sunshack.org/data/sh/2.1/infoserver.central/data/syshbk/collections/fins/I0616-1.html>
which seems to describe a number of workarounds to deal with parity
errors in CPU caches.
I seem to recall that other non-trivial systems manufacturers had
already been discovering that increased cache (and/or perhaps memory)
sizes of the time were running into an increased likelihood of parity
errors, especially for systems operating at altitude (eg Denver etc)
and ECC became increasingly common from that time forward.
rick jones
--
The computing industry isn't as much a game of "Follow The Leader" as
it is one of "Ring Around the Rosy" or perhaps "Duck Duck Goose."
- Rick Jones
these opinions are mine, all mine; HP might not want them anyway... :)
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...
Bill Todd
> Bill Todd <bill...@metrocast.net> wrote:
>> Thomas Womack wrote:
>
>>> I am very unconvinced by the need for parity memory; it only lets
>>> you tell if the memory has become flaky, whilst with a modern OS,
>>> machines are reliable enough that random crashes imply flaky
>>> memory, and something like prime95 or msieve tests memory hard
>>> enough to convince you.
>
>> A fair number of people see value in catching flaky memory *before*
>> it has the opportunity to corrupt other information in the system
>> (of particular importance in a file server, for example, where
>> trashing random portions of the data is bad enough but trashing
>> random metadata can lose everything).
>
> My wetware dimms have too many parity errors of their own to have the
> specifics down pat, but I seem to recall around the turn of the
> century/millenium there was a big dust-up over parity errors in the
> caches on systems like Enterprise 10000's causing folks like eBay to
> hit the front page of the newspaper. Finding specifics via web
> searches is a bit difficult - perhaps the Anne & Lynn Wheeler bot will
> kick-in with something?-) but until then I did find this:
>
> <http://www.sunshack.org/data/sh/2.1/infoserver.central/data/syshbk/collections/fins/I0616-1.html>
>
> which seems to describe a number of workarounds to deal with parity
> errors in CPU caches.
This may (or may not) have been related to the infamous 'zinc whiskers'
problem. My own very vague recollection was that they had tried to get
by without any parity (let alone ECC) and got bitten badly by that, but
I could easily be mistaken (it may instead have been that they tried to
get away without mirroring their cache and thus could not recover from a
parity error there).
>
> I seem to recall that other non-trivial systems manufacturers had
> already been discovering that increased cache (and/or perhaps memory)
> sizes of the time were running into an increased likelihood of parity
> errors, especially for systems operating at altitude (eg Denver etc)
> and ECC became increasingly common from that time forward.
I'd certainly prefer ECC RAM to mere parity RAM in a server and wasn't
entirely sure which Nick meant to refer to originally here, but parity
is probably better than nothing in terms of catching problems earlier
rather than later even if you then have to crash because you can't
recover from them as ECC might allow you to.
- bill
Terje Mathisen
> Kim Enkovaara wrote:
> > Here in europe we
>> pay already over $7/gallon and that does not depend on the car. Just
>> rise the taxes in gasoline enough, and you start to see energy efficient
>> cars.
>
> Yes, but remember, we have somewhat different situation in the US that
> we have to take into account. We have a much larger (physically)
> country and it is more spread out than most of Europe. We have to take
> this into account.
OTOH, population density is still significantly higher than several
European countries.
People living in northern Norway (or Sweden/Finland) have exactly the
same distance issues as someone in Montana or the Mid-West.
During my army time up north in Finnmark (a few hours south of North
Cape), I was amazed by the distances people were willing to drive in an
afternoon, just to maybe meet some others.
After a year in Utah I noticed that parts of the US is very similar. :-)
Thomas Womack
The machine I use as my file server is the one that I know has memory
flaky enough that I can't do four-week linear algebra runs on it;
losing a bit of one filesystem because a once-a-year failure happens
to corrupt a critical block before it corrupts one of thousands of
blocks which cause the machine to crash is a once-in-decades risk I'm
happy to run. It's bitten me once, when I thought a memory problem
was a disc problem and ran chkdsk while every 1024th bit of memory
read came back zero.
And whilst that left the FS structure comprehensively rogered, I was
able to recover the interesting directories by grepping /dev/hda to
find the dirent as a cluster of recognisable filenames, then following
the lists of sectors from there to get the files back.
Tom
Bill Todd
That kind of error (and subsequent effort) may be acceptable to you for
your personal data, but most businesses would likely not find it so -
especially when they could guard against it relatively inexpensively.
And recovery from backup is becoming increasingly unpleasant as data
sizes burgeon, leaving aside whatever new data accumulated since the
most recent backup that can't be recovered that way.
But that's just my own take on the situation, and I'd be happy to hear
others.
- bill
nm...@cam.ac.uk
>>
>> And whilst that left the FS structure comprehensively rogered, I was
>> able to recover the interesting directories by grepping /dev/hda to
>> find the dirent as a cluster of recognisable filenames, then following
>> the lists of sectors from there to get the files back.
>
>That kind of error (and subsequent effort) may be acceptable to you for
>your personal data, but most businesses would likely not find it so -
>especially when they could guard against it relatively inexpensively.
>And recovery from backup is becoming increasingly unpleasant as data
>sizes burgeon, leaving aside whatever new data accumulated since the
>most recent backup that can't be recovered that way.
You have given two reasons to want checking on memory (and, as Rick
says, caches - and elsewhere). Those of us who are professionals
and/or very experienced don't want to waste the time trying to track
down obscure errors that can easily and automatically be detected.
It is a common delusion of the inexperienced that memory errors are
always 'hard' - 90% of them are, but some are absolute b*gg*rs.
Also, one simple error can lead to widespread and nasty trashing.
That happened to me the other day - recovering the files I had
erroneously deleted took an hour - but I had to reinstall and
reconfigure (manually, for GUIs) because the system had trashed
itself in the short window when it was missing files. Days lost.
[ For reasons outside my control, I wasn't using a system with
snapshots or other reliable, instantaneous backup technology. ]
ECC may be overkill, but you can't seem to get parity (except on
write-through caches). And there is a strong correlation between
the use of ECC and the checking in other parts of the system.
Even on my home computers, I specify a decent configuration - and,
of course, two disks. Doesn't everyone use automated disk-to-disk
backup? :-)
Regards,
Nick Maclaren.
Kim Enkovaara
>
> Yes, but remember, we have somewhat different situation in the US that
> we have to take into account. We have a much larger (physically)
> country and it is more spread out than most of Europe. We have to take
> this into account.
In terms of population density, US is twice as dense compared to where I
live for example (Finland).
Population density for US is 31 people/km^2, in sweden the same figure
is 20, in finland 16 and in norway 12. The number for whole europe would
be 114 which is quite high tough.
--Kim
Bill Todd
> In article <rtGdnSDlu6zhvtbX...@metrocastcablevision.com>,
> Bill Todd <bill...@metrocast.net> wrote:
>> Thomas Womack wrote:
>>> In article <9bGdnWBP3Ni1NdfX...@metrocastcablevision.com>,
>>>
>>> And whilst that left the FS structure comprehensively rogered, I was
>>> able to recover the interesting directories by grepping /dev/hda to
>>> find the dirent as a cluster of recognisable filenames, then following
>>> the lists of sectors from there to get the files back.
>> That kind of error (and subsequent effort) may be acceptable to you for
>> your personal data, but most businesses would likely not find it so -
>> especially when they could guard against it relatively inexpensively.
>> And recovery from backup is becoming increasingly unpleasant as data
>> sizes burgeon, leaving aside whatever new data accumulated since the
>> most recent backup that can't be recovered that way.
>
> You have given two reasons to want checking on memory (and, as Rick
> says, caches - and elsewhere). Those of us who are professionals
> and/or very experienced don't want to waste the time trying to track
> down obscure errors that can easily and automatically be detected.
> It is a common delusion of the inexperienced that memory errors are
> always 'hard' - 90% of them are, but some are absolute b*gg*rs.
Another reason to favor such checks is simply to help eliminate that
possibility when something mysterious needs to be tracked down. This is
one reason why many people find Sun's ZFS file system so attractive:
undetected (as opposed to detected but uncorrectable) disk errors are
extremely rare but can wreak havoc when they occur, so ZFS's end-to-end
checks (similar to those which NetApp quietly instituted quite a while
ago) help eliminate that possibility when debugging a problem (they also
catch more common problems like link errors due to bad connections that
slip by the CRC checks on the links).
- bill
nm...@cam.ac.uk
Spot on. Been there - torn my hair out investigating that!
Regards,
Nick Maclaren.
Terje Mathisen
>> checks (similar to those which NetApp quietly instituted quite a while
>> ago) help eliminate that possibility when debugging a problem (they also
>> catch more common problems like link errors due to bad connections that
>> slip by the CRC checks on the links).
>
> Spot on. Been there - torn my hair out investigating that!
You need real end-to-end checks everywhere, the simplistic tcpip
checksum has caught (mostly silently) a _lot_ of faulty network hardware
over the years.
I've reported previously about how we had error rates of around one
wrong byte per 1-5 GB of data, most probably caused by a race condition
in an ethernet controller fifo.
We found the errors by doing full readback & compare, until we switched
on end-to-end network checksums.
With tcpip checksum offload those error would probably not have been
detected, which is why don't totally trust those tcpip offload engines
most fast network cads have these days.
Robert Myers
> Another reason to favor such checks is simply to help eliminate that
> possibility when something mysterious needs to be tracked down. This is
> one reason why many people find Sun's ZFS file system so attractive:
> undetected (as opposed to detected but uncorrectable) disk errors are
> extremely rare but can wreak havoc when they occur, so ZFS's end-to-end
> checks (similar to those which NetApp quietly instituted quite a while
> ago) help eliminate that possibility when debugging a problem (they also
> catch more common problems like link errors due to bad connections that
> slip by the CRC checks on the links).
Some routine integrity checks implemented in software probably would
have saved me from most of my problems (network errors and disk
errors). I don't think ECC would have helped much, and I don't need
mission-critical RAS.
In fact, as a result of this conversation, I think I'm going to add
some easy safety nets and warning buzzers. If there are memory
problems, I suspect they'd be swept up in the net.
Robert.
Kai Harrekilde-Petersen
> On Tue, 30 Jun 2009 03:56:15 +0100, Andrew Reilly
> <andrew-...@areilly.bpc-users.org> wrote:
>
>> On Mon, 29 Jun 2009 19:32:26 -0700, MitchAlsup wrote:
>>
>>> It is currently thought that
>>> unplugging wall warts (those 9V and 12V plastic boxes that plug into the
>>> AC and powed portable devices) when not in use would eliminate the need
>>> for 25-50 coal fired power plants over here in the USA.
>>
>> I wonder whether that holds for the newer, switch-mode ones, or just the
>> old iron-cored transformers?
