"Linux is better than any proprietary Unix"
Greg Douglas
The emergence of Linux as an industry standard, and the
fact that Linux is better than any proprietary version
of Unix, led SGI to reassess its position in the Windows
NT market, Vrolyk said.
-- John R. Vrolyk, senior vice president of SGI's computer
systems business unit.
Still selling those Origins, John?
This company is doomed.
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Greg Douglas : Linux is only free
Reputable Systems : if your time is worthless.
http://www.reputable.com : - MrNutty.
Harry Tanovich
become NUMA systems!
EVILjosh
> The emergence of Linux as an industry standard, and the
> fact that Linux is better than any proprietary version
> of Unix, led SGI to reassess its position in the Windows
> NT market, Vrolyk said.
The optimist in me could ask what better means.
Thing is that these kinds of statements totally undermine
confidence. What does he think of IRIX?
> -- John R. Vrolyk, senior vice president of SGI's no
> tact unit.
Yours,
Josh...
_/_/_/ _/ _/ _/ _/
_/_/ _/ _/ _/ _/
_/_/_/ _/ _/ _/_/
Arthur Hagen
> > The emergence of Linux as an industry standard, and the
> > fact that Linux is better than any proprietary version
> > of Unix, led SGI to reassess its position in the Windows
> > NT market, Vrolyk said.
>
> The optimist in me could ask what better means.
> Thing is that these kinds of statements totally undermine
> confidence. What does he think of IRIX?
That it will be "integrated into linux", appearently. Which makes me
suspect that these guys might not know the difference between programs
and operating systems.
It's indeed a sad day. Upto now I have heartily recommended IRIX
systems, and purchased my share. Until, if ever, SGI can show some more
focus on core business instead of jumping on the linux bandwagon and
make a second attempt at internet products (remember WebFORCE and
Cosmo*, anyone?), my advice will be "no, we/you do NOT want to buy new
products from SGI right now".
*sigh*
Regards,
--
*Art
Bev Brandt
a...@broomstick.com wrote:
> That it will be "integrated into linux", appearently. Which makes me
> suspect that these guys might not know the difference between programs
> and operating systems.
>
> It's indeed a sad day. Upto now I have heartily recommended IRIX
> systems, and purchased my share. Until, if ever, SGI can show some
more
> focus on core business instead of jumping on the linux bandwagon and
> make a second attempt at internet products (remember WebFORCE and
> Cosmo*, anyone?), my advice will be "no, we/you do NOT want to buy new
> products from SGI right now".
>
> *sigh*
>
> Regards,
> --
> *Art
>
Yikes. The last thing I need to hear is someone who agrees with me! ;-)
We lease our equipment and our SGI leases are up at the end of the
year. We either hang on to Irix-based equipment (which is likely) or
look elsewhere. Problem is, I'm backed into an Irix corner with Xinet
(I could run it on Solaris. I could shoot myself in the foot, too.)
None of my favorite "stuff" has been ported to Linux. (Well, I could
run Helios on Linux. I could shoot myself in the other foot, too.) Or I
could run NT. (I'm out of feet, might as well put the gun to my head.)
My situation is not desperate, I'm not really looking to make a major
move. But what if I were? It's not like I have a big ol' budget and
it's not like a server or two is going to make a dent in anyone's
marketing plan. But a server or two from me, some from the guy down the
street, a few workstations from those people over there and...But all
us small fry are hanging on to our pocket books right now.
Most of the time, when a company announces a new strategy or serious
cutbacks, their stock price goes *up.* (Such are the vultures of Wall
Street. It happened with regularity with the old McDonnell Douglas. You
remember them don't you...itty bitty company that got *bought* *out* by
someone called...let me see...Boeing, oh yeah.) But because Linux is a
virtual unknown to the suits and because there has been announcement
upon announcement from SGI and because that makes the company *seem*
desperate -- well, SGI is in the crapper right now and that isn't A
Good Thing.
I'm trying to cop a "wait and see" attitude, but boy is it tough. Every
day I come down on a different side of the fence. "SGI is dying," Bad
Bev says. "No, this is going to turn them around!" Good Bev says.
I really need to see a doctor about those voices. In the meantime, I
ain't buying A THING!
--
Beverly Brandt
bev brandt at mindspring dot com
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
cos...@hepcat.org
> I really need to see a doctor about those voices. In the meantime, I
> ain't buying A THING!
yep. i will never do business with sgi again, ever. ive canceled all
support contracts and have advised everyone else to do the same and to tell
the now trimmed work force over there at sgi they can take their new linux
offerings and shove em up their ass.
at least ive got a nice array of quality, well designed/built workhorses
from a once great company known as silicon graphics.
fuck sgi, fuck linux, and fuck the slashdot masses.
peace.
Pat Wilson
>Most of the time, when a company announces a new strategy or serious
>cutbacks, their stock price goes *up.* (Such are the vultures of Wall
>Street. It happened with regularity with the old McDonnell Douglas. You
>remember them don't you...itty bitty company that got *bought* *out* by
>someone called...let me see...Boeing, oh yeah.) But because Linux is a
>virtual unknown to the suits and because there has been announcement
>upon announcement from SGI and because that makes the company *seem*
>desperate -- well, SGI is in the crapper right now and that isn't A
>Good Thing.
SGI has been flailing for several years now (remember them jumping
onto the NT bandwagon with both feet 2-3 years ago)? And they
haven't made many friends lately as they've changed kernel internals
and made the details unavailable...
We won't mention the pathetic name and logo change.
Damn shame. SGI _used_ to be a very cool company. Too bad they
decided that market share was more important than excellence (no
offense to SGI employees - SGI has consistently hired top techincal
folks).
--
Pat Wilson
paw.n...@dartmouth.edu
Kristoffer Lawson
: But that article is very exciting! It says that the O2 and Octane will
: become NUMA systems!
I think it mixes the terms NUMA (meaning the way memory is handles in
the O2) and ccNUMA/S2MP/whatever which is the Origin cluster-like design.
I doubt it means the O2 will be getting an architecture like the Origin.
--
- ---------- = = ---------//--+
| / Kristoffer Lawson | www.fishpool.com
+-> | se...@fishpool.com | - - --+
|-- Fishpool Creations Ltd - / |
+-------- = - - - = --------- /~setok/
I R A Darth Aggie
<0UBs3.117$17.2454@ultra> wrote:
+ fuck sgi, fuck linux, and fuck the slashdot masses.
Said like someone who never used linux. So, what *are* you going to
use when your SGI's die? Solaris? some flavor of BSD? or an abacus?
inquiring minds would like to know.
James
--
Consulting Minister for Consultants, DNRC
The Bill of Rights is paid in Responsibilities - Jean McGuire
To cure your perl CGI problems, please look at:
<url:http://www.perl.com/CPAN/doc/FAQs/cgi/idiots-guide.html>
David Evans
Going forward, SGI will shift its focus to the Linux operating
systems for Intel-based platforms, while maintaining its investment
in and support for MIPS-based systems into 2002, Vrolyk said.
This makes it sound like some time in 2002 support for MIPA IRIX will cease,
meaning that one gets about three years of software support for a shiny new
Origin 2000.
--
David Evans (NeXTMail/MIME OK) dfe...@bbcr.uwaterloo.ca
Computer/Synth Junkie http://bbcr.uwaterloo.ca/~dfevans/
University of Waterloo "Default is the value selected by the composer
Ontario, Canada overridden by your command." - Roland TR-707 Manual
cos...@hepcat.org
> On Thu, 12 Aug 1999 15:38:36 GMT, cos...@hepcat.org <cos...@hepcat.org>, in
> <0UBs3.117$17.2454@ultra> wrote:
> + fuck sgi, fuck linux, and fuck the slashdot masses.
> Said like someone who never used linux. So, what *are* you going to
> use when your SGI's die? Solaris? some flavor of BSD? or an abacus?
> inquiring minds would like to know.
yes, i've never used linux. in fact, i don't know what linux is!! i've only
heard that it cures cancer and promotes world peace. just remember kids,
DONT DRINK THE KOOL-AID!
sgi's die? theres a novel concept. funny thing is, ive had sgi machines
at home since 1993 and not one has died. now intel shitboxes on the other
hand...
peace.
I R A Darth Aggie
<Y_Es3.206$17.2524@ultra> wrote:
+ sgi's die? theres a novel concept.
As if. I've seen 'em die in many different ways...fortunately, those where
under a university program, so we resurrected them...
The best was the burnt powersupply in an Onyx. Smelled that one when I
walked in the building, and said "oh, electronic fire". Yuck.
Alastair
Alastair
>We lease our equipment and our SGI leases are up at the end of the
>year. We either hang on to Irix-based equipment (which is likely) or
>look elsewhere.
I'm sure we're all in a similar position - that is, 'wait and see'. If
the SGI's are working well for you, there's no reason to switch.
The original quote was a very poor choice of words - I'd say it might
be missing words like 'model' or 'grow' e.g.
the 'Linux model is better than any proprietary Unix'
or
'Linux can grow to be better than any proprietary Unix'
Whatever. It was taken out of context anyway.
There was a commitment to new MIPS CPU's until 2002 so you should be
'safe' to re-lease next year. Since transitions are likely, it is in
SGI's interest to make this easy and desirable.
I am obviously in a positive mood just now :-)
But have concerns as well, ofcourse :-(
Bye,
Alastair
> Not only that but:
>
> Going forward, SGI will shift its focus to the Linux operating
> systems for Intel-based platforms, while maintaining its investment
> in and support for MIPS-based systems into 2002, Vrolyk said.