>
> I think it quite specifically doesn't, just as most TVs on stand-by
> now consume about 1W rather than the received wisdom of "almost as
> much as they do when they are on".
Do you have any data to support that claim?
I recently bought a 40" Sony LCD TV, which claims 0.17W standby power.
Guess what, I measured the standby power to be 14W, unless I power it
off - then it falls to <0.1W - not exactly what I had expected.
Kai
--
Kai Harrekilde-Petersen <khp(at)harrekilde(dot)dk>
Terje Mathisen
> I recently bought a 40" Sony LCD TV, which claims 0.17W standby power.
> Guess what, I measured the standby power to be 14W, unless I power it
> off - then it falls to <0.1W - not exactly what I had expected.
That's a known bug with most current sets:
Standby power is indeed quite low, however by default the set will try
to receive the electronic program info, and to do that it turns itself
on for a significant fraction of each second. :-(
The place I read about this they claimed that US authorities considered
modifying the criteria to be what end users would get in a default
configuration.
Duh!
nm...@cam.ac.uk
Robert Myers <rbmye...@gmail.com> wrote:
>
>Some routine integrity checks implemented in software probably would
>have saved me from most of my problems (network errors and disk
>errors). I don't think ECC would have helped much, and I don't need
>mission-critical RAS.
>
>In fact, as a result of this conversation, I think I'm going to add
>some easy safety nets and warning buzzers. If there are memory
>problems, I suspect they'd be swept up in the net.
From actual experience, don't bet on it :-( They will detect most
hard and frequent errors, but a lot of damage can occur before the
first visible sign.
But I agree with your strategy - ECC is no substitute for such checks.
Regards,
Nick Maclaren.
Ken Hagan
<k...@harrekilde.dk> wrote:
> "Ken Hagan" <K.H...@thermoteknix.com> writes:
>
>> I think it quite specifically doesn't, just as most TVs on stand-by
>> now consume about 1W rather than the received wisdom of "almost as
>> much as they do when they are on".
>
> Do you have any data to support that claim?
I've measured my own TV (a late 1990s Panasonic model) at 1W with one
of those gizmos. It's a rather small dataset, I'll grant you that.
I've also had a few friends report similar findings on their own kit,
although some brands are better than others.
> I recently bought a 40" Sony LCD TV, which claims 0.17W standby power.
> Guess what, I measured the standby power to be 14W, unless I power it
> off - then it falls to <0.1W - not exactly what I had expected.
Not what I'd expect either. Presumably this is substantially less than
the "full on" consumption, but it is still disappointing.
If Terje is correct, that's a very lame design. Most TVs will be on at
least once each day and listings cover the next week or so. If you want
a full set of listings for "today" (or even most of the coming week) as
soon as you turn the set on, then simply *remembering* yesterday's
information would be adequate.
Kai Harrekilde-Petersen
> On Wed, 01 Jul 2009 18:20:41 +0100, Kai Harrekilde-Petersen
> <k...@harrekilde.dk> wrote:
>
>> "Ken Hagan" <K.H...@thermoteknix.com> writes:
>>
>>> I think it quite specifically doesn't, just as most TVs on stand-by
>>> now consume about 1W rather than the received wisdom of "almost as
>>> much as they do when they are on".
>>
>> Do you have any data to support that claim?
>
> I've measured my own TV (a late 1990s Panasonic model) at 1W with one
> of those gizmos. It's a rather small dataset, I'll grant you that.
> I've also had a few friends report similar findings on their own kit,
> although some brands are better than others.
I measured <1W on my 1999 model TV (25" Sony, FWIW). But I would find
it hard to argue that it's a "current" TV - rather an oudated TV.
>> I recently bought a 40" Sony LCD TV, which claims 0.17W standby power.
>> Guess what, I measured the standby power to be 14W, unless I power it
>> off - then it falls to <0.1W - not exactly what I had expected.
>
> Not what I'd expect either. Presumably this is substantially less than
> the "full on" consumption, but it is still disappointing.
Full on is on the order of 170-200W. I didn't pay much attention to
this number, since I don't watch TV for hours on end.
> If Terje is correct, that's a very lame design. Most TVs will be on at
> least once each day and listings cover the next week or so. If you want
> a full set of listings for "today" (or even most of the coming week) as
> soon as you turn the set on, then simply *remembering* yesterday's
> information would be adequate.
Indeed. I wonder if I can turn off the update of the EPG in some way.
Stephen Sprunk
> Stephen Sprunk wrote:
>> Stephen Fuld wrote:
>>> Robert Myers wrote:
>>>> Something like the fleet fuel efficiency averages of US CAFE standards
>>>> would work pretty well.
>>> No, they don't, or at least didn't. We have experience with these and
>>> they don't reduce gasoline consumption as much as you would think. The
>>> reason is that given higher MPG cars, the price of making a trip
>>> declines and, surprise, surprise, people drive more. They move farther
>>> away from work, make more trips, etc.
>>>
>>> The most reliable way to decrease the usage of something is to raise its
>>> price. When gasoline was $4.00 a gallon, you couldn't buy a Toyota
>>> Prius. When it dropped to about $2.30, you couldn't sell one.
>>
>> Logically, though, increasing the price of gas will lead to customers
>> buying higher-mileage cars, with the same result.
>
> Not quite. Higher mileage cars plus higher cost gas is, for sake of
> argument cost neutral for a given trip. Just higher mileage cars with
> no increased gas cost means lower cost for a longer trip.
True, but you're assuming that lowering the per-mile cost will induce
people to drive more. To a point, I'm sure that's correct, but it is
mitigated by the _time_ that longer trips take, which is unaffected by
the price of gas or their car's mileage. Most gas is consumed
commuting, and the average commute all over the world is 45 minutes
(each way), regardless of mode or price. Making those 45 minutes cost
half as much doesn't mean everyone will move an hour and a half away
from work.
>> The problem with that, or with your original plan, is that any US
>> politician who voted for it would find himself out of a job after the
>> next election, and they know it. The American public has grown
>> dependent on cheap oil and, like any addict, will lash out at anyone who
>> gets in the way of their fix -- even if they know, intellectually, that
>> getting sober is better for them long-term.
>>
>> The other issue, which I've never seen fully discussed (usually because
>> debates are so heated that they never get there) is what to do with all
>> that "carbon tax" revenue...
>
> You missed one of my big points. The 100% of the revenue from the
> increased gas tax going back to the consumer in the form of a per capita
> creditable tax credit.
I was speaking to "carbon tax" plans in general.
> Thus, it is cost neutral to the person using the average amount of gas.
> And it is a tax that can be avoided by buying a more fuel efficient car,
> etc. The hope is that this would allow the Republican's to vote for it,
> as it isn't a net tax increase
There's a lot of folks, from both parties, ticked off at the idea
Washington takes our money and then makes us beg to get it back. And,
realistically, what are the odds that the refund will last beyond the
next budget crisis?
> and the Democrats would accept is even though they don't get any more
> revenue out of it because they want to see reduction in gas usage.
The tree-huggers you're thinking of want all fossil fuel (and, in
extreme cases, all fuel) consumption eliminated, not just a penalty on
folks who consume "above average" amounts.
> This is designed explicitly to solve the political problems you mention
> above.
IMHO, it'll never fly with folks from _either_ side, nor with the
plurality of the populace that is independent.
>> Nuclear power is our best choice in the short term; it's cheap, it's
>> reliable, it's well understood, there's plenty of fuel, and it doesn't
>> have any net emissions. The tree-huggers are against it because of the
>> "waste" storage problem, but the only reason we _have_ a "waste" problem
>> is Carter's ban on breeder reactors which can use that "waste" as fuel.
>
> Agreed, but that is a different issue from petroleum usage, which was
> the subject of my first post.
With plug-in hybrids being the obvious next step, and battery technology
approaching the point that all-electric cars are becoming possible, you
cannot separate automotive consumption from general energy consumption.
It's becoming just one more consumer plugged into the grid.
There are also folks working on turning coal into oil, which will tie
the two markets very closely. And, of course, if this stupid hydrogen
car idea ever takes off, that is going to be sucking up electricity from
the grid as well, either directly for hydrolysis or indirectly by
competing for natural gas with peaking plants (a major source of
electricity).
You just can't separate them, though it's attractive to pretend that you
can because it makes the models simpler. In the end, though, it's all
just energy in arbitrary form.
S
--
Stephen Sprunk "Stupid people surround themselves with smart
CCIE #3723 people. Smart people surround themselves with
K5SSS smart people who disagree with them." --Isaac Jaffe
Stephen Sprunk
> Morten Reistad wrote:
>> The current hybrids still think petrol, or they go all overboard with
>> all-electric. We don't have the batteries for this, yet. But one of
>> those all-electric drive trains, with added integrated combustion
>> engine and generator, plus some battery capacity could revolutionise
>> automotion.
>
> That is called the Chevy Volt. It is called an "extended range
> electric".
It's called a plug-in hybrid, and many different companies are working
on them. It's just a series hybrid drive train with the possibility of
an external charging source.
Chevy will probably be first to market, with the Volt, but they aren't
the only ones with such a car under development, and after-market
conversion kits for the Prius have been available for years.
> All electric drive train with plug in rechargeable batteries
> and the 1.4 liter engine from their European division connected to a
> generator. Range on batteries is projected to be 200 miles with
> automatic start-up of the gas engine for longer trips.
Um, no. Range on battery only is claimed to be 40 miles, which covers
75% of Americans' daily commute. The original design had a 12gal fuel
tank, which would have given a total range of 640mi, but rumors have it
they've since cut the tank back to 6-7gal, which would reduce the total
range to only 340-390mi. (Figures assume GM's claimed 50mpg is true.)
>> After all, we use a significant part of time in our cars
>> near home, work or other places where we could charge a car; even if
>> that radius is down to 20 km. Having plug-ins that could handle this
>> driving would make a real dent in petrol consumption.
>
> That's GM's plan.
More importantly, GM's (actually, Obama's) long-term plan is to make
_all_ cars plug-in hybrids. So far, though, the automakers are just
testing the waters with expensive dedicated models or special packages
that add thousands of dollars to the price.
MitchAlsup
> Chevy will probably be first to market, with the Volt, but they aren't
> the only ones with such a car under development, and after-market
> conversion kits for the Prius have been available for years.
Tesla anyone?
Stephen Fuld
> Stephen Fuld wrote:
>> Morten Reistad wrote:
>>> The current hybrids still think petrol, or they go all overboard with
>>> all-electric. We don't have the batteries for this, yet. But one of
>>> those all-electric drive trains, with added integrated combustion
>>> engine and generator, plus some battery capacity could revolutionise
>>> automotion.
>> That is called the Chevy Volt. It is called an "extended range
>> electric".