>
>This makes it sound like some time in 2002 support for MIPA IRIX will cease,
>meaning that one gets about three years of software support for a shiny new
>Origin 2000.
I imagine that would be crazy. I would hope SGI could clarify this
point. I would expect some 'years' after this.
Another quote ;
SGI will continue to introduce new MIPS processors into 2002, as
previously planned, Vrolyk said.
You can't introduce new MIPS 'into 2002' and only maintain support
'into 2002'.
Alastair
>And they haven't made many friends lately as they've changed kernel
>internals and made the details unavailable...
Perhaps if you posted a specific question about what you want to know.
>Damn shame. SGI _used_ to be a very cool company. Too bad they
>decided that market share was more important than excellence
>(no offense to SGI employees - SGI has consistently hired top
>techincal folks).
Agreed. SGI have a lot of very good technical folk. Still.
Although they've lost some to companies like NVidia.
As for 'market share' vs 'excellence', it's a question of economics, I
guess. I'm constantly amazed my the way publicly traded company's
behave :-/
Defining what's 'cool' is the path to hell :-)
Bye,
David Evans
Alastair <alas...@calliope.demon.co.uk> wrote:
>
>I imagine that would be crazy. I would hope SGI could clarify this
>point. I would expect some 'years' after this.
>
Likewise. The article doesn't come accross as a very well thought-out PR
move, does it?
EVILjosh
> Said like someone who never used linux.
You can do better than that.
> So, what *are* you going to
> use when your SGI's die?
Of what?
EVILjosh
> yes, i've never used linux. in fact, i don't know what linux is!! i've only
> heard that it cures cancer and promotes world peace. just remember kids,
> DONT DRINK THE KOOL-AID!
Belluzo == Jones.
EVILjosh
> As if. I've seen 'em die in many different ways...fortunately, those where
> under a university program, so we resurrected them...
>
> The best was the burnt powersupply in an Onyx. Smelled that one when I
> walked in the building, and said "oh, electronic fire". Yuck.
I don't think you understand what daniel is saying.
matt...@spire-inc.com
saving so that I could get my own SGI/Maya animation workstation. Now
it seems that when my goal is finally in sight the whole picture is
getting cloudy again.
I don't even pretend to understand big business, but it certainly
cannot be GOOD business to be giving off so many negative signals and
alienating the core customers.
Something that I've noticed Silicon Graphics users and Macintosh users
had in common was that they took pride in the tools they used to work
with. And even a sort of pride at daring to buck the system and use
something that was ahead of the pack. And at least to the best of my
knowledge, Apples reversal of its financial woes included focusing on
areas where they knew they had an established customer base, such as
education and graphics, and build from its base. I've heard nothing but
rave reviews from artists using the new Power Macs. I doubt very many
of them would be willing to settle for anything less.
So why does it seem now that SGI finds itself in a similar situation,
and seems to be bent on driving away its former patrons and advocates?
With the immense talent of human resources they have at their disposal,
it just seems they could find a way to pool the resources of their
different divisions and deliver a more satisfying solution than they
seemed to have opted for.
Matt Smith
Digital Artist
I R A Darth Aggie
<37B37D30...@noorg.org> wrote:
+
+ > Said like someone who never used linux.
+
+ You can do better than that.
Why? that's about all the effort I need to expend to lampoon someone who's
mind is closed, and is in the process of throwing a temper tantrum.
Arthur Hagen
> Likewise. The article doesn't come accross as a very well thought-out PR
> move, does it?
And the stock market reacted accordingly. No surprises there.
Regards,
--
*Art
skyw...@my-deja.com
>
> > use when your SGI's die?
>
> Of what?
Well mine would die of embarrassment
if linux was ported to them. Or maybe
disgust?
>
> Yours,
>
> Josh...
> _/_/_/ _/ _/ _/ _/
> _/_/ _/ _/ _/ _/
> _/_/_/ _/ _/ _/_/
>
Beau Vrolyk
While I'm certain that Margaret Quan made every effort to get this right,
this isn't what I actually said.
I did say that Linux has emerged as an industry standard. (With growth
rates of 212% last year, over 12 million copies installed, and an install
rate of over 1 million per month that's totally justifyable.) I also said
that Linux will become better, note the future tense, due to the massive
invesment that a large number of companies are making (including SGI) along
side the Linux Community.
Most importantly, I will repeat my position that the competition is NOT
between Windows NT and Linux. At SGI we almost never encounter a customer
who has any doubt about the type of operating system they wish to use. Long
before we ever talk to them they have decided that they either wish to use
Windows NT or a form of Unix. The competition is between versions of Unix
and Unix-like operating systems, like Linux.
In this competitive space Irix (our os) finds itself competing with Solaris,
HP/UX, AIX, etc.... While Irix is, and will remain for many many years, the
best Unix version for Technical computing and large systems, it is
inappropriate for IA-32 based machines and other small system. For this
reason, we are adding a number of the best features of Irix to Linux, and
deploying in on small systems.
I would call your attention to an announcement we just made with Ohio
Supercomputer Center, attached at the end of this message, in which we've
built a 128 CPU cluster of IA-32 Four CPU machines running Linux. This is
exactly the direction we'll be taking. Some time, far in the future, Linux
will grow up to challenge Irix as a large machine OS. Between now and then
it will swamp all the small machine Unix versions as it expands.
Hope this clears this issue up.
Greg Douglas wrote in message <37B1CC7A...@reputable.com>...
>http://www.eet.com/story/OEG19990810S0019
>
>The emergence of Linux as an industry standard, and the
>fact that Linux is better than any proprietary version
>of Unix, led SGI to reassess its position in the Windows
>NT market, Vrolyk said.
>
>-- John R. Vrolyk, senior vice president of SGI's computer
>systems business unit.
>
>Still selling those Origins, John?
>This company is doomed.
>
>--
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>Greg Douglas : Linux is only free
>Reputable Systems : if your time is worthless.
>http://www.reputable.com : - MrNutty.
>SGI TO INSTALL 128-PROCESSOR LINUX CLUSTER AT OSC
> 08.13.99
>SCIENCE AND ENGINEERING NEWS
> HPCwire
>===============================================================
>============
>
> Mountain View, CA -- SGI announced that it will install the company's
>first 128-processor Linux cluster at the Ohio Supercomputer
>Center (OSC),
>bringing new technologies to the Ohio research and education
>community. As
>the adoption of Linux systems expands across all marketplaces, Ohio
>scientists, educators and engineers can begin to use the
>state's largest
>Beowulf cluster as a starting point into scalable high-performance
>computing.
>
> "Presenting new technologies, like the 128-processor Linux
>cluster, to
>Ohio is central to OSC's mission," said Charlie Bender,
>executive director,
>OSC. "This type of project provides the center with an
>opportunity to expand
>its role as a statewide resource by bringing even more
>scalable computing
>power to Ohio's scientists and engineers. This collaborative
>project with
>SGI will help us assist researchers using Linux at their
>desktop to use the
>high-performance computing systems found both at OSC and the National
>Science Foundation supercomputer centers."
>
> Beowulf clusters like OSC's are specialized supercomputers that are
>gaining popularity in the technical and enterprise computing
>market because
>of their high performance at a relatively low cost. Beowulf
>clusters are
>used for solving very specific types of problems through what
>is known as
>parallel decomposing.
>
> Ohio's research community will be able to access the Beowulf cluster
>through OARnet, a division of OSC. OARnet is the state's
>high-performance
>network providing Internet connectivity to more than a million
>Ohioans. As a
>leader in computing and networking, OSC is a state-supported
>resource for
>Ohio's scientists and engineers with an impressive array of
>machines and
>visualization equipment including Cray T3E and Cray T94
>supercomputers, two
>Origin 2000 servers and several Silicon Graphics 320 and
>Silicon Graphics
>540 visual workstations.
>
> "SGI continues to pave the way for highly scalable, parallel Linux
>solutions in both technical and commercial markets. Our
>cluster solutions
>offer the unique capability to scale both high and wide for the right
>price-performance to meet our customers' application and budget
>requirements," said Jan Silverman, vice president of
>marketing, Computer
>Systems Business Unit, SGI. "Together with OSC, we hope to focus our
>attention on building better management tools, workload
>balancing and high
>availability for these cost-effective clusters."
>
> The OSC Linux cluster will consist of 32 SGI 1400L servers,
>each with four
>500 MHz Intel Pentium III Xeon processors. Preloaded with the SGI Linux
>Environment with Rad Hat Linux 6.0, the SGI 1400L server is an
>enterprise-class server designed to fulfill customer needs for
>comprehensive
>and cost-effective solutions that merge SGI's expertise and
>innovation in
>scalability, bandwidth and performance with industry-standard
>components and
>operating systems.
>
> The Ohio Supercomputer Center is a statewide resource
>located in Columbus,
>Ohio. High-performance computing and networking converge at OSC to help
>position Ohio as a technology state. OSC works to bring the
>latest education
>and information technologies to the state through its
>involvement with the
>Ohio science and engineering community and its role in state
>and national
>advanced technology collaborations.
>
> For further information visit http://www.osc.edu or
http://www.sgi.com
Beau Vrolyk
David,
We have announced that we will be introducing new MIPS based system running
Irix through 2002. Those are the ones that we've formally committed to.
There will be many more after than, but we don't normally announce products
over three years in advance. Those systems will be supported for many years
after their introduction in 2002 and beyond, at least 15 years under the
terms of some contracts. Hopefully, this will clear this up.