>
> It's called a plug-in hybrid, and many different companies are working
> on them. It's just a series hybrid drive train with the possibility of
> an external charging source.
No, the Chevy Volt is not an extended range hybrid. The gasoline engine
does not connect to the drive train and is only used to recharge the
batteries. I guess there is some confusion, (Wikipedia called it a
hybrid) but I think it is useful to distinguish between cars that have
the gas engine partially responsible for some of the direct drive and
ones that don't. Not having the direct connection saves a lot of fairly
complex mechanical bits and pieces, and allows easier physical placement
of the gasoline engine, or eventual replacement of that engine with
better technology (a turbine, or even fuel cells) sometime in the future.
> Chevy will probably be first to market, with the Volt, but they aren't
> the only ones with such a car under development, and after-market
> conversion kits for the Prius have been available for years.
I agree that there are existent conversion kits for "typical" hybrids
and that extended range hybrids(basically "typical" hybrids with more
batteries and a plug in charger) will soon be available fro several
manufacturers, but that is not what the Volt is going to be.
>> All electric drive train with plug in rechargeable batteries
>> and the 1.4 liter engine from their European division connected to a
>> generator. Range on batteries is projected to be 200 miles with
>> automatic start-up of the gas engine for longer trips.
>
> Um, no. Range on battery only is claimed to be 40 miles, which covers
> 75% of Americans' daily commute.
You're right. I could swear I head and interview with Bob Lutz claiming
200, but apparently I was wrong.
snip
> More importantly, GM's (actually, Obama's) long-term plan is to make
> _all_ cars plug-in hybrids. So far, though, the automakers are just
> testing the waters with expensive dedicated models or special packages
> that add thousands of dollars to the price.
Well, even if you discount the cost of the batteries, which is high, but
expected to come down, having both fairly large electric and gasoline
engines seems more costly than having only one. But if you save enough
on the gas, that extra cost will be made up eventually.
Stephen Fuld
Tesla is a pure electric - no gasoline engine at all. Thus it is in a
different class than either the Volt or say a Prius converted to a
plug-in hybrid.
I'm impressed with the Tesla, but going from a few $100,000 sports cars
to a mass production vehicle is a big step. I hope they make it.
Stephen Fuld
> Stephen Fuld wrote:
>> Stephen Sprunk wrote:
>>> Stephen Fuld wrote:
>>>> Robert Myers wrote:
>>>>> Something like the fleet fuel efficiency averages of US CAFE standards
>>>>> would work pretty well.
>>>> No, they don't, or at least didn't. We have experience with these and
>>>> they don't reduce gasoline consumption as much as you would think. The
>>>> reason is that given higher MPG cars, the price of making a trip
>>>> declines and, surprise, surprise, people drive more. They move farther
>>>> away from work, make more trips, etc.
>>>>
>>>> The most reliable way to decrease the usage of something is to raise its
>>>> price. When gasoline was $4.00 a gallon, you couldn't buy a Toyota
>>>> Prius. When it dropped to about $2.30, you couldn't sell one.
>>> Logically, though, increasing the price of gas will lead to customers
>>> buying higher-mileage cars, with the same result.
>> Not quite. Higher mileage cars plus higher cost gas is, for sake of
>> argument cost neutral for a given trip. Just higher mileage cars with
>> no increased gas cost means lower cost for a longer trip.
>
> True, but you're assuming that lowering the per-mile cost will induce
> people to drive more.
Its not an assumption, but what happened the last time we significantly
increased CAFE standards.
> To a point, I'm sure that's correct, but it is
> mitigated by the _time_ that longer trips take, which is unaffected by
> the price of gas or their car's mileage.
I agree that time is a mitigating issue. But it doesn't eliminate the
effect. this is based on past history.
> Most gas is consumed
> commuting, and the average commute all over the world is 45 minutes
> (each way), regardless of mode or price. Making those 45 minutes cost
> half as much doesn't mean everyone will move an hour and a half away
> from work.
I don't have a good source to refer you to, but I believe that average
commute in the US had been increasing prior to the latest economic mess.
>>> The problem with that, or with your original plan, is that any US
>>> politician who voted for it would find himself out of a job after the
>>> next election, and they know it. The American public has grown
>>> dependent on cheap oil and, like any addict, will lash out at anyone who
>>> gets in the way of their fix -- even if they know, intellectually, that
>>> getting sober is better for them long-term.
>>>
>>> The other issue, which I've never seen fully discussed (usually because
>>> debates are so heated that they never get there) is what to do with all
>>> that "carbon tax" revenue...
>> You missed one of my big points. The 100% of the revenue from the
>> increased gas tax going back to the consumer in the form of a per capita
>> creditable tax credit.
>
> I was speaking to "carbon tax" plans in general.
If you mean something like the cap and trade plan passed by the House, I
am missing some knowledge about it. Perhaps you can help me. I
understand how it is supposed to work for fixed sources like power
plants, but for gasoline, who buys the credits? The oil companies?
that doesn't seem right. I heard one estimate that it would raise the
cost of gas by about 35 cents, which is hardly enough to make much
difference.
>> Thus, it is cost neutral to the person using the average amount of gas.
>> And it is a tax that can be avoided by buying a more fuel efficient car,
>> etc. The hope is that this would allow the Republican's to vote for it,
>> as it isn't a net tax increase
>
> There's a lot of folks, from both parties, ticked off at the idea
> Washington takes our money and then makes us beg to get it back.
Agreed. That is why I proposed that no has to beg. It comes back
automatically.
> And,
> realistically, what are the odds that the refund will last beyond the
> next budget crisis?
You are more cynical than I am. :-(
>> and the Democrats would accept is even though they don't get any more
>> revenue out of it because they want to see reduction in gas usage.
>
> The tree-huggers you're thinking of want all fossil fuel (and, in
> extreme cases, all fuel) consumption eliminated, not just a penalty on
> folks who consume "above average" amounts.
I think you are missing something here. If the people who consume above
average reduce, then the average comes down. Thus everyone has an
incentive to reduce. Its just that the people who use more have a
bigger incentive, which is what we want.
>> This is designed explicitly to solve the political problems you mention
>> above.
>
> IMHO, it'll never fly with folks from _either_ side, nor with the
> plurality of the populace that is independent.
You may be right, but trying is better than implementing something that
we know, by experience won't work.
>
>>> Nuclear power is our best choice in the short term; it's cheap, it's
>>> reliable, it's well understood, there's plenty of fuel, and it doesn't
>>> have any net emissions. The tree-huggers are against it because of the
>>> "waste" storage problem, but the only reason we _have_ a "waste" problem
>>> is Carter's ban on breeder reactors which can use that "waste" as fuel.
>> Agreed, but that is a different issue from petroleum usage, which was
>> the subject of my first post.
>
> With plug-in hybrids being the obvious next step, and battery technology
> approaching the point that all-electric cars are becoming possible, you
> cannot separate automotive consumption from general energy consumption.
> It's becoming just one more consumer plugged into the grid.
You are more optimistic than I am about the rapid adoption of the
various electric transportation technologies. I have also heard
conflicting claims about how much additional baseline load will be put
on the power grid.
> There are also folks working on turning coal into oil, which will tie
> the two markets very closely.
Yeah, and they have been for decades. Sort of like clean coal and
nuclear fusion power.
> And, of course, if this stupid hydrogen
> car idea ever takes off, that is going to be sucking up electricity from
> the grid as well, either directly for hydrolysis or indirectly by
> competing for natural gas with peaking plants (a major source of
> electricity).
I agree the Hydrogen car is a non-starter for several reasons.
> You just can't separate them, though it's attractive to pretend that you
> can because it makes the models simpler. In the end, though, it's all
> just energy in arbitrary form.
Perhaps in the end, but we are far from the end. For at least the next
decade, I predict there will still be a significant difference. Of
course, you are free to make your own predictions.
Stephen Sprunk
> Stephen Sprunk wrote:
>> Stephen Fuld wrote:
>>> That is called the Chevy Volt. It is called an "extended range
>>> electric".
>>
>> It's called a plug-in hybrid, and many different companies are working
>> on them. It's just a series hybrid drive train with the possibility of
>> an external charging source.
>
> No, the Chevy Volt is not an extended range hybrid. The gasoline engine
> does not connect to the drive train and is only used to recharge the
> batteries. I guess there is some confusion, (Wikipedia called it a
> hybrid) but I think it is useful to distinguish between cars that have
> the gas engine partially responsible for some of the direct drive and
> ones that don't. Not having the direct connection saves a lot of fairly
> complex mechanical bits and pieces, and allows easier physical placement
> of the gasoline engine, or eventual replacement of that engine with
> better technology (a turbine, or even fuel cells) sometime in the future.
You're thinking of parallel hybrids, where the electric motor/generator
is in parallel with the ICE. In a series hybrid, the generator and
motor are in series with the ICE.
(ICE = Internal Combustion Engine, if you aren't familiar with that TLA)
>> Chevy will probably be first to market, with the Volt, but they aren't
>> the only ones with such a car under development, and after-market
>> conversion kits for the Prius have been available for years.
>
> I agree that there are existent conversion kits for "typical" hybrids
> and that extended range hybrids(basically "typical" hybrids with more
> batteries and a plug in charger) will soon be available fro several
> manufacturers, but that is not what the Volt is going to be.
See above. All of the hybrids on the market today are parallel; the
Volt (and many other planned plug-in models) will be series.
>>> All electric drive train with plug in rechargeable batteries
>>> and the 1.4 liter engine from their European division connected to a
>>> generator. Range on batteries is projected to be 200 miles with
>>> automatic start-up of the gas engine for longer trips.
>>
>> Um, no. Range on battery only is claimed to be 40 miles, which covers
>> 75% of Americans' daily commute.
>
> You're right. I could swear I head and interview with Bob Lutz claiming
> 200, but apparently I was wrong.
You're probably thinking of the Tesla Roadster, which reportedly has a
range of 244mi.
Lutz was on Leno a couple of months ago, and Leno was (stupidly) bashing
him for not being able to make a car with the same range for a quarter
the price, despite the fact that the batteries Tesla's using cost more
than the Volt's entire target selling price.
>> More importantly, GM's (actually, Obama's) long-term plan is to make
>> _all_ cars plug-in hybrids. So far, though, the automakers are just
>> testing the waters with expensive dedicated models or special packages
>> that add thousands of dollars to the price.
>
> Well, even if you discount the cost of the batteries, which is high, but
> expected to come down, ...
This is, IMHO, a great area for part of any "carbon tax" receipts to be
spent: give the money to the NSF for a series of "X Prize" style battery
competitions to improve efficiency and reduce cost.
It's happening on its own, with significant advances being made every
year, but a few billion dollars in prize money will motivate a lot of
really smart people to speed things up. (And, unlike research grants, a
competition guarantees results or nothing is spent; even fiscal
conservatives like me would have a hard time arguing with that.)
> ... having both fairly large electric and gasoline engines seems more
> costly than having only one.
An electric engine is commonly called a "motor".
Yes, a series hybrid needs to have a much larger generator and motors,
since the ICE's full power must flow through both to reach the wheels,
unlike with a parallel hybrid.
OTOH, it does remove the need for a lot of heavy, complicated parts such
as the transmission, and it allows the ICE to be smaller and tuned to
run for optimal efficiency at all times (when it's not off completely).
It also makes much more sense when you start thinking of plug-in
models, where you're not expecting the ICE to be used much in the first
place; you want the ICE and generator to be as small as possible so you
can afford more size/weight for the batteries.
> But if you save enough on the gas, that extra cost will be made up
> eventually.
With the Volt running 50mpg on the ICE, something GM's never managed to
accomplish in the US market, and consuming no gas at all for the first
40mi after a charge (which, again, is enough for 75% of Americans' daily
commute), it's pretty easy to see where the expected payback is.
Of course, your home electric bill will go up if you're recharging your
car every day, but, per kWh of motive output, grid power is much cheaper
and produces far less pollution than gas. If increased demand causes
the grid price to go up, that will encourage investment in
pollution-free energy generation (whether wind, solar, nuclear, etc.);
OTOH, that would happen anyway if even a minimal carbon tax were passed.
Stephen Fuld
> Stephen Fuld wrote:
>> Stephen Sprunk wrote:
>>> Stephen Fuld wrote:
>>>> That is called the Chevy Volt. It is called an "extended range
>>>> electric".
>>> It's called a plug-in hybrid, and many different companies are working
>>> on them. It's just a series hybrid drive train with the possibility of
>>> an external charging source.
>> No, the Chevy Volt is not an extended range hybrid. The gasoline engine
>> does not connect to the drive train and is only used to recharge the
>> batteries. I guess there is some confusion, (Wikipedia called it a
>> hybrid) but I think it is useful to distinguish between cars that have
>> the gas engine partially responsible for some of the direct drive and
>> ones that don't. Not having the direct connection saves a lot of fairly
>> complex mechanical bits and pieces, and allows easier physical placement
>> of the gasoline engine, or eventual replacement of that engine with
>> better technology (a turbine, or even fuel cells) sometime in the future.
>
> You're thinking of parallel hybrids, where the electric motor/generator
> is in parallel with the ICE. In a series hybrid, the generator and
> motor are in series with the ICE.
I'd never heard the terms serial and parallel applied to hybrids, but if
that is the accepted terminology, then OK. In any event, it appears our
disagreement was not of substance, but of terminology.
> (ICE = Internal Combustion Engine, if you aren't familiar with that TLA)
I figured it out, but thanks. :-)
>>> Chevy will probably be first to market, with the Volt, but they aren't
>>> the only ones with such a car under development, and after-market
>>> conversion kits for the Prius have been available for years.
>> I agree that there are existent conversion kits for "typical" hybrids
>> and that extended range hybrids(basically "typical" hybrids with more
>> batteries and a plug in charger) will soon be available fro several
>> manufacturers, but that is not what the Volt is going to be.
>
> See above. All of the hybrids on the market today are parallel; the
> Volt (and many other planned plug-in models) will be series.
And AFAIK, the Volt will be the first series one.
>>>> All electric drive train with plug in rechargeable batteries
>>>> and the 1.4 liter engine from their European division connected to a
>>>> generator. Range on batteries is projected to be 200 miles with
>>>> automatic start-up of the gas engine for longer trips.
>>> Um, no. Range on battery only is claimed to be 40 miles, which covers
>>> 75% of Americans' daily commute.
>> You're right. I could swear I head and interview with Bob Lutz claiming
>> 200, but apparently I was wrong.
>
> You're probably thinking of the Tesla Roadster, which reportedly has a
> range of 244mi.
>
> Lutz was on Leno a couple of months ago, and Leno was (stupidly) bashing
> him for not being able to make a car with the same range for a quarter
> the price, despite the fact that the batteries Tesla's using cost more
> than the Volt's entire target selling price.
>
>>> More importantly, GM's (actually, Obama's) long-term plan is to make
>>> _all_ cars plug-in hybrids. So far, though, the automakers are just
>>> testing the waters with expensive dedicated models or special packages
>>> that add thousands of dollars to the price.
>> Well, even if you discount the cost of the batteries, which is high, but
>> expected to come down, ...
>
> This is, IMHO, a great area for part of any "carbon tax" receipts to be
> spent: give the money to the NSF for a series of "X Prize" style battery
> competitions to improve efficiency and reduce cost.
I have no problem with an X prize type of thing, but that cost would be
trivial (in today's budget terms), and not require any additional taxes.
i.e. it would be lost in the rounding error.
> It's happening on its own, with significant advances being made every
> year, but a few billion dollars in prize money will motivate a lot of
> really smart people to speed things up. (And, unlike research grants, a
> competition guarantees results or nothing is spent; even fiscal
> conservatives like me would have a hard time arguing with that.)
>
>> ... having both fairly large electric and gasoline engines seems more
>> costly than having only one.
>
> An electric engine is commonly called a "motor".
>
> Yes, a series hybrid needs to have a much larger generator and motors,
> since the ICE's full power must flow through both to reach the wheels,
> unlike with a parallel hybrid.
>
> OTOH, it does remove the need for a lot of heavy, complicated parts such
> as the transmission, and it allows the ICE to be smaller and tuned to
> run for optimal efficiency at all times (when it's not off completely).
> It also makes much more sense when you start thinking of plug-in
> models, where you're not expecting the ICE to be used much in the first
> place; you want the ICE and generator to be as small as possible so you
> can afford more size/weight for the batteries.
Precisely. And it does allow easier substitution of a non ICE. I
understand that a turbine would have greater efficiency than an ICE See,
I am learning the TLAs :-)
Stephen Sprunk
> Stephen Sprunk wrote:
>> Stephen Fuld wrote:
>>>> Chevy will probably be first to market, with the Volt, but they aren't
>>>> the only ones with such a car under development, and after-market
>>>> conversion kits for the Prius have been available for years.
>>>
>>> I agree that there are existent conversion kits for "typical" hybrids
>>> and that extended range hybrids(basically "typical" hybrids with more
>>> batteries and a plug in charger) will soon be available fro several
>>> manufacturers, but that is not what the Volt is going to be.
>>
>> See above. All of the hybrids on the market today are parallel; the
>> Volt (and many other planned plug-in models) will be series.
>
> And AFAIK, the Volt will be the first series one.
First in the auto industry, probably, though we'll have to see if GM has
any more delays or if some other maker can push things up any.
Diesel locomotives have been series hybrids for several decades; none of
this is truly novel technology.
>>>> More importantly, GM's (actually, Obama's) long-term plan is to make
>>>> _all_ cars plug-in hybrids. So far, though, the automakers are just
>>>> testing the waters with expensive dedicated models or special packages
>>>> that add thousands of dollars to the price.
>>> Well, even if you discount the cost of the batteries, which is high, but
>>> expected to come down, ...
>>
>> This is, IMHO, a great area for part of any "carbon tax" receipts to be
>> spent: give the money to the NSF for a series of "X Prize" style battery
>> competitions to improve efficiency and reduce cost.
>
> I have no problem with an X prize type of thing, but that cost would be
> trivial (in today's budget terms), and not require any additional taxes.
> i.e. it would be lost in the rounding error.
That depends on how many series there are, how many contests are in
each, and how big the prizes must be to motivate the potential
contestants. There are at least a dozen different fields, not just
batteries, where one could justify spending public money to make
industry (and consumers) "greener". Even in fields where green
alternatives exist, the price is usually much higher and it'd be worth
spending money to cut the gap (or even reverse it) so that consumers
would switch. And I suspect you're going to need hundreds of millions,
perhaps billions, for each prize if the goals are going to be
significant advances in the state of the art...
>> Yes, a series hybrid needs to have a much larger generator and motors,
>> since the ICE's full power must flow through both to reach the wheels,
>> unlike with a parallel hybrid.
>>
>> OTOH, it does remove the need for a lot of heavy, complicated parts such
>> as the transmission, and it allows the ICE to be smaller and tuned to
>> run for optimal efficiency at all times (when it's not off completely).
>> It also makes much more sense when you start thinking of plug-in
>> models, where you're not expecting the ICE to be used much in the first
>> place; you want the ICE and generator to be as small as possible so you
>> can afford more size/weight for the batteries.
>
> Precisely. And it does allow easier substitution of a non ICE. I
> understand that a turbine would have greater efficiency than an ICE
An ECE like a gas turbine would be good, though thermal efficiency isn't
much better than a diesel ICE. The main benefits come from simplicity,
reduced size, and fuel flexibility but are partially offset by increased
manufacturing cost (special materials, tight tolerances). Turbines
don't work in traditional cars or parallel hybrids because they can't
vary their speed quickly enough, but in a series hybrid where they just
turn a generator at a fixed, optimum speed...
Some folks have also envisioned a car whose hood and roof are covered
with solar panels, recharging the batteries for free while parked all
day in the office/mall/etc. parking lot. That could conceivably double
the electric range and/or reduce demand on the grid for charging.
There are also folks pushing for electrified freeways using some sort of
inductive track embedded in the pavement; cars would recharge themselves
as they drove and have virtually unlimited electric range.
There are lots of creative things one can do with a series hybrid which
aren't really feasible with a parallel hybrid.
Stephen Sprunk
> Stephen Sprunk wrote:
>> Stephen Fuld wrote:
>>>> Logically, though, increasing the price of gas will lead to customers
>>>> buying higher-mileage cars, with the same result.
>>>
>>> Not quite. Higher mileage cars plus higher cost gas is, for sake of
>>> argument cost neutral for a given trip. Just higher mileage cars with
>>> no increased gas cost means lower cost for a longer trip.
>>
>> True, but you're assuming that lowering the per-mile cost will induce
>> people to drive more.
>
> Its not an assumption, but what happened the last time we significantly
> increased CAFE standards.
To what period are you referring? CAFE started at 18mpg in 1978,
increased gradually to 27mpg in 1984, dropped back to 26mpg in 1986,
rose to 27.5mph in 1990, and has stayed there ever since.
I can pull up the average miles driven per capita, but it's kind of a
pain so I'd like to know which specific years to look at.
>> Most gas is consumed commuting, and the average commute all over the
>> world is 45 minutes (each way), regardless of mode or price. Making
>> those 45 minutes cost half as much doesn't mean everyone will move an
>> hour and a half away from work.
>
> I don't have a good source to refer you to, but I believe that average
> commute in the US had been increasing prior to the latest economic mess.