Beau
John R "Beau" Vrolyk
SGI
David Evans wrote in message <7ov1f8$9la$1...@watserv3.uwaterloo.ca>...
> Not only that but:
>
> Going forward, SGI will shift its focus to the Linux operating
> systems for Intel-based platforms, while maintaining its investment
> in and support for MIPS-based systems into 2002, Vrolyk said.
>
>This makes it sound like some time in 2002 support for MIPA IRIX will
cease,
>meaning that one gets about three years of software support for a shiny new
>Origin 2000.
>
John R. Mashey
was one of the architects of MIPS, and managed the UNIX port that
turned into both RISC/os & IRIX on MIPS ... so I'm one of the godfathers
of this stuff ... and nevertheless, I am helping us get into Linux on
Intel... & I'll explain why & how].
1) I sit right next to Vrolyk, and in plenty of meetings with him,
and the EETimes quote sounds pretty weird/misquoted to me.
Thsi happens all teh time, sometimes because we don't communicate so well,
and sometimes because press doesn't transmit what is said so well.
2) Rather than trying to answer all these comments piecemeal, why don't
I just tell you what we're doing, in more detail, i.e., from
an engineering view. Some fo this confusion coems from the fact that
textual press releases don't use the roadmap charts I have sitting on my
machines...
3) We announced, in the Spring, at least 2 more MIPS CPUs beyond the ones
on the earlier roadmap. Specifically, the latter of these appears
towards the end of 2002, meaning sold throughout 2003. There is a design
for one more beyond that, but it's not yet committed. I sit in on
design reviews of these things.
4) MIPS-based machines run MIPS-IRIX, and we have a whole stack of
enhancements to IRIX that people are working on. I expect to still be
selling new MIPS-based machines in 2004/2005 - in fact, there's a major, brand-new system design coming out next year (can't say more) that uses MIPS-IRIX, and from past experience, what all this means is that we'll still be enhancing IRIX in 2003, and supporting/maintaining it for years thereafter.
There is an extremely clever [well, I helped a little with it, so I may'
be biased) architecture that will let us cost-effectively continue
MIPS-based systems for a long time, in parallel with some IA-64 machines
that otherwise use the same hardware.
5) During the last few years, we had a whole lot of customers tell us that
while they loved IRIX, sooner or later they were going to go NT;
we had even more ISVs tell us this. It turns out that not all of those
who said that actually did, and it also turns out that a lot of
people keep telling us they'd really love to have Linux systms from us ...
if certain IRIX features were there.
We also discovered there was a strong bifurcation in market between people who demand that it be NT, and people who demand that it be anything but NT...
and that some of the people who were telling us NT really meant:
lower-cost platform with lots of software, and we don't need every IRIX bell and whistle, and actually Linux is getting close.
6) Linux *is* better than IRIX in some ways: it's cheaper, and it's getting
terrific application momentum. IRIX is a whole lot better in many other ways,
and *nobody* here is under the illusion that Linux, as it stands right now,
is a direct replacement for IRIX machines, especially for people who use the
cool stuff that IRIX does.
7) We think, by appropriately working within the Linux community, that we
can take important IRIX "genes" and offer them to the community, or
make them addons, and that "over time", we can help Linux be a useful
choice for a higher percentage of people who use IRIX. When I say,
"over time", I mean: there are some things that people do with Irix systems
that could just as easily move to Linux today. There are some other
things where it will be 2005 or later before any Linux-based machine
could be a reasonable replacement.
8) I talk to customers a lot. I show them engineering roadmaps, and it's
a lot easier to do with slides than words, but it makes it 100% clear
that we are:
a) Doing MIPS & IRIX for a *lonnnng time*, and enhancing both.
b) We are firing up LInux work, and providing intellectual
property and work to help Linux, *over time* become suitable for
our customers (at lower cost, and with more applications).
c) It does our customers no good whatsoever if we sell the greatest
OS ever, but it doesn't run the applications they want. We *know*
how to do scalability, high-performance, graphics, 64/32-bit, etc,
and we have a bunch of good people working, soem in IRIX, some in
Linux, and some in IRIX, with the long-term expectation that
those new features will be able to be moved over in a few years.
9) Anyway, my roadmaps show a long period of overlap & coexistence between
IRIX & Linux. There will be IRIX machines still running a *decade* from now.
We aren't going to try to make custoerms do flash cuts or crazed transitions:
we're not nuts; we are going to try to bring some of the great things from
IRIX into the Linux environment, and hopefully getting lower costs and
long-term broader software availability.
Does that help? It doesn't cover everything, and there are still several
things to decide, but what I've described is true to the best of my
knowledge (and it happens that I do the keeping of the top-level roadmaps,
and present them to customers quite often).
[Note: this is generated on an IRIX O2; at home I have an Indy (IRIX),
and a 320 (NT); I will be getting another disk for the 320 so I can dual-boot
it with Linux, but of course, I will personally have at least one IRIX
machine until it dies [if only for old showcase files, and masses of
past history about MIPS, going back to 1985].
--
-john mashey DISCLAIMER: <generic disclaimer: I speak for me only...>
EMAIL: ma...@sgi.com DDD: 650-933-3090 FAX: 650-933-4392
USPS: SGI 40U-005,
2011 N. Shoreline Blvd, Mountain View, CA 94043-1389
Jeremy Allison
>[I'm a Chief Scientist at SGI, started doing UNIX stuff in 1973,
>was one of the architects of MIPS, and managed the UNIX port that
>turned into both RISC/os & IRIX on MIPS ... so I'm one of the godfathers
>of this stuff ... and nevertheless, I am helping us get into Linux on
>Intel... & I'll explain why & how].
...The rest of John's post deleted....
Thanks, John for talking about the IRIX plans. I just came from
LinuxWorld Expo where many people were making the same mistake
of thinking that support for Linux means no support for IRIX.
From a completely engineering perspective (ie. no I'm *NOT*
speaking for SGI, just making a personal observation)
what I see here day to day is work being done on Linux in
*addition* to the work being done on IRIX.
I'm employed by SGI to develop Samba (yes, one of those 'evil' Linux
applications :-). Samba is developed first and formost on IRIX.
We test Samba on IRIX benchmark Samba on IRIX, develop new Samba
features on IRIX. When Samba is working correctly on IRIX I then
port it to Linux (RedHat 6.0 actually) in the same way I do to
Solaris, AIX and HPUX. I then additionally benchmark Samba on one
of the new 1400L boxes (that's the new Linux server).
The main samba.org Web server is Apache on IRIX running on an O200 !
As I said, this is just a personal perspective from a developer
in SGI, but all *large* (ie. multi-terrabyte) Samba deployments
that I know about are IRIX based.
Regards,
Jeremy Allison,
(just speaking for myself this time, not SGI or the Samba Team).
David Evans
Beau Vrolyk <bvr...@sgi.com> wrote:
>
>We have announced that we will be introducing new MIPS based system running
>Irix through 2002. Those are the ones that we've formally committed to.
>There will be many more after than, but we don't normally announce products
>over three years in advance.
I just read John Mashey's response saying essentially the same thing--good
to hear. It's a shame that the article botched the interpretation as badly
as it did.
David Evans
John R. Mashey <ma...@mash.engr.sgi.com> wrote:
>4) MIPS-based machines run MIPS-IRIX, and we have a whole stack of
>enhancements to IRIX that people are working on. I expect to still be
>selling new MIPS-based machines in 2004/2005
[ etc.]
All good news to reduce the FUD that's around. That EE article was one
of the biggest FUD producers I've seen in a while!
Now all we need are those licenses for IRIX and MIPSPro on old, slow boxes
for hobbyist use. ;-)
John R. Mashey
|> I just read John Mashey's response saying essentially the same thing--good
|> to hear. It's a shame that the article botched the interpretation as badly
|> as it did.
If you deal with the press a lot, this is the way it goes...
There are good ones, and bad ones, and they're all under tight deadlines,
and dealing with very complicated, fast-changing businesses; we sometimes
don't communicate clearly enough; sometimes we invest energy in educating
reporters, and about the time they get up speed, they switch to a complete
different column area, and we have to start again. But that's life.
Arthur Hagen
> 3) We announced, in the Spring, at least 2 more MIPS CPUs beyond the ones
> on the earlier roadmap. Specifically, the latter of these appears
> towards the end of 2002, meaning sold throughout 2003. There is a design
> for one more beyond that, but it's not yet committed. I sit in on
> design reviews of these things.
Now this is good and reassuring news indeed, if the new processors will
be backed up by new IRIX development (instead of being meant for the
consumer RISC market only).
> 4) MIPS-based machines run MIPS-IRIX, and we have a whole stack of
> enhancements to IRIX that people are working on. I expect to still be
> selling new MIPS-based machines in 2004/2005 - in fact, there's a major, brand-new system design coming out next year (can't say more) that uses MIPS-IRIX, and from past experience, what all this means is that we'll still be enhancing IRIX in 2003, and supporting/maintaining it for years thereafter.
Even more good news. Thanks for clarifying. I don't think many people
here fears linux as such, but I am sure many people will have a hard
time swallowing it as a direct replacement for IRIX in any forseeable
future. I am sure linux too will continue to develop and grow,
especially after Linus Torvald's latest announcements, and I think it
may be a nice addition to the market. But it's not mature enough to
replace IRIX for quite some time yet, and there may also be legal
implications preventing some installations from using linux at all.
> c) It does our customers no good whatsoever if we sell the greatest
> OS ever, but it doesn't run the applications they want. We *know*
> how to do scalability, high-performance, graphics, 64/32-bit, etc,
> and we have a bunch of good people working, soem in IRIX, some in
> Linux, and some in IRIX, with the long-term expectation that
> those new features will be able to be moved over in a few years.
Once upon a time, SGI paid or subsidised external developers to port key
applications to IRIX, and provided extensive help and support for
vendors wishing to do so. I believe that the availability of a program
like Adobe Photoshop was essential when people bought machines like the
Indy a few years back. Other programs that was ported and then dropped
include (but are not limited to) Netscape Servers (the SGI version is
still an old o32 application), Zmail, Adobe Illustrator, Xing
Streamworks, Real Audio servers and players, (only 5.0 available for
IRIX, G2 is the standard now), Word Perfect, Lightwave (no plugins),
Insignia Softwindows (no Windows 98) and various others. I tend to
believe that if more key applications were made available for IRIX,
there would be less demand for Windows NT and/or Linux. The workstation
*users* are usually most concerned with having the applications
available (and updated), whereas the system *administrators* may be more
concerned with other factors, including network compatibility,
administrative issues, upgradability, price/performance and management
policies. The end user's wish list is important, but not always
desirable or easily implementable if it means changing operating
systems.
> 9) Anyway, my roadmaps show a long period of overlap & coexistence between
> IRIX & Linux. There will be IRIX machines still running a *decade* from now.
> We aren't going to try to make custoerms do flash cuts or crazed transitions:
> we're not nuts; we are going to try to bring some of the great things from
> IRIX into the Linux environment, and hopefully getting lower costs and
> long-term broader software availability.
How about taking pieces of the Linux environment and incorporating into
IRIX? SGI-supported key Linux and/or GNU apps running on IRIX seems to
me to be a good thing. Samba is a first step - how about supported
versions of other software usually associated with Linux, like Apache,
php, IP masquerading, GIMP and others?
But thank you very much for clarifying some of the issues. I still hope
for IRIX and MIPS to be a vital part of SGI's future, and wish you luck
with the expansion into the linux market.
Regards,
--
*Art
Dave Olson
| "John R. Mashey" wrote:
| > 3) We announced, in the Spring, at least 2 more MIPS CPUs beyond the ones
| > on the earlier roadmap. Specifically, the latter of these appears
| > towards the end of 2002, meaning sold throughout 2003. There is a design
| > for one more beyond that, but it's not yet committed. I sit in on
| > design reviews of these things.
|
| be backed up by new IRIX development (instead of being meant for the
| consumer RISC market only).
I'll take a stab at answering for John. Yes, these are the R10000
descendents, and therefore are by sgi along with our silicon
partners, and are intended primarily (but not exclusively) for our
systems.
Not related to the MIPS business plans and markets.
| Once upon a time, SGI paid or subsidised external developers to port key
| applications to IRIX, and provided extensive help and support for
| vendors wishing to do so. I believe that the availability of a program
(speaking for me, not for the company). Yes, this is probably the case,
but it doesn't come cheap, and there is always the risk that your market
becomes small enough that even assistance of various forms isn't enough
to get new ports, or new versions of old ports. That may or may not be
why the number of apps has dropped off, but I'd certainly guess it's related.
On a side note, there are new versions of netscape server, even though
they were still o32. There's a lot of history behind that...
--
Dave Olson, Silicon Graphics
http://reality.sgi.com/olson ol...@sgi.com
Carl Bergfors
is right now. Right now IRIX is probably the best one. But in the future
Linux will be the sole Unix os. It is rapidly growing in a rate that will
not be checked by any Unix or Microsoft system.
Many people here are afraid that IRIX will go away but it will eventually
and be replaced by Linux, a Linux that can do everything that IRIX can and
more. It should be clear that people see Linux as a salvation from
Microsoft. The battle is really between Linux and Microsoft in the long
run. In the short run the battle is between Linux and proprietary Unices
but that battle is really over. Linux won. The good thing is that Linux is
the only os that can defeat Microsoft and it will defeat Microsoft. Linux
is Unix with a vengeance. Without Linux the Unix world would have
succumbed to Microsoft. We should support Linux since it can not be
stopped and since its ultimate achievement will be the overthrowing of
Microsoft
Carl
In article <37B1CC7A...@reputable.com>, Greg Douglas
cos...@hepcat.org
ahahahahahahahahahahahahahaha.....
you drank the KOOL-AID man...
seriously, i think it's time you get back to the linux advocacy groups and
get back to the real business at hand--namely, engaging in mass hysteria
and self promotion with the other fucknuts.
calgon take me away.
peace.
John R. Mashey
1) IMHO, IRIX is one of the best UNIXes out there, and definitely the
best for certain classes of work, and it will be around a long time.
2) Linux has some complementary virtues, and it is much easier to
get IRIX technical goodies into the Linux universe, than it is to
get them into the NT universe, or to get the applications from either of
those places onto IRIX at any reasonable cost.
Maybe it is hard for me to be extreme since I've been doing UNIX work for
>25 years, and have seen a lot of UNIX versions come and go.
--
Arthur Hagen
> 1) IMHO, IRIX is one of the best UNIXes out there, and definitely the
> best for certain classes of work, and it will be around a long time.
>
> 2) Linux has some complementary virtues, and it is much easier to
> get IRIX technical goodies into the Linux universe, than it is to
> get them into the NT universe, or to get the applications from either of
> those places onto IRIX at any reasonable cost.
Out of curiosity, how is it easier to port an IRIX goodie to Linux than
a Linux goodie to IRIX?
Regards,
--
*Art
Arthur Hagen
>
> Is Linux better than any proprietary Unix? It does not matter really if it
> is right now. Right now IRIX is probably the best one. But in the future
> Linux will be the sole Unix os. It is rapidly growing in a rate that will
> not be checked by any Unix or Microsoft system.
> Many people here are afraid that IRIX will go away but it will eventually
> and be replaced by Linux, a Linux that can do everything that IRIX can and
> more. It should be clear that people see Linux as a salvation from
> Microsoft. The battle is really between Linux and Microsoft in the long
> run. In the short run the battle is between Linux and proprietary Unices
> but that battle is really over. Linux won. The good thing is that Linux is
> the only os that can defeat Microsoft and it will defeat Microsoft. Linux
> is Unix with a vengeance. Without Linux the Unix world would have
> succumbed to Microsoft. We should support Linux since it can not be
> stopped and since its ultimate achievement will be the overthrowing of
> Microsoft
You forget that some people don't *want* a single winner of the battle.
Some of us believe that the future is in diversity, not in having to
choose The One operating system, no matter what it is called.
This is one of the reasons why unix *has* been successful so far - the
ability to choose different operating systems with different window
managers and different programs. As long as the protocols are
compatible, the operating systems themselves can differ as much as they
want. This is freedom. Everybody running one OS isn't.
A 2000cc straight six might be a good car engine for many purposes, but
I don't want it to be the only engine available for every car on the
planet.
I sure hope to see a plentitude of different operating systems to choose
from in the future, not just one.
Regards,
--
*Art
EVILjosh
> Is Linux better than any proprietary Unix? It does not matter really if it
> is right now. Right now IRIX is probably the best one. But in the future
> Linux will be the sole Unix os.
Doubt it.
> It is rapidly growing in a rate that will not be checked by any Unix or
> Microsoft system.
> Many people here are afraid that IRIX will go away but it will eventually
> and be replaced by Linux, a Linux that can do everything that IRIX can and
> more.
Well, SGI is talking about integrating more IRIX into the Linux base
right? Still though, IRIX is IRIX.
> It should be clear that people see Linux as a salvation from
> Microsoft.
The way you present this, Linux will be the next Windows. Thanks but
I prefer a choice.
> The battle is really between Linux and Microsoft in the long
> run. In the short run the battle is between Linux and proprietary Unices
> but that battle is really over. Linux won.
Right-o. This freys as vendors start adding their functionality to the
Linux code bases.
> The good thing is that Linux is the only os that can defeat Microsoft
> and it will defeat Microsoft. Linux is Unix with a vengeance.
So is IRIX, SunOS, Solaris, DECunix, AIX, UNICOS etc etc etc.
> Without Linux the Unix world would have succumbed to Microsoft.
> We should support Linux since it can not be
> stopped and since its ultimate achievement will be the overthrowing of
> Microsoft
The way you put it, Linux is replacing what Microsoft is.
Jon Leech
>> The battle is really between Linux and Microsoft in the long
>> run. In the short run the battle is between Linux and proprietary Unices
>> but that battle is really over. Linux won.
>
>Right-o. This freys as vendors start adding their functionality to the
>Linux code bases.
GPL protects the Linux kernel to some degree; but perhaps more
significantly, nobody is dominant in the Linux market (possible
exception for Red Hat, who is not a hardware vendor and is highly
committed to open-sourcing everything on their distribution).
With nobody dominant, vendors who "innovate" in proprietary and
incompatible ways are likely to be ignored by commercial ISVs who want
maximum market share - and are certain to be ignored by open source
developers. This provides strong pressure to at least get your
interfaces accepted into the mainstream, if not your implementation.
>> The good thing is that Linux is the only os that can defeat Microsoft
>> and it will defeat Microsoft. Linux is Unix with a vengeance.
>
>So is IRIX, SunOS, Solaris, DECunix, AIX, UNICOS etc etc etc.
But unlike etc etc., Linux is not limited to a single hardware
platform and a single OS vendor.
>> Without Linux the Unix world would have succumbed to Microsoft.