It's been relatively stable for the century we've been collecting the
data, and the same number keeps popping up in other countries' studies
as well. That's just how long, on average, human beings are willing to
commute to work -- psychology, not technology or economics.
What changes is how far one can go in that 45 minutes, and that goes up
and down with road capacity, congestion, development patterns, migration
of employers, etc. And, of course, it's just an average.
>>>> The other issue, which I've never seen fully discussed (usually because
>>>> debates are so heated that they never get there) is what to do with all
>>>> that "carbon tax" revenue...
>>>
>>> You missed one of my big points. The 100% of the revenue from the
>>> increased gas tax going back to the consumer in the form of a per capita
>>> creditable tax credit.
>>
>> I was speaking to "carbon tax" plans in general.
>
> If you mean something like the cap and trade plan passed by the House, I
> am missing some knowledge about it. Perhaps you can help me. I
> understand how it is supposed to work for fixed sources like power
> plants, but for gasoline, who buys the credits? The oil companies? that
> doesn't seem right. I heard one estimate that it would raise the cost
> of gas by about 35 cents, which is hardly enough to make much difference.
Cap&trade systems are a horribly complicated and convoluted system that
only barely resembles what I'm talking about.
A true carbon tax would quite simply apply an excise tax (at a fixed
rate per amount of carbon) on the production or import of any fossil
fuel. There would be no way for consumers to avoid it since it'd be
levied directly on the producers and included in the price. There would
be no "credits", just like one can't avoid the excise tax on liquor by
purchasing credits from someone who doesn't drink.
>>> Thus, it is cost neutral to the person using the average amount of gas.
>>> And it is a tax that can be avoided by buying a more fuel efficient car,
>>> etc. The hope is that this would allow the Republican's to vote for it,
>>> as it isn't a net tax increase
>>
>> There's a lot of folks, from both parties, ticked off at the idea
>> Washington takes our money and then makes us beg to get it back.
>
> Agreed. That is why I proposed that no has to beg. It comes back
> automatically.
Still, the idea of getting taxed just so that they can send the money
back to us is going to generate a lot of resistance.
>> And, realistically, what are the odds that the refund will last beyond
>> the next budget crisis?
>
> You are more cynical than I am. :-(
Perhaps. A pessimist is rarely disappointed, though.
>>> and the Democrats would accept is even though they don't get any more
>>> revenue out of it because they want to see reduction in gas usage.
>>
>> The tree-huggers you're thinking of want all fossil fuel (and, in
>> extreme cases, all fuel) consumption eliminated, not just a penalty on
>> folks who consume "above average" amounts.
>
> I think you are missing something here. If the people who consume above
> average reduce, then the average comes down.
OTOH, you're paying the people who consume below average, which gives
them more money to spend on gas. The result may well reduce deviation
from the average on _both_ sides or perhaps even increase it.
> Thus everyone has an incentive to reduce. Its just that the people who
> use more have a bigger incentive, which is what we want.
The people who use more tend to be the folks to whom the incentive will
matter least. Gas would have to cost five to ten times what it does
today before I'd really start to care enough about the cost to change my
behavior -- but you'd get the poor people out of their cars and onto the
buses and trains (or even bicycles) long before that.
>>> This is designed explicitly to solve the political problems you mention
>>> above.
>>
>> IMHO, it'll never fly with folks from _either_ side, nor with the
>> plurality of the populace that is independent.
>
> You may be right, but trying is better than implementing something that
> we know, by experience won't work.
Expending the effort on a plan we know won't pass will reduce the
attention that can be garnered by other plans in the future. In fact,
this is a common tactic of politicians: preempt a reasonable law you
don't like by proposing one so hideous that nobody will even consider
the reasonable one (i.e. guilt by association).
>> With plug-in hybrids being the obvious next step, and battery technology
>> approaching the point that all-electric cars are becoming possible, you
>> cannot separate automotive consumption from general energy consumption.
>> It's becoming just one more consumer plugged into the grid.
>
> You are more optimistic than I am about the rapid adoption of the
> various electric transportation technologies. I have also heard
> conflicting claims about how much additional baseline load will be put
> on the power grid.
It'd be absolutely massive, bordering on apocalyptic, if adoption is
sudden and widespread. That's one of the reasons that I support
building many more nuclear power plants, transmission capacity, etc.
before the wave hits. The folks in that industry are well aware of the
looming problem, and they'd be busy preparing for it if the damn
politicians would get out of their way.
Still, I don't think it's going to be "rapid". We have a long way to go
before the prices come down to reasonable levels and they become a
significant fraction of the fleet -- but I think that's a reasonable
thing to fund, since it benefits us all with cleaner air.
>> There are also folks working on turning coal into oil, which will tie
>> the two markets very closely.
>
> Yeah, and they have been for decades. Sort of like clean coal and
> nuclear fusion power.
The latter two are, for now, scientific theory. Oil from coal is proven
technology and was even used in Germany during WWII. However, it will
require tens of billions of dollars in investment to get the kind of
volume that would be required to make a difference, and foreign oil is
just too cheap (and too volatile) for anyone to invest in it. If the
price of oil stays above $100/bbl for long enough, though, you'll see
coal conversion plants being built all over the country, particularly in
places with types of coal that aren't wanted by power or steel plants.
Bernd Paysan
> An ECE like a gas turbine would be good, though thermal efficiency isn't
> much better than a diesel ICE.
Humid air turbines (especially TOPHAT) have a significantly better thermal
efficiency (~60%), in combination with an even better power-to-weight ratio.
So far, scaling down is a bit of a problem - 1MW, even 500kW works (whereas
a combined gas/steam engine works only well above 50MW). But this may be
just an implementation detail, the scaling down of normal turbines worked
down to the 30W region (MIT nano turbine). TOPHAT turbines also have a
larger range of efficient operating conditions as usual turbines.
> The main benefits come from simplicity,
> reduced size, and fuel flexibility but are partially offset by increased
> manufacturing cost (special materials, tight tolerances). Turbines
> don't work in traditional cars or parallel hybrids because they can't
> vary their speed quickly enough, but in a series hybrid where they just
> turn a generator at a fixed, optimum speed...
Turbines also don't work well in traditional cars, because small turbines
rotate so fast that mechanical transmission with gears and such have
problems. This is also something that doesn't pose a problem for a generator
directly attached to the axis.
--
Bernd Paysan
"If you want it done right, you have to do it yourself"
http://www.jwdt.com/~paysan/
Robert
> matter least. Gas would have to cost five to ten times what it does
> today before I'd really start to care enough about the cost to change my
> behavior -- but you'd get the poor people out of their cars and onto the
> buses and trains (or even bicycles) long before that.
This is an interesting bit. Efficient anything is great and all, if your
ROI is there.
But better still is just rational zoning. I am american, and things are
bad.
Houses here, mall there, grocery the other way. Offices way over the other
way,
with the factories down wind.. Cities are built for cars, not for people!
I have been in asia the last 6 years. Here is it much better. I can look
out the window
and see a quicky mart, a fruit shop, a bar, 2 restaurants, dry cleaner, etc.
In china I shared
my floor of a tower with a civil engineering firm, an insurance firm, an
export firm, and 2
other families house!
Mixed use. I walk everywhere. Take a moto every week to the big market,
and a taxi
with friends when going out. If everything is handy, why drive? No car
payment,
insurance, repairs, parking, depreciation, inspections, plates, etc.
And then there is telecommuting.
Why, exactly, do we need all of these cars?!?
If you look at personal energy usage, cars are maybe first? So, we get rid
of most via convenience..
Then heating/cooling, hot water - Solar thermal is much more effective than
solar power.
insulation is cheap.
refrigerators - shop across the street. Buy fresh. Maybe a small one for
beer and snacks.
for all the rest - big tv's, computers, well, getting down to small
potatoes but every bit
helps.
We do not need taxes, credits, or a change of life style.
We do need smarter cities, designed for people, optimized for convenience.
Put a bus or train stop in every kilometer, and things get better still.
Then the problem almost solves itself.
Rick Jones
> An ECE like a gas turbine would be good, though thermal efficiency isn't
> much better than a diesel ICE. The main benefits come from simplicity,
> reduced size, and fuel flexibility but are partially offset by increased
> manufacturing cost (special materials, tight tolerances). Turbines
> don't work in traditional cars or parallel hybrids because they can't
> vary their speed quickly enough, but in a series hybrid where they just
> turn a generator at a fixed, optimum speed...
How well do turbines handle potholes?
rick jones
--
portable adj, code that compiles under more than one compiler
these opinions are mine, all mine; HP might not want them anyway... :)
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...
Eugene Miya
Well most people don't have a clue what DARPA is and does (that includes
most people in comp.arch). The majority of the US DOD doesn't have a clue
about DARPA.
In article <b7022f05-2242-4bee...@y34g2000yqd.googlegroups.com>,
Robert Myers <rbmye...@gmail.com> wrote:
>On Jun 29, 8:10=A0pm, Andrew Reilly <andrew-newsp...@areilly.bpc-
>users.org> wrote:
>> On Mon, 29 Jun 2009 13:08:47 -0700, Robert Myers wrote:
>> > 1. There is little point in the DoD, whose needs change overnight,
>> > building hardware that can be programmed only with great difficulty by
>> > hard-to-find software specialists with a very high skill level,
>>
>> I don't get that. Why should the DoD require their systems to be built
>> by monkeys?
A number of amusing memoirs exist to answer that. Most uniformed
services, especially in the past, would have preferred monkeys.
>> NASA uses elite specialists to design their space systems,
If you only knew......
>> large-scale civil works are designed by elite specialists. That's how
>> good stuff works. =A0Of course, there's nothing wrong with having programs
>> to develop tools to help these guys, that's just common sense that
>> applies in all engineering fields. =A0Also no point calling for tenders
>>for systems that rely on luck or magic. =A0Maybe that's what you were saying.
>>
>My experience has been that code winds up in the hands of all kinds of
>people with all kinds of skill levels, almost no matter what kinds of
>controls are put into place. If you're doing cutting edge science or
...
>to some real-time event (a damaged shuttle circling the earth?), you
>may be out of luck.
8^(
>Software shouldn't be written by monkeys, but it should be written in
>a way that monkeys would have to work hard to do serious damage. That
Then it's a hardware problem.
>describes almost no serious code that I've come across, and anyone in
>the business has to have seen a software train wreck brought on by
>someone with good intentions who doesn't understand what they're
>doing.
Like the Disneyworld Monorail?
>Even wizards have a hard time coding concurrent processes safely,
>witness some of the scary discussions in this forum.
Comp.risks is only slightly better. I'm up at Microsoft next week.
>As to luck, almost all software depends on it, so far as I can tell,
>since the only software I know of that is routinely formally verified
>is flight control software.