>> We should support Linux since it can not be
>> stopped and since its ultimate achievement will be the overthrowing of
>> Microsoft
>
>The way you put it, Linux is replacing what Microsoft is.
As a widely used OS, sure. As a way to dominate the PC market with
inferior design via threats, strongarming, and buyouts, unlikely.
Jon (*not* speaking for my employer)
John R. Mashey
|> "John R. Mashey" wrote:
|>
|> > 1) IMHO, IRIX is one of the best UNIXes out there, and definitely the
|> > best for certain classes of work, and it will be around a long time.
|> >
|> > 2) Linux has some complementary virtues, and it is much easier to
|> > get IRIX technical goodies into the Linux universe, than it is to
|> > get them into the NT universe, or to get the applications from either of
|> > those places onto IRIX at any reasonable cost.
|>
|> Out of curiosity, how is it easier to port an IRIX goodie to Linux than
|> a Linux goodie to IRIX?
Your words were symmetric, mine weren't.
1) We can take any source code that is SGI IP and offer it to Linux-land,
or do the equivalent work on Linux, or (in some cases) compile our
source and offer it is a binary addon.
2) There are *applications* on IRIX unavailable on NT or Linux.
3) There are many more *applications* that are available on NT,
that are not runnable on IRIX. There are probably less of such on
Linux at this instant, but there are likely to be a lot more very fast,
if I can believe the NDAs I've gotten from software vendors, and if I
believe what I see when I got into Fry's and look on the shelves.
Most of the applications of concern arte those that are only given out
in binary from.
--
-john mashey EMAIL: ma...@sgi.com DDD: 650-933-3090 FAX: 650-933-4392
USPS: SGI 1600 Amphitheatre Pkwy, ms 005, Mountain View, CA 94043-1351
John R. Mashey
|> With nobody dominant, vendors who "innovate" in proprietary and
|> incompatible ways are likely to be ignored by commercial ISVs who want
|> maximum market share - and are certain to be ignored by open source
|> developers. This provides strong pressure to at least get your
|> interfaces accepted into the mainstream, if not your implementation.
a) Once upon a time, long ago and far away (in NJ), we all had
copies of the source, and we all diddled it, and it got
a bit chaotic. [At one point in 1974/75, a department of 30 of us
had 15-20 *slightly* different versions of the shell, for starters.
This didn't last of course.)
b) UNIX has *always* acted under Darwinian evolution, that is:
a) There is some kind of UNXI that gets popular.
b) It mutates into all sorts of thigns, and then it gets too
crazy.
c) And then, there is selection & consolidation, back to a).
However, unlike the earlier waves of this, there is one thing
strikingly different about Linux (on Intel, anyway): there is
already a pretty strong binary standard that got there without
quite so many of the convolutions that the rest of us did,
and there is building a strong base of binary applications that
platforms are expected to run. This didn't happen:
- during the RISC wars [since there were different architectures]
- during the early X86 & 68K days [I'm not sure, but didn't]
- during the minicomputer days [everybody had the source, and in any
case, binary ISV software wasn't as strong a factor then]
I am not so naive as to think there won't be issues anyway, but I observe
that there is a *much* stronger incentive for multiple vendors to
follow the binary interfaces that are there, at least.
Tore H. Larsen
>
> Once upon a time, SGI paid or subsidised external developers to port key
> applications to IRIX, and provided extensive help and support for
> vendors wishing to do so. I believe that the availability of a program
> Indy a few years back. Other programs that was ported and then dropped
> include (but are not limited to) Netscape Servers (the SGI version is
> still an old o32 application), Zmail, Adobe Illustrator, Xing
> Streamworks, Real Audio servers and players, (only 5.0 available for
> IRIX, G2 is the standard now), Word Perfect, Lightwave (no plugins),
> Insignia Softwindows (no Windows 98) and various others. I tend to
> believe that if more key applications were made available for IRIX,
> there would be less demand for Windows NT and/or Linux. The workstation
Arthur,
I tend to agree with you. Hopefully lots of good things comes out of
KDE/qt which would lead to good application also for Irix. Personally I
would like to se StarOffice and WordPerfect 8.0 for Irix. My linux box
has them, and I'd like to use them at work.
> How about taking pieces of the Linux environment and incorporating into
> IRIX? SGI-supported key Linux and/or GNU apps running on IRIX seems to
> me to be a good thing. Samba is a first step - how about supported
> versions of other software usually associated with Linux, like Apache,
> php, IP masquerading, GIMP and others?
In the coming Mediabase 4.0 both streaming QuickTime 4.0 and
RealNetworks G2 servers as options.
Gimp and Apache can be found on freeware.sgi.com, or latest and gratest
Freeware CD's. KDE 1.1.1 works great on the workstations we have tested
(Indigo2, O2, Indy). No reason for it not running on any 6.5 platform.
Whether we will support it or not I know nothing about. It's opensource.
Most of the features of Linux, including IP masquerading can be done
using Gauntlet and NS Proxy afaik.
http://www-europe.sgi.com/solutions/internet/products/gauntlet/
Regards,
Tore
--
Tore H. Larsen | Switch: +47 55 54 38 30
SGI AS | Bus.fax: +47 55 54 38 33
Bergen - Norway | Phone: +47 55 54 38 39
Systems Support Engineer | Cellul: +47 918 33 670
SGI Global Services Organization | E-mail: to...@sgi.com
Martin Knoblauch
EVILjosh wrote in message <37B8B04B...@noorg.org>...
>
>
>> The battle is really between Linux and Microsoft in the long
>> run. In the short run the battle is between Linux and proprietary Unices
>> but that battle is really over. Linux won.
>
>Right-o. This freys as vendors start adding their functionality to the
>Linux code bases.
>
I do not even think there is a "battle" between Linux and the "commercial"
Unixes.
They have enough fun battling between themselves instead of coming up with
an "industrie standard" Unix [of course, we all claim to be that :-)]. Some
of the vendors are just now realizing that Linux will be that "industry
standard" and that it benefits them to participate in it. Good move, if you
care about my opinion :-)
>> The good thing is that Linux is the only os that can defeat Microsoft
>> and it will defeat Microsoft. Linux is Unix with a vengeance.
>
>So is IRIX, SunOS, Solaris, DECunix, AIX, UNICOS etc etc etc.
>
the problem with above Unixes is that they are all kind of niche players
competing against each other for parts of a small cake instead of competing
for the big volume of dough now occupied by NT. Linux does compete and grow
in that space. The others will just squeeze each other to death.
The other problem with the proprietary Unixes is that they are (today) all
bound to single HW platforms (maybe with the exceptrion of Solaris). Even if
some are planning to migrate/port to IA64, I *personally* doubt that they
have a big chance to get more of the cake than they have today.
Linux, with all its current deficiencies (file systems, SMP scalability,
...) is the one candidate to prevent the total world domination of NT
(again, my *personal* opinion). We just have to make sure that none of the
distributors gets to greedy and tries to rule the world alone.
Martin
--
------------------------------------------------------------
Martin Knoblauch
Compaq Computer EMEA BV
E-Mail: Martin.K...@Compaq.com
Std.Disclaimer: Not speaking for COMPAQ in any form on this medium
EVILjosh
> >> The good thing is that Linux is the only os that can defeat Microsoft
> >> and it will defeat Microsoft. Linux is Unix with a vengeance.
> >
> >So is IRIX, SunOS, Solaris, DECunix, AIX, UNICOS etc etc etc.
>
> platform and a single OS vendor.
Mainly referring to "Linux is Unix with a vengeance" here.
Just remember, we MUST/SHOULD all support Linux now. ;)
skyw...@my-deja.com
claim to be speaking for SGI now you are not really disclaiming it? :)
> -john mashey EMAIL: ma...@sgi.com DDD: 650-933-3090 FAX: 650-933-4392
> USPS: SGI 1600 Amphitheatre Pkwy, ms 005, Mountain View, CA
94043-1351
>
Harry Tanovich
>
> In the coming Mediabase 4.0 both streaming QuickTime 4.0 and
> RealNetworks G2 servers as options.
Mediabase? Whose product is that? Aren't we talking about SGI products here?
Kristoffer Lawson
: You forget that some people don't *want* a single winner of the battle.
: Some of us believe that the future is in diversity, not in having to
: choose The One operating system, no matter what it is called.
: This is one of the reasons why unix *has* been successful so far - the
I would say it is the opposite. The fact that there are many different
environments, many different competing vendors, many different APIs
(there is no real standard audio API f.ex.) is exactly the reason why
UNIX is failing to compete with Windows. Simply, a large, unified
environment is much easier to sell than a little bit of this and a little
bit of that. You may not like that but that's how things are.
OTOH I really see no problem with SGI going Linux if they can integrate
all of the features of IRIX into it. What then is the problem? I admit that
currently Linux is lagging far behind IRIX but apparently SGI wants to do
something about that. The only real worry I still have is that then SGI
would need to release the source to many of its core technologies, including
ccNUMA. SGI might sell hardware, but the software is also a big part of
that. If competitors can get their grubby hands on the software there is
no telling how they might use it to their advantage.
--
- ---------- = = ---------//--+
| / Kristoffer Lawson | www.fishpool.com
+-> | se...@fishpool.com | - - --+
|-- Fishpool Creations Ltd - / |
+-------- = - - - = --------- /~setok/
Scott Elyard
<martin.k...@compaq.com> wrote:
> I do not even think there is a "battle" between Linux and the "commercial"
> Unixes.