Depends on lucky hardware. Flight software also has other constraints
and factors which most landlubbing programmers don't have such as
stripping dynamic memory out of Ada.
>> > 2. Software design methodologies and languages currently in use make
>> > most software highly resistant to extreme optimization and artificially
>> > limit the usefulness of energy-efficient approaches like stream
>> > computation. =A0Adapting to the installed base makes the problem hopeless.
>> > =A0Time to throw the installed base overboard and start with a clean sheet
>> > of paper.
Throwing the base out is sometimes insufficient. Sometimes you must
execute, terminate, or retire the humans (I mean management).
>> Doesn't that happen anyway? =A0With the possible exception of a few dusty-
>> deck numeric libraries, it seems to me that most code (perhaps outside of
>> banking) is rewritten every ten years or so anyway, just because that's
>> easier for the new hires than talking to the old guys who wrote the last
>> version. =A0How much code are we talking about in this installed base, an=
>d
>> who does it belong to? =A0
>>
>Eugene Miya would be the one to give an answer to that. "Installed
>base" is his phrase, at least in this forum.
I don't claim the phrase. It came from Datamation. Just track down IBM
mainframes types like Lynn. DARPA on the other hand, I have worked with
them. They aren't just about computers. They have their own misguidance.
C and C++ don't optimize as well as Fortran. You'd be amazed where
LINPACK gets used: the library not the benchmark.
>> A large, related problem, I think, is that quite a lot of research now
>> seems to finish with code and graphs: the actual workings never find
>> their way into published papers in english and maths. =A0That's going to =
>be
>> a big problem for those who would otherwise write new versions using new
>> tools, later.
>
>The problem is money. The pressure to get out a colored graph with
>the right answer is overwhelming. Failure to do so can be hazardous
>to funding. Advancing the common pool of knowledge, creating code
>that can be understood by future generations, or creating a solid
>foundation for future work all have low priorities because no one gets
>a pay raise for such contributions.
What graph?
Virtualization will eliminate the requirement for source and even much
object code.
>> I.e., I think that the installed base argument is mostly a red herring.
Bigger deal during the Cold War.
>> The new and interesting (research) problems are already being solved with
>> the new and interesting tools. =A0The new tools don't solve all of the
>> problems, of course, but useful progress seems to be being made. =A0(Check
>> outhttp://blip.tv/file/812787if you have a spare 152-odd minutes to get
>> an overview of how immutable-by-default data structures, and STM are
>> useful for concurrency in Clojure. =A0There are plenty of dissenting view=
>s
>> out there, of course...)
>
>I will find the spare minutes.
--
Looking for an H-912 (container).
Stephen Sprunk
> Stephen Sprunk <ste...@sprunk.org> wrote:
>> An ECE like a gas turbine would be good, though thermal efficiency isn't
>> much better than a diesel ICE. The main benefits come from simplicity,
>> reduced size, and fuel flexibility but are partially offset by increased
>> manufacturing cost (special materials, tight tolerances). Turbines
>> don't work in traditional cars or parallel hybrids because they can't
>> vary their speed quickly enough, but in a series hybrid where they just
>> turn a generator at a fixed, optimum speed...
>
> How well do turbines handle potholes?
I've never seen any specific studies on that. There have been several
prototype cars by various manufacturers over the years, and Chrysler had
a near-production run of their 1963 Turbine Car. AFAICT, though,
nobody's ever actually sold one to the general public, and automakers
generally don't release much operational data about their prototypes.
There are production buses with turbine engines in successful, daily use
in various countries. However, several other companies have failed to
bring turbine buses to market due to "reliability problems", which might
be due to how well a 50,000 RPM turbine handles (or doesn't handle)
potholes.
The closest you could get to real-world data about how turbines handle
sharp forces like that would be from carrier-based fighter aircraft, but
their maintenance regime is so different than consumer automobiles that
I doubt you could draw any useful conclusions.
EricP
> Rick Jones wrote:
>> Stephen Sprunk <ste...@sprunk.org> wrote:
>>> An ECE like a gas turbine would be good, though thermal efficiency isn't
>>> much better than a diesel ICE. The main benefits come from simplicity,
>>> reduced size, and fuel flexibility but are partially offset by increased
>>> manufacturing cost (special materials, tight tolerances). Turbines
>>> don't work in traditional cars or parallel hybrids because they can't
>>> vary their speed quickly enough, but in a series hybrid where they just
>>> turn a generator at a fixed, optimum speed...
>> How well do turbines handle potholes?
>
> I've never seen any specific studies on that. There have been several
> prototype cars by various manufacturers over the years, and Chrysler had
> a near-production run of their 1963 Turbine Car. AFAICT, though,
> nobody's ever actually sold one to the general public, and automakers
> generally don't release much operational data about their prototypes.
In 1999, EV1 series hybrid turbine + battery + electric motor
http://en.wikipedia.org/wiki/General_Motors_EV-1#EV1_series_hybrid
No word in that article about turbine reliability,
though they say it was manufactured by Williams International
so you might be able to chase down some numbers.
Eric
Robert Myers
> In article <b7022f05-2242-4bee-92fa-ee26ef1d5...@y34g2000yqd.googlegroups.com>,
> Robert Myers <rbmyers...@gmail.com> wrote:
>
> >On Jun 29, 8:10=A0pm, Andrew Reilly <andrew-newsp...@areilly.bpc-
> >users.org> wrote:
>
> >Software shouldn't be written by monkeys, but it should be written in
> >a way that monkeys would have to work hard to do serious damage. That
>
> Then it's a hardware problem.
>
Perfect software (were such a thing possible) wouldn't solve the
world's ills, but it would be really, really nice, even if hardware
remains inherently imperfect.
> C and C++ don't optimize as well as Fortran. You'd be amazed where
> LINPACK gets used: the library not the benchmark.
I don't think I would, really. It's a large class of problems, but
there is also a large class of (important) problems for which linpack-
style computing will never get accurate answers. Those are problems
where nonlinear interaction between the largest scales and the
smallest scales is decisive. It should be obvious, without going into
detail, that you can't do that right without global bandwidth.
Robert.
Bernd Paysan
> How well do turbines handle potholes?
Turbo-chargers are small turbines, too. I'm not aware of any significant
reliability problems associated with turbo-chargers and potholes. Standard
techniques to reduce forces can be applied.
As I suggested in another thread, TOPHAT gives you a higher efficiency, but
it also gives you more robustness in this case, too. A tophat turbine has an
operating pressure of about 20 bar, and can be operated with air bearings,
and at 20 bar, air bearings can stand quite high shock accelerations. A
turbine operated without excessive stress can work for hundreds thousand
hours, which is a lot more than you need for a car (10k hours operating is
sufficient for a car).
Stephen Sprunk
>> The people who use more tend to be the folks to whom the incentive will
>> matter least. Gas would have to cost five to ten times what it does
>> today before I'd really start to care enough about the cost to change my
>> behavior -- but you'd get the poor people out of their cars and onto the
>> buses and trains (or even bicycles) long before that.
>
> This is an interesting bit. Efficient anything is great and all, if
> your ROI is there.
It's not ROI per se, but whether your "incentives" are large enough to
make someone care. As people get richer, their energy consumption
increases, but it's still decreasing as a fraction of their total
income. All these "tax the rich into consuming less" plans assume it's
a constant or even growing fraction of their income.
> But better still is just rational zoning. I am american, and things are
> bad. Houses here, mall there, grocery the other way. Offices way over
> the other way, with the factories down wind.. Cities are built for cars,
> not for people!
>
> I have been in asia the last 6 years. Here is it much better. I can
> look out the window and see a quicky mart, a fruit shop, a bar, 2
> restaurants, dry cleaner, etc. In china I shared my floor of a tower
> with a civil engineering firm, an insurance firm, an export firm, and 2
> other families house!
I'm not sure I'd like to have mixed use within a single floor, but
within a building appears to be sufficient. The standard pattern in the
US before the 1950s was to have retail on the first floor of every
building and either commercial or residential on the floors above.
However, in a (quite successful) effort to force people into cars and
the suburbs, zoning laws were changed to make such mixed-use buildings
illegal. A given parcel of land is now either industrial, retail,
commercial (offices), or residential -- no mixing allowed.
> Mixed use. I walk everywhere. Take a moto every week to the big
> market, and a taxi with friends when going out. If everything is
> handy, why drive?
Exactly. I lived in a mixed-use urban area for years and I loved it; I
could walk to virtually every service and store I needed on a daily
basis. Scooters and motorcycles were extremely popular, and any store
with goods too large to carry on one offered delivery, usually for free.
> No car payment, insurance, repairs, parking, depreciation, inspections,
> plates, etc.
There were a few things I still needed a car for that won't go away,
plus there's a social stigma in the US to being carless, and
unfortunately cars have very high fixed costs. Once you have to pay
those fixed costs, you're tempted to use the car a lot more than you
really need to in order to "get your money's worth".
> And then there is telecommuting.
A reasonable solution for many professionals, at least a few days a
week. However, many jobs inherently require employees' physical
presence every day. Even jobs that are compatible still lose quite a
bit of productivity if you don't see your coworkers in person at least
once a week if for no other reason than the loss of social bonds. Email
and phone calls do have their benefits, but they also have their
limitations; a lot of things are simply done a lot more efficiently in
person.
> Why, exactly, do we need all of these cars?!?
In areas where mixed-use zoning is still (or newly) possible, you find
walkable neighborhoods, lower per capita energy consumption, lower car
ownership rates (because a family only needs one, not two or more), and
healthier people (because they walk more).
> Then heating/cooling, hot water - Solar thermal is much more effective
> than solar power.
True, but there's just not enough roof space to do it for high-density
buildings. And the time when hot water and air is needed most, just
before dawn, is when solar heating is least effective.
> insulation is cheap.
... if you put it in during construction. It costs a _lot_ more to add
into existing buildings, to the point where it's often cheaper to tear
them down and re-build than it is to retrofit.
> refrigerators - shop across the street. Buy fresh. Maybe a small one
> for beer and snacks.
Even if every person had a grocery store within walking distance, I
doubt you could convince them to give up their refrigerator/freezer or
even downsize significantly.
> for all the rest - big tv's, computers, well, getting down to small
> potatoes but every bit helps.
We have big TVs, computers, etc. for a reason. Unless you find a
better, cheaper way to solve those problems, they aren't going away.
> We do not need taxes, credits, or a change of life style.
What you're proposing _is_ a change of lifestyle. High-density,
mixed-use urban living is very different from low-density, single-use
suburban living. I've done both _in the same city_ and there's no
comparison.
Singles and empty-nesters will adopt the new lifestyle in fairly large
numbers if the artificial scarcity is solved and prices come down;
families with small children will be a much, much tougher sale.
> We do need smarter cities, designed for people, optimized for convenience.
> Put a bus or train stop in every kilometer, and things get better still.
A kilometer? Even in the 'burbs, we've got a bus stop every quarter
mile or so, feeding into train stations about a mile apart. Still,
traffic congestion isn't bad enough to make transit time-competitive
unless you're headed to (or through) downtown during rush hour.
> Then the problem almost solves itself.
It all rests on the zoning problem, and since that is quite deliberate
and with substantial political and economic powers behind it, I don't
predict much change.
Bill Todd
...
> It's not ROI per se, but whether your "incentives" are large enough to
> make someone care. As people get richer, their energy consumption
> increases, but it's still decreasing as a fraction of their total
> income. All these "tax the rich into consuming less" plans assume it's
> a constant or even growing fraction of their income.
It's possible that you mistook a general proposal for something more
specifically aimed. The idea is to tax *everyone* (at least on average)
into consuming less, using whatever rate it takes to drive down total
consumption by the desired amount.
It really won't matter if some of the richer people don't care as long
as total consumption decreases as needed. And in the process a good
deal of wealth will get transferred from those rich people who (by your
own definition) don't care about losing it to people who can benefit
significantly from it, thereby creating a desirable side-effect as well.
So it's a painless tax for those sufficiently rich to ignore it while
benefiting those sufficiently motivated by it to minimize their
consumption and by so doing get far more back in refunds than they pay
in fuel taxes. True, some small percentage of people will need to
change their actual lifestyle or means of livelihood (rather than merely
their choice of vehicle and/or recreational use of it) if they simply
can't afford the amount of fuel that these currently require, but that's
the price paid for having allowed our politicians to ignore the issue
for so long.
- bill
Robert
> It all rests on the zoning problem, and since that is quite deliberate
> and with substantial political and economic powers behind it, I don't
> predict much change.
It needs to be rational. Current cities are not, and the proposed
solutions are also flawed.
The status quo needs to change a bit. Eventually it will.
Making LED light bulbs is great, but maybe I save 1 K watt a day.
A car on the highway uses maybe 20 hp to maintain speed, this is near
20 KW. A hot water heater is a short circuit!
Green power would be nice, but first don't waste half of what you use!
Eugene Miya
Robert Myers <rbmye...@gmail.com> wrote:
>On Jul 6, 7:30=A0pm, eug...@cse.ucsc.edu (Eugene Miya) wrote:
>> In article <b7022f05-2242-4bee-92fa-ee26ef1d5...@y34g2000yqd.googlegroups=
>.com>,
>> Robert Myers =A0<rbmyers...@gmail.com> wrote:
>> >On Jun 29, 8:10=3DA0pm, Andrew Reilly <andrew-newsp...@areilly.bpc->> >a way that monkeys would have to work hard to do serious damage. =A0That
>>
>=A0 =A0 Then it's a hardware problem.
>>
>Perfect software (were such a thing possible) wouldn't solve the
>world's ills, but it would be really, really nice, even if hardware
>remains inherently imperfect.
I will remember that as I sit in Redmond at the end of next week.
Intel is, of course, just down the street from me.
>> C and C++ don't optimize as well as Fortran. =A0You'd be amazed where
>> LINPACK gets used: the library not the benchmark.
>
>I don't think I would, really. It's a large class of problems, but
>there is also a large class of (important) problems for which linpack-
>style computing will never get accurate answers. Those are problems
>where nonlinear interaction between the largest scales and the
>smallest scales is decisive. It should be obvious, without going into
>detail, that you can't do that right without global bandwidth.
One presumes said programmer interaction know linear from non-linear but
the average programmer would rather quickly cheat than nitty gritty stuff.
I was told an amusing 6600 story (LLL) but it would take too long to
repeat here. I have to drop a person at SFO shortly.
Robert Myers
> In article <48424573-f7d1-4440-b236-ccea8c717...@o7g2000yqb.googlegroups.com>,
> Robert Myers <rbmyers...@gmail.com> wrote:
>
> >I don't think I would, really. It's a large class of problems, but
> >there is also a large class of (important) problems for which linpack-
> >style computing will never get accurate answers. Those are problems
> >where nonlinear interaction between the largest scales and the
> >smallest scales is decisive. It should be obvious, without going into
> >detail, that you can't do that right without global bandwidth.
>
> One presumes said programmer interaction know linear from non-linear but
> the average programmer would rather quickly cheat than nitty gritty stuff.
> I was told an amusing 6600 story (LLL) but it would take too long to
> repeat here. I have to drop a person at SFO shortly.
>
"For a successful technology, reality must take precedence over
public relations, for nature cannot be fooled."
from "Appendix F - Personal observations on the reliability of the
Shuttle" by R. P. Feynman
(relegated to an appendix, of course)
And, from WIkipedia: The concept known as the law of the instrument,
Maslow's hammer, or a golden hammer is an over-reliance on a familiar
tool; as Abraham Maslow said in 1962, "When the only tool you have is
a hammer, it is tempting to treat everything as if it were a nail."
Robert.
Niels Jørgen Kruse
> The closest you could get to real-world data about how turbines handle
> sharp forces like that would be from carrier-based fighter aircraft, but
> their maintenance regime is so different than consumer automobiles that
> I doubt you could draw any useful conclusions.
Don't forget the turbine driven main battle tanks.
--
Mvh./Regards, Niels J�rgen Kruse, Vanl�se, Denmark
Rick Jones
> Stephen Sprunk <ste...@sprunk.org> wrote:
> > The closest you could get to real-world data about how turbines
> > handle sharp forces like that would be from carrier-based fighter
> > aircraft, but their maintenance regime is so different than
> > consumer automobiles that I doubt you could draw any useful
> > conclusions.
> Don't forget the turbine driven main battle tanks.
Good point - still, since those are tracked, while they do indeed
bounce around quite a bit, apart from perhaps their last moments, do
they get quite the same "jolts" as a wheeled vehicle gets from a
pothole?
That a turbocharger is in essence a turbine had escaped me but given I
have one under the hood of my car I guess it is a reasonable existence
proof of surviving potholes. Of course, having said that, I'm sure
the turbo will fail within a week :)
rick jones
--
oxymoron n, Hummer H2 with California Save Our Coasts and Oceans plates
Eugene Miya
Robert Myers <rbmye...@gmail.com> wrote:
>"For a successful technology, reality must take precedence over
>public relations, for nature cannot be fooled."
>
>from "Appendix F - Personal observations on the reliability of the
>Shuttle" by R. P. Feynman
I know. I was the first person to post it to the net when it came out.
I had to OCR my copy from Caltech. Required reading was my
Constellation comment. Section 3 was on software.
>(relegated to an appendix, of course)
That was an involved interesting story in its own right.
>And, from WIkipedia: The concept known as the law of the instrument,
>Maslow's hammer, or a golden hammer is an over-reliance on a familiar
>tool; as Abraham Maslow said in 1962, "When the only tool you have is
>a hammer, it is tempting to treat everything as if it were a nail."
Actually said before Maslow got it.
You want to trust wikipedia?
Eugene Miya
Rick Jones <rick....@hp.com> wrote:
>Niels J?rgen Kruse <nos...@ab-katrinedal.dk> wrote:
>> Stephen Sprunk <ste...@sprunk.org> wrote:
>> > The closest you could get to real-world data about how turbines
>> > handle sharp forces like that would be from carrier-based fighter
>> > aircraft, but their maintenance regime is so different than
>> > consumer automobiles that I doubt you could draw any useful
>> > conclusions.
>
>> Don't forget the turbine driven main battle tanks.
>
>Good point - still, since those are tracked, while they do indeed
>bounce around quite a bit, apart from perhaps their last moments, do
>they get quite the same "jolts" as a wheeled vehicle gets from a
>pothole?
>
>That a turbocharger is in essence a turbine had escaped me but given I
>have one under the hood of my car I guess it is a reasonable existence
>proof of surviving potholes. Of course, having said that, I'm sure
>the turbo will fail within a week :)
A horse owning co-worker used to simulate the inside of turbines on our Cray.
The horse is from a slightly different thread. Some of you guys have
obsolete concepts of turbine engines. The USS Hopper has marine turbine
engines. Those who live in Silicon Valley and commute on CA 237 pass a
gas turbine power plant just before merging into I880, to the North. A
small turbine went into a version of Bell jet packs on a guy's back and
could fly him for up to 30 minutes. Turbine engines are every where.
Some of you guys are behind the times. Reminds me of another a.f.c.
poster who is clueless about diesels.
cjt
it. Too bad they didn't go into production.
Robert Myers
>
> Actually said before Maslow got it.
>
> You want to trust wikipedia?
Not really. I wanted to see if you'd rag me about using wikipedia.
The "when the only tool you have is a hammer" saying fits my fears
about the future of numerical simulation and the current generation of
so-called supercomputers.
Robert.
Joe Pfeiffer
> In article <792c5266-83f4-40c5...@m11g2000yqh.googlegroups.com>,
> Robert Myers <rbmye...@gmail.com> wrote:
>>"For a successful technology, reality must take precedence over
>>public relations, for nature cannot be fooled."
>>
>>from "Appendix F - Personal observations on the reliability of the
>>Shuttle" by R. P. Feynman
>
> I know. I was the first person to post it to the net when it came out.
> I had to OCR my copy from Caltech. Required reading was my
> Constellation comment. Section 3 was on software.
>
>>(relegated to an appendix, of course)
>
> That was an involved interesting story in its own right.
A story that appears in one of the books that make up his autobiography
(they're at school, I'm at home, I don't remember which one).
>>And, from WIkipedia: The concept known as the law of the instrument,
>>Maslow's hammer, or a golden hammer is an over-reliance on a familiar
>>tool; as Abraham Maslow said in 1962, "When the only tool you have is
>>a hammer, it is tempting to treat everything as if it were a nail."
>
> Actually said before Maslow got it.
>
> You want to trust wikipedia?
So who did say it first? When? Comments like this are why I find
wikipedia-bashing to be really annoying.
I'm not sure what "trust" means in this context. I certainly don't
"trust" anything short of the original source for anything, and that
typically needs to be verified. But as secondary sources go, wikipedia
is awfully, awfully good.
Eugene Miya
cjt <chel...@invalid.invalid> wrote:
>Chrysler prototyped a steam turbine car decades ago -- I had a ride in
>it. Too bad they didn't go into production.
What were their problems? Any idea?
I know guys who own old Stanley Steamers. They aren't turbines but they are very interesting
cars and likely will be running when gas is too expensive for many people.
Not the 60s gas turbine?
Eugene Miya
Robert Myers <rbmye...@gmail.com> wrote:
>On Jul 8, 9:37=A0pm, eug...@cse.ucsc.edu (Eugene Miya) wrote:
>> Actually said before Maslow got it.