Oh, but there is. On the low-end, Linux can unquestionably be used to
replace Sun's sendmail and http servers. Linux is, finally, for good or
ill, being pushed into the SMP arena as well, where it will be competing
with every other SMP unix out there.
> They have enough fun battling between themselves instead of coming up with
> an "industrie standard" Unix [of course, we all claim to be that :-)]. Some
> of the vendors are just now realizing that Linux will be that "industry
> standard" and that it benefits them to participate in it. Good move, if you
> care about my opinion :-)
SGI has apparently embraced the new standard (or would-be standard, since
there is no official standards body that even recognises Linux as a Unix)
Others, like Sun, have done little more than play lip-service to it,
possibly because Linux more directly competes with the likes of Sun than
with the likes of SGI.
> >> The good thing is that Linux is the only os that can defeat Microsoft
> >> and it will defeat Microsoft. Linux is Unix with a vengeance.
> >
> >So is IRIX, SunOS, Solaris, DECunix, AIX, UNICOS etc etc etc.
> >
>
> the problem with above Unixes is that they are all kind of niche players
> competing against each other for parts of a small cake instead of competing
> for the big volume of dough now occupied by NT. Linux does compete and grow
> in that space. The others will just squeeze each other to death.
In a way, you're correct; each are niche players.
Suns run networks, databases and other boring business work.
IBMs run databases and other boring banking work.
HPs run engineering and manufacturing applications.
SGIs run engineering/visulaisation/supercomputing applications as well as
the boring Lucas-Hollywood stuff.
And yet they fiercely compete against one another. In a way, this is
good. Ideally this keeps prices moderately sane. But it is bad, in that
it also forced each into their respective corners, and limits direct
competition with one another, even though they make a show of it.
> The other problem with the proprietary Unixes is that they are (today) all
> bound to single HW platforms (maybe with the exceptrion of Solaris). Even if
> some are planning to migrate/port to IA64, I *personally* doubt that they
> have a big chance to get more of the cake than they have today.
Maybe. I dunno.
> Linux, with all its current deficiencies (file systems, SMP scalability,
> ...) is the one candidate to prevent the total world domination of NT
> (again, my *personal* opinion). We just have to make sure that none of the
> distributors gets to greedy and tries to rule the world alone.
I don't see that as happening, personally.
--
Scott Elyard ~~~ooOOoo~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~|
| Am I wrong? Big Deal. The discussion is all that matters.|
| There is no noise like human talk. IRIX, BeOS, & MacOS.|
^~~~~~~~~~~~~~~~~~sc...@nyetspam.stonebug.net~~~~~~~~~~~~~~'
I R A Darth Aggie
<37B94794...@noorg.org> wrote:
+ Mainly referring to "Linux is Unix with a vengeance" here.
+ Just remember, we MUST/SHOULD all support Linux now. ;)
Well, you *could* support one of the *BSD's...
James
--
Consulting Minister for Consultants, DNRC
The Bill of Rights is paid in Responsibilities - Jean McGuire
To cure your perl CGI problems, please look at:
<url:http://www.perl.com/CPAN/doc/FAQs/cgi/idiots-guide.html>
EVILjosh
> I would say it is the opposite. The fact that there are many different
> environments, many different competing vendors, many different APIs
> (there is no real standard audio API f.ex.) is exactly the reason why
> UNIX is failing to compete with Windows. Simply, a large, unified
> environment is much easier to sell than a little bit of this and a little
> bit of that. You may not like that but that's how things are.
But that is *BORING*.
EVILjosh
> Well, you *could* support one of the *BSD's...
With pleasure. I pick OpenBSD.
I used to run NetBSD too on an 660AV which I liked alot.
Scott Elyard
> > Well, you *could* support one of the *BSD's...
>
> With pleasure. I pick OpenBSD.
> I used to run NetBSD too on an 660AV which I liked alot.
A Quadra? Interesting. I didn't know there was a UNIX flavor available
for it. Did X run on it?
Penio Penev
| I tend to agree with you. Hopefully lots of good things comes out of
| KDE/qt which would lead to good application also for Irix. Personally I
| would like to se StarOffice and WordPerfect 8.0 for Irix. My linux box
| has them, and I'd like to use them at work.
I fail to see why a small department would not install a Linux server
so that people with IRIX desktops can do 'rsh linux wp -display irix:0 &'
Ditto for WABI/Linux from Caldera -- runs win16 apps rock solid,
including winword2 and winword6.
--
Penio Penev <Pe...@pisa.Rockefeller.edu> 1-212-327-7423
mapl...@gamers.org
Btw, wrt to the subject line - such a statement is utter
balderdash. You know it, I know, we all know, so I wish
a few senior execs would just come out and SAY it. It's about
time the myth of Linux being the all-encompasing UNIX god OS
was quashed. It's being hyped far too much.
ma...@mash.engr.sgi.com (John R. Mashey) wrote:
> 3) There are many more *applications* that are available on NT,
> that are not runnable on IRIX. There are probably less of such on
> Linux at this instant, but there are likely to be a lot more very
fast,
> if I can believe the NDAs I've gotten from software vendors, and if I
> believe what I see when I got into Fry's and look on the shelves.
> Most of the applications of concern arte those that are only given out
> in binary from.
What people are afraid of is IRIX being dropped in favour of Linux.
I do not want that and I believe many others do not want that either.
For the SGI UNIX machines, I want IRIX, not Linux - the latter
is not secure enough or scalable enough IMO.
Has SGI ditched all its IRIX plans? I seriously hope not, or it
means much of the PR which has been released over the past 2
years has been false.
Commitments were made to merging IRIX with Unicos, which I thought
was a fantastic idea.
Commitments were made to scale IRIX to 9000+ CPUs (or have you
all forgotten the original Cray and ASCI press releases?).
Has this been ditched?
I like the GUI IRIX has. I do not like the messyness of Linux.
Btw, why was the volumizer guy laid off recently?
And has the next-gen IR stuff (Bali?) really been scrapped?
It seems to me and many others that the previous PR output which
stated SGI was going to keep its leading gfx tech was false.
Where is Octane's new gfx? It's way behind schedule now IMO.
If/When I ever get an Octane or Onyx2 of my own, I want IRIX
on it with the latest MIPS CPU(s), not Linux running on Intel.
In fact, if SGI *really* wanted the give customers the best
possible performance, it should actually be supporting the Russian
E2K CPU which is MUCH better than IA64 - all the folks over there
need to get it going is $40M. Team up with IBM, get E2K online
and stomp on IA64. Why not? Russian CPUs have been better than
western designs for years.
If SGI really wants to help its *own* products, instead
of switching to everybody else's, why not do the following
two steps which should have been done *years* ago:
1. Make IRIX free, for ALL versions, not just old ones.
2. Make ALL Alias/Wavefront products free to students and academia.
3. Stop asking stupid prices for old products. Currently, Indys,
Indigo2s, etc. are at least 2X more than typical 2nd-hand
prices asked by ordinary 2nd-hand dealers like ITI, MCE,
Reputable, Mashek, XSInt, WUI, etc. and that's *after* the
recent so-called 90% price cuts in Indigo2 stuff.
4. Scrap the license code requirements for products like CosmoWorlds,
CosmoCode, and especially the Impressario Postcript Renderer. Is
SGI *really* making that much money out of these petty license
requirements??
Why is it IRIX 6.5.4 costs $300 in the US (a quote received
by a friend of mine), but more like $940 where I am in the UK?
(a quote I received from SGI UK) Such differences are ridiculous.
All this hoohaa about Linux being the better choice, when all this
time there are many things SGI could have done to help its own
products along. I was told a long time ago by Steve Proffitt that
many of SGI's dumb pricing and configuration practices would be
stopped, yet they still continue today. *Why* for example does
a 4MB TRAM upgrade for Octane/SE cost $4500?? That's insane!
Why are systems still being sold with disks as small as 4GB
and RAM as small as 128MB? Why are O2s still so expensive compared
to the power offered by the VW320 and VW540? And where on earth
is R7000 for O2? (R5200 just doesn't cut it).
Oh John, we want to believe, we really do, but SGI has said some
astoundingly shocking things in recent days when it comes to
reassuring customers old and new.
For over a decade, SGI has maintained order-of-magnitude improvements
in its high-end gfx technology. That era appears to be at an end
for no readily apparent logical reason. Is SGI not even going to
bother fighting against Wildcat4500 and PixelFusion? The more I
talk to people, the more it seems many SGI gfx experts have left for
other companies.
SGI has made commitments and statements in past press releases
about the future of IRIX, MIPS, Cray, etc. IMO it should be legally
bound to stick to those commitments.
Ian.
SGI Network Admin, University of Central Lancashire, Preston, England,
PR1 2HE.
mape...@yahoo.com | Tel: (+44 -0) 1772 893297, Fax: (+44 -0) 1772
892913
"There is no magic, only stuff." - Nakor, "The King's Buccaneer" (R.E.
Feist)
Doom Help Service (DHS): http://doomgate.gamers.org/dhs/
SGI/Future Technology/N64: http://www.futuretech.vuurwerk.nl/
BSc Dissertation (Doom): http://doomgate.gamers.org/dhs/diss/
Andrew Walton
>
> If SGI really wants to help its *own* products, instead
> of switching to everybody else's, why not do the following
> two steps which should have been done *years* ago:
>
> 1. Make IRIX free, for ALL versions, not just old ones.
>
> 2. Make ALL Alias/Wavefront products free to students and academia.