>>
>> You want to trust wikipedia?
>
>Not really. I wanted to see if you'd rag me about using wikipedia.
Ah trolling.
The adjacent window is a googled wikipedia page on the Chrysler gas turbine.
The collection of photos is good, some good links. It notes about 50 were made. It has a
few references.
Wikipedia isn't bad for thing nicely structured, well understood data. I've corrected and
sample started various pages (Network Systems Corp needs more and its catching on).
However, I would not trust it for descriptions for say the NSA, weapons of mass destruction,
etc. and the like for say policy inputs.
>The "when the only tool you have is a hammer" saying fits my fears
>about the future of numerical simulation and the current generation of
>so-called supercomputers.
He who has the gold calls how the gold is used. To a point (doesn't seem to work as well at
the smaller scales), the old supercomputer guys, the physicists who also think they started
the Web (they irk Berners-Lee as well as they poopooed him when he first published) also
think they (their problems) drive a subserviant computer industry. That trickle down
electronics they figure benefited general life. Fortunately, Gordon Moore commented against
that, but I don't have access to my quote DB, but it appeared in Crystal Fire.
There's also "laws" of ease of use in how people find data. People take the easy way out.
Come up with money and you can build your own supercomputers: Gilmore did and the Chudnowskis did.
Joe Pfeiffer
>>
> Chrysler prototyped a steam turbine car decades ago -- I had a ride in
> it. Too bad they didn't go into production.
Are you sure you aren't thinking of a gas turbine? That was a
research project that started in the mid 1950s, and continued until
their 1979 brush with bankruptcy. It actually came very close to going
into production in the mid 1960s (the 1963 Turbine Car was a test for
that) -- unfortunately, for all its theoretical advantages, as a
practical powerplant it suffered from low power and low gas mileage (if
it had had high power, of course, the low gas mileage wouldn't have been
a problem).
Robert Myers
>
> Wikipedia isn't bad for thing nicely structured, well understood data. I've corrected and
> sample started various pages (Network Systems Corp needs more and its catching on).
> However, I would not trust it for descriptions for say the NSA, weapons of mass destruction,
> etc. and the like for say policy inputs.
>
> Robert Myers wrote:
> >The "when the only tool you have is a hammer" saying fits my fears
> >about the future of numerical simulation and the current generation of
> >so-called supercomputers.
> Eugene Miya wrote:
> He who has the gold calls how the gold is used. To a point (doesn't seem to work as well at
> the smaller scales), the old supercomputer guys, the physicists who also think they started
> the Web (they irk Berners-Lee as well as they poopooed him when he first published) also
> think they (their problems) drive a subserviant computer industry. That trickle down
> electronics they figure benefited general life. Fortunately, Gordon Moore commented against
> that, but I don't have access to my quote DB, but it appeared in Crystal Fire.
> There's also "laws" of ease of use in how people find data. People take the easy way out.
> Come up with money and you can build your own supercomputers: Gilmore did and the Chudnowskis did.
A great deal depends on what you think supercomputers are for. I've
been outside the barbed-wire for a long time, so I have little insight
into what the DoD might be thinking.
I also don't understand much about the politics of big egos, although
I've certainly had enough exposure to the big egos of big egos. I
read about the Chudnowskis when a profile about them appeared in the
New Yorker in 1992. What Gilmore did seems fundamentally
uninteresting.
If you don't have to pay the electricity bill or worry about moving
lots of data around fast, building a "supercomputer" with parts that
can be had for almost nothing is not a particular challenge.
Alll of the "supercomputers" that get the headlines leave out the
hardest part of the problem--routing, switching, and moving arbitrary
amounts of data anywhere in the machine quickly--by simply defining it
out of existence. Apparently, that's acceptable to whoever writes
the checks, and there's not much I can do about it.
Just as it took people a long time to wake up to the fact that the
cost of computers is no longer driven by execution units, it seems to
be taking people a long time to wake to the fact that, relative to
bytes/second, flops/second are essentially free, so no one should be
impressed by lots of flops/second when the bytes/flop are vanishingly
small. It's looking as if stream processors will actually make the
situation worse, as the market is driven by people who want to do lots
of operations on the same small set of data.
Problems with good data locality are the easiest. Does that surprise
anyone? If someone claims that problems with good data locality are
necessarily the most important, put your hand on your wallet, because
you're about to be had. I can't resist pointing out that there's a
neat parallel to your criticism of wikipedia. The fact that it's
really great for some kinds of things doesn't mean that it's good or
even benign in all circumstances.
Robert.
Eugene Miya
>>>"For a successful technology, reality must take precedence over
>>>public relations, for nature cannot be fooled."
>>>
>>>from "Appendix F - Personal observations on the reliability of the
>>>Shuttle" by R. P. Feynman
>>
>> I know. I was the first person to post it to the net when it came out.
>> I had to OCR my copy from Caltech. Required reading was my
>> Constellation comment. Section 3 was on software.
>>
>>>(relegated to an appendix, of course)
>>
>> That was an involved interesting story in its own right.
>
>A story that appears in one of the books that make up his autobiography
>(they're at school, I'm at home, I don't remember which one).
A limited version appears in Chris Sykes' No Ordinary Genius book. I
think a slightly more complete might appear elsewhere where he gives
slightly more nods to the bureaucrats (possibly the last Nova).
We bussed past his office last evening (Lauritsen hall).
>>>And, from WIkipedia: The concept known as the law of the instrument,
>>>Maslow's hammer, or a golden hammer is an over-reliance on a familiar
>>>tool; as Abraham Maslow said in 1962, "When the only tool you have is
>>>a hammer, it is tempting to treat everything as if it were a nail."
>>
>> Actually said before Maslow got it.
>>
>> You want to trust wikipedia?
>
>So who did say it first? When? Comments like this are why I find
>wikipedia-bashing to be really annoying.
I think attribution to Maslow in '62 is just as annoying as wikipedia bashing.
Wikipedia has aspects which deserved to be bashed, and in the long
wrong, it has the potential to make it better than Britania or Colliers
or other commercial work. Whom first? If I had time, I might consider
the popular culture question, but like Kilroy, I think the hammer
comment goes back to at least WWII.
>I'm not sure what "trust" means in this context. I certainly don't
>"trust" anything short of the original source for anything, and that
>typically needs to be verified. But as secondary sources go, wikipedia
>is awfully, awfully good.
I would not use two "awfully"s, but it is very good for certain kinds of
information. However, one of the most recent corrections I applied was
a simple typo, and it got rejected. The person doing the rejection
wasn't in the USA. He was very apologetic on the exchange of a couple
of messages.
In the end, it's we who are the editors. Wikipedia's advantage isn't
merely editing but also the attached less formal Discussion page. Some
people, like older centralized mainframes types, tend to like single
threaded narratives (a complaint heard from one of the Unisys guys).
They have to learn to fork and divide. Let them have their editing wars.
More time for the rest of us to work on getting other more important
things right. We have to be vigilant about wikipedia.
So called B and C rated articless are likely in many cases Start-level.
I should go check a Drake's Bay references to see if academics battle
over him discovering that Bay.
Eugene Miya
Robert Myers <rbmye...@gmail.com> wrote:
>On Jul 9, 4:02=A0pm, eug...@cse.ucsc.edu (Eugene Miya) wrote:
>> Wikipedia isn't bad for thing nicely structured, well understood data.
>
>> >The "when the only tool you have is a hammer" saying fits my fears
>> >about the future of numerical simulation and the current generation of
>> >so-called supercomputers.
>
>> Eugene Miya wrote:
>em to work as well at
>think they started
>> the Web (they irk Berners-Lee as well as they poopooed him when he first =
>published) also
>> think they (their problems) drive a subserviant computer industry. =A0Tha=
>t trickle down
>> electronics they figure benefited general life. =A0Fortunately, Gordon Mo=
>ore commented against
>> that, but I don't have access to my quote DB, but it appeared in Crystal =
>Fire.
>> There's also "laws" of ease of use in how people find data. =A0People tak=
>e the easy way out.
> and the Chudnowskis did.
>
>A great deal depends on what you think supercomputers are for.
Well I know what they are used for.
>I've been outside the barbed-wire for a long time, so I have little insight
>into what the DoD might be thinking.
The DOD as a whole doesn't do much despite a so-called modernization
program. Select Agencies and Depts in the DOD go beyond mainframes.
>I also don't understand much about the politics of big egos, although
>I've certainly had enough exposure to the big egos of big egos.
Mostly about the control of money.
>I read about the Chudnowskis when a profile about them appeared in the
>New Yorker in 1992.
A cute non-mathematical article which does a little toward a balanced
view (the author interviews 1-2 critics).
> What Gilmore did seems fundamentally uninteresting.
What John did is quite interesting. Banks and other organizations were
funding work on the old 56-bit DES. When EFF went public, all that work
stopped. I muled half a dozen copies of the Cracking DES book to both
CIA and NSA, so whether or not they had them before before, they have it now.
A public copy is the National Crypto Museum library was one of the 6
copies I carried there. However, they chose not to sell copies of
Cracking DES (too bad, the EFF could use the revenue). He spent comparatively
little money to make a point on an obsolete standard.
>If you don't have to pay the electricity bill or worry about moving
>lots of data around fast, building a "supercomputer" with parts that
>can be had for almost nothing is not a particular challenge.
You need code. W/o softtware it's merely a paperweight.
>Alll of the "supercomputers" that get the headlines leave out the
>hardest part of the problem--routing, switching, and moving arbitrary
>amounts of data anywhere in the machine quickly--by simply defining it
>out of existence. Apparently, that's acceptable to whoever writes
>the checks, and there's not much I can do about it.
Stop reading only headlines.
>Just as it took people a long time to wake up to the fact that the
>cost of computers is no longer driven by execution units, it seems to
>be taking people a long time to wake to the fact that, relative to
>bytes/second, flops/second are essentially free, so no one should be
>impressed by lots of flops/second when the bytes/flop are vanishingly
>small. It's looking as if stream processors will actually make the
>situation worse, as the market is driven by people who want to do lots
>of operations on the same small set of data.
Bytes? Looking in the wrong place.
>Problems with good data locality are the easiest. Does that surprise
>anyone? If someone claims that problems with good data locality are
>necessarily the most important, put your hand on your wallet, because
>you're about to be had. I can't resist pointing out that there's a
>neat parallel to your criticism of wikipedia. The fact that it's
>really great for some kinds of things doesn't mean that it's good or
>even benign in all circumstances.
It all depends whom you are reading.
Eugene Miya
to be a DARPA or NSF program manager and deal out various millions of $$,
apply.
Go to their web site.
And show DARPA you know what you are talking about.
I have to get back to getting NSF program managers (yes, you have to
move to the DC area, Arlington).