>
> 3. Stop asking stupid prices for old products. Currently, Indys,
> Indigo2s, etc. are at least 2X more than typical 2nd-hand
> prices asked by ordinary 2nd-hand dealers like ITI, MCE,
> Reputable, Mashek, XSInt, WUI, etc. and that's *after* the
> recent so-called 90% price cuts in Indigo2 stuff.
There is a difference between second-hand and remanufactured (warranty
and product liability - I assume) - perhaps not enough to justify the
price difference.
>
> 4. Scrap the license code requirements for products like CosmoWorlds,
> CosmoCode, and especially the Impressario Postcript Renderer. Is
> SGI *really* making that much money out of these petty license
> requirements??
The vast majority of the license fee for Impressario goes to Adobe, the
PostScript rendering engine is essentially their product.
>
> Why is it IRIX 6.5.4 costs $300 in the US (a quote received
> by a friend of mine), but more like $940 where I am in the UK?
> (a quote I received from SGI UK) Such differences are ridiculous.
Your almost certainly not comparing the same thing, the cost of Irix
depends upon the size and type of system, in the UK the list price is
260 pounds for a workstation (O2, Octane and deskside Onyx), for an SGI
2100 the price is 870 pounds and still more for an Origin2000.
--
Andrew Walton
Applications Consulting Engineer
SGI (EMEA)
Tel +44 118 925 7676
Fax +44 118 925 7505
sasha e.
>Is Linux better than any proprietary Unix? It does not matter really if it
>is right now. Right now IRIX is probably the best one. But in the future
>not be checked by any Unix or Microsoft system.
I find it exceedingly hard to believe that Linux will _ever_ completely
eradicate every other version of UNIX out there.. it's been the dream of
the UNIX community to have one version of UNIX out there that'll be
everything to everyone, but for now and for the forseeable future, that's
just not the case.
I'm fond of Linux, and use it at work as my desktop, where it's quite
sufficient. However, all of the machines I work on are Solaris, with a
few exceptions (IRIX boxes that haven't been sent to Sun, FreeBSD ftp
sites, etc.) The Linux development folks are working very hard to make
Linux a viable enterprise-class server OS by improving SMP performance,
NFS, journalling filesystems, etc., with a lot of these improvements
coming from SGI (yay sgi!) Don't expect me to toss my Ultra Enterprise
3500s anytime soon though :)
>Many people here are afraid that IRIX will go away but it will eventually
>and be replaced by Linux, a Linux that can do everything that IRIX can and
>more. It should be clear that people see Linux as a salvation from
>Microsoft. The battle is really between Linux and Microsoft in the long
>run. In the short run the battle is between Linux and proprietary Unices
>but that battle is really over. Linux won. The good thing is that Linux is
As I said before, I don't think Linux has "won" anything at this point..
sure, they have a ton of momentum, and they've taken over as the UNIX
desktop of choice, but in the server market, Linux still has a long, long
way to go before they'll top any of the big commercial vendors, or the
various BSDs (see ftp.cdrom.com).
>the only os that can defeat Microsoft and it will defeat Microsoft. Linux
>is Unix with a vengeance. Without Linux the Unix world would have
>succumbed to Microsoft. We should support Linux since it can not be
>stopped and since its ultimate achievement will be the overthrowing of
>Microsoft
...and then we shall have a dictatorship of the proletariat! Viva Linux!
Viva Che! C'mon.. the market decided that Windows was the way to go for
the desktop market, pushing aside better options linux NeXTStep (sniff..),
Apple, and others. They're not evil, they're just really successful.
Let's see how all of these hardcore Linux bigots feel once MS-Linux is
available at your local CompUSA...
cynical as always,
-s
Allan Schaffer
>
>Has SGI ditched all its IRIX plans? I seriously hope not, [...]
>And has the next-gen IR stuff really been scrapped?
No. Totally, entirely, no. And you'll be happy to know, we're
fighting back to squash all this FUD.
If you want a better understanding of SGI's Business strategy, our
plans for the next-generation midrange and High-End Graphics, our OS
plan, the NVIDIA partnership, etc.. here is the REAL DEAL:
http://reality.sgi.com/performer/perf-99-08/siggraph99-summary.html
This is a posting I made earlier today to our Performer mailing list
(summarizing our "Friends of Performer" user group meeting at
SIGGRAPH last week), so it is very Performer-centric, and it is quite
long, however it includes a good amount of what one needs to know to
understand SGI's plan for Graphics and our vision for success.
Enjoy. And I'd welcome comments, send them to al...@sgi.com
Allan
--
Allan Schaffer al...@sgi.com
Silicon Graphics http://reality.sgi.com/allan
Allan Schaffer
>
>Has SGI ditched all its IRIX plans? I seriously hope not, [...]
>And has the next-gen IR stuff really been scrapped?
No. Totally, entirely, no. And you'll be happy to know, we're
fighting back to squash all this FUD.
To get a better understanding of SGI's Business strategy, our plans
for the next-generation midrange and High-End Graphics, our OS plan,
the NVIDIA partnership, etc.. here is some highly-recommended
reading:
Penio Penev
| Has SGI ditched all its IRIX plans? I seriously hope not, or it
| means much of the PR which has been released over the past 2
| years has been false.
| Commitments were made to merging IRIX with Unicos, which I thought
| was a fantastic idea.
| Commitments were made to scale IRIX to 9000+ CPUs (or have you
| all forgotten the original Cray and ASCI press releases?).
| Has this been ditched?
Good questions. But where are the answers?
| I like the GUI IRIX has. I do not like the messyness of Linux.
Very true. But I'd say that inst on LInux is of a higher priority.
| Where is Octane's new gfx? It's way behind schedule now IMO.
Yeah, wasn't this supposed to be announced at SIGGRAPH?
| In fact, if SGI *really* wanted the give customers the best
| possible performance, it should actually be supporting the Russian
| E2K CPU which is MUCH better than IA64 - all the folks over there
| need to get it going is $40M.
With all the intellectual property? I highly doubt it. BTW, is the
E2K sampling? What feature size do they use? Do they use copper?
SOI? Which fabs do they produce it in?
| Team up with IBM, get E2K online
| and stomp on IA64. Why not?
H1 and H2 were better that IA64 too, remember? Why were _they_
dropped?
| Russian CPUs have been better than
| western designs for years.
Why, then, did SGI pay a fine recently for selling computers to the
Russians (and why were they buying it, if they have better ones)? And
why, then, was /Serge moaing and groaning recently that when he was in
Russia, he could not use the latest-and-greatest machines to do
computational chemistry because of the embargo -- didn't he have
access to those better CPUs?
| Why is it IRIX 6.5.4 costs $300 in the US (a quote received
| by a friend of mine), but more like $940 where I am in the UK?
| (a quote I received from SGI UK) Such differences are ridiculous.
This questions you need to address to Her Majesty's Tax Office. SGI
has repeatedly stated that they make _exactly_ the same profit on
every piece of service and/or equipment, regardless of geographic
region.
| Why are systems still being sold with disks as small as 4GB
| and RAM as small as 128MB? Why are O2s still so expensive compared
| to the power offered by the VW320 and VW540?
Because they run IRIX, of course -- you get more value, you pay more
money.
| And where on earth
| is R7000 for O2?
Isn't that a question for QED?
| For over a decade, SGI has maintained order-of-magnitude improvements
| in its high-end gfx technology. That era appears to be at an end
| for no readily apparent logical reason. Is SGI not even going to
| bother fighting against Wildcat4500 and PixelFusion?
But that's the whole story with NVIDIA, right?
mapl...@gamers.org
Penio Penev <pe...@venezia.rockefeller.edu> wrote:
> Good questions. But where are the answers?
"Dear John..."
:D
> Yeah, wasn't this supposed to be announced at SIGGRAPH?
It's depressing to see Octane's basic tris/sec, etc. so far
behind now. I can still remember when IMPACT was an order of
magnitude faster than most other competitors when it came out.
One can't keep beating off the PC competition forever by boasting
about 48bit RGBA and 4:4:4:4. Wildcat4500 will change all that.
> With all the intellectual property? I highly doubt it. BTW, is the
> E2K sampling? What feature size do they use? Do they use copper?
> SOI? Which fabs do they produce it in?
The Russian team wants to do a full joint venture with a western
firm which has the money. See the press release on the CIC
at Berkeley.
> H1 and H2 were better that IA64 too, remember? Why were _they_
> dropped?
They were not *significantly* better. Besides, some of the people
designing IA64 came from MIPS anyway. Lots of staff poaching in
winter of 1996.
20% better isn't enough to justify the much higher cost of
non-mass-produced CPUs. In an ideal world, Intel's mass production
muscle would be fabbing H2s, but that's life. The decisions were
based on expected 2nd-gen IA64 performance. If Intel's schedule
slips, then SGI has R18K to plug the gap, but it won't be what
H2 would have been.
I was pretty visceral about all that stuff at the time. My thanks
to John Mashey for clearing it up.
> Why, then, did SGI pay a fine recently for selling computers to the
> Russians (and why were they buying it, if they have better ones)? ...
If it's what I think you're referring to, it had to do with sales of
supercomputers and nothing to do with CPU designs.
> ... And
> why, then, was /Serge moaing and groaning recently that when he was in
> Russia, he could not use the latest-and-greatest machines to do
> computational chemistry because of the embargo -- didn't he have
> access to those better CPUs?
No, he doesn't have access to those better CPUs, because the
Russian team don't have a fab partner. Read the PR.
E2K is between 3 and 5 times faster than IA64 and runs at higher
clock speeds anyway. Last I heard, they were talking to Sun
Microsystems about a partnership.
There's actually a number of technologies in Russia which have
been far superior to western tech for many years. My favourite
is the titanium laser welding technique used on Russian nuclear
subs - the number of required passes of weld head is waaaay lower
than the method used by the US and other nations, for reasons
kept very secret I imagine. :D
E2K runs all x86 code anyway btw.
> This questions you need to address to Her Majesty's Tax Office. SGI
> has repeatedly stated that they make _exactly_ the same profit on
> every piece of service and/or equipment, regardless of geographic
> region.
I don't believe this I'm afraid. Even without the UK VAT tax
of 17.5% the price is still 500 UKP, ie. $800. Something stinks.
IRIX should be sold for the fab/stamp cost of the CD media, manuals,
box and time, etc. The sw itself should be free IMO.
> Because they run IRIX, of course -- you get more value, you pay more
> money.
That's nonsense. Are you suggesting there's a $6000 markup just
because IRIX is on the machine?
O2 is too expensive. The hikes for R12K are far too high. R7K should
be used instead in O2 - it'd be *much* faster than R12K at the same
clock speed (especially for fp) and much, much cheaper.
> Isn't that a question for QED?
Actually probably not. They designed the chip, but I doubt they're
at the top of the pack when it comes to SGI deciding whether or
not to use it and who to fab it if it were (eg. NEC).
> But that's the whole story with NVIDIA, right?
NO.
NVIDIA are doing the next mid-range stuff. Read the performer
news bulletin by Allan Schaffer.
Btw, to Allan: your bulletin was very interesting, but I have to
say, why is it SO different from the official PR? Especially
in terms of clarity. SGI should get a new PR team.
Allan Schaffer
>
>Btw, to Allan: your bulletin was very interesting, but I have to
>say, why is it SO different from the official PR? Especially
>in terms of clarity. SGI should get a new PR team.
I'll take that as a compliment. :-)
(IMHO) Official PR is done in such a way to get the information
across quickly, with an eye towards being easily quotable by the
mainstream press. This is just the way it's done -- announce the
news, quote the exec, add in a sound bite, a few more details, and
close with a reference to forward looking statements and SEC Form
10-F.
In forums like these, the rest of us have the advantage of being able
to discuss the changes (and the thinking behind them) at length, and
without having to be so formal about it all. So you can't fault the
PR people *too* much, they're "serving different markets" as one
might say.
Personally, I think it's pretty darn cool that our Senior VP (Beau
Vrolyk) is hanging out here in comp.sys.sgi.misc occasionally to
help spread the word too..
Serguei Patchkovskii
: E2K is between 3 and 5 times faster than IA64 and runs at higher
: clock speeds anyway. Last I heard, they were talking to Sun
: Microsystems about a partnership.
[...]
: E2K runs all x86 code anyway btw.
I think we should get this straight: E2k does not run *anything*
at any speed, for the good reason that there are no E2k's around.
(and no, simulations do not count - you can simulate pretty much
anything; the set of things you can actually build is substantially
smaller; the set of things which do get built is smaller still).
E2k is vapour - may be very a very hot vapour, but vapour
nonevertheless.
/Serge.P
--
home page: http://www.cobalt.chem.ucalgary.ca/ps/
Tor Arntsen
Arthur Hagen <a...@broomstick.com> writes:
>It's indeed a sad day. Upto now I have heartily recommended IRIX
[...]
It's just so completely ridiculous to see all this whining and stupid
statements coming up whenever someone mentions SGI and Linux in the
same sentence. To get so hung up in these more or less (usually less)
interesting press statements is just super childish. SGI has stated
again and again that the MIPS line will continue for *quite* a while,
and that means IRIX too. And as we all know Linux can't, which means
won't, replace IRIX for big SMP systems for a *long* time to come.
So can we just cut the crap, please?
-Tor
Scott Elyard
<pe...@venezia.rockefeller.edu> wrote:
> | Russian CPUs have been better than
> | western designs for years.
>
> Why, then, did SGI pay a fine recently for selling computers to the
> Russians (and why were they buying it, if they have better ones)?
Just because you have a great CPU design, doesn't mean you've got the fab
facilities to realise it.
> And
> why, then, was /Serge moaing and groaning recently that when he was in
> Russia, he could not use the latest-and-greatest machines to do
> computational chemistry because of the embargo -- didn't he have
> access to those better CPUs?
Not if you can't produce them. And then there's the software itself. I
imagine that's not strictly trivial to develop from scratch.
Russian (and Soviet) high-end engineering has always been impressive,
especially when it comes down to getting the most out of limited
resources. It has always made US examples, by comparison, seem downright
wasteful and inefficient.
SGI teaming up with the Russians. Now that does bring a Grinchy grin to
my face.
Arthur Hagen
> It's just so completely ridiculous to see all this whining and stupid
> statements coming up whenever someone mentions SGI and Linux in the
> same sentence. To get so hung up in these more or less (usually less)
> interesting press statements is just super childish. SGI has stated
> again and again that the MIPS line will continue for *quite* a while,
> and that means IRIX too.
Yes, but what we are concerned about is just what "continue" implies
with respect to IRIX.
* Continued support?
Most definitely, for many years to come.
* Continued implementation for new CPU's / systems?
Yes, SGI has explained to us that this will happen, and
that's good news.
* Continued program development?
No straight answer from SGI on this one. "Will you sell bananas?"
gets answered with "we have excellent apples and soon we will
have nice oranges too. And we already TOLD you nobody wants
a million dollar cucumber".
Regards,
--
*Art
Yes, we have no bananas...
Michael A. Thompson
Remember when SGI did the NT move and everybody said: WHY NT!!!!
PUT LINUX ON IT!!! Now they are putting Linux on it and they say:
WHY LINUX!!!
Strange.....
Michael
Tor Arntsen wrote:
> In article <37B25A72...@broomstick.com>,
> Arthur Hagen <a...@broomstick.com> writes:
>
> >It's indeed a sad day. Upto now I have heartily recommended IRIX
> [...]
>
> It's just so completely ridiculous to see all this whining and stupid
[...]
>
>
> So can we just cut the crap, please?
>
> -Tor
--
----------------------------------
Michael A. Thompson
[IRIX - NeXTStep - Linux - MacOS - Windows]
Home: (940)382-2086
E-Mail: mat...@jove.acs.unt.edu
----------------------------------
EVILjosh
> Your almost certainly not comparing the same thing, the cost of Irix
> depends upon the size and type of system, in the UK the list price is
> 260 pounds for a workstation (O2, Octane and deskside Onyx), for an SGI
> 2100 the price is 870 pounds and still more for an Origin2000.
This is AWE vs. ASE? I've not done an inst step/list of both.
EVILjosh
> > With pleasure. I pick OpenBSD.
> > I used to run NetBSD too on an 660AV which I liked alot.
>
> A Quadra? Interesting. I didn't know there was a UNIX flavor available
> for it. Did X run on it?
I didn't configure X on that system. As it was, it was a little
slow but I didn't mind. It ran in console mode; I like(d) that.
Andy Catero
>
> Penio Penev <pe...@venezia.rockefeller.edu> wrote:
> > Good questions. But where are the answers?
>
"Dear John..."
Is SGI going to release a roadmap for the cpu path for the o2
if there is one? I know the o2 isn't SGI's main focus but
it would be nice to know.
| Why are systems still being sold with disks as small as 4GB
| and RAM as small as 128MB? Why are O2s still so expensive compared
| to the power offered by the VW320 and VW540?
> > Because they run IRIX, of course -- you get more value, you pay more
> > money.
>
> That's nonsense. Are you suggesting there's a $6000 markup just
> because IRIX is on the machine?
I can't help but think SGI believes their digital media api's are worth
that extra 6000 amount of money for irix, so people will buy o2's over
their
visual pc's to get the os and api adavantage. The 2 reasons I can think
of to buy an o2 is for video purposes and for s/w development for the
larger machines.
Lumis.com benchmarks make the o2 look very bad in the 3d modelling and
while the
o2 r1000 rendering scores aren't bad, they lose on price/performance. It
would
be interesting to know if the o2 had the 30 to 50 million dollars in
sales like the visual pc.
I doubt it.
>
> O2 is too expensive. The hikes for R12K are far too high. R7K should
> be used instead in O2 - it'd be *much* faster than R12K at the same
> clock speed (especially for fp) and much, much cheaper.
>
It would be interesting to hear from someone in the know how a chip that
cost 140.00 in mass
quantities gets sold for 2,250. SGI will give a credit of 750.00 to
trade in
my old R5000 that didn't even cost that much for SGI to buy. The markup
on this chip
is very high. Even with the price of the cache memory, this still seems
way too high.
The new r5200 didn't require any patches or os revision to work, so I
wonder
what took SGI so long to upgrade from the R5000?
I want to upgrade my R5000 as I am sure many others do, so hopefully
with this
restructuring SGI will be able to lower their costs. If they can make
complete
linux boxes excluding graphics using the same mips chips for around
1,000. Least,
SGI could sell us just the chip for 1,000. See the liux cobalt raq
servers.
> > Isn't that a question for QED?
>
> Actually probably not. They designed the chip, but I doubt they're
> at the top of the pack when it comes to SGI deciding whether or
> not to use it and who to fab it if it were (eg. NEC).
QED was shipping the R5271 at lower speeds for awhile before SGI went
with the R5200.
At least that is what their press release said.
Andy Catero
Alexis Cousein
> Yes, but what we are concerned about is just what "continue" implies
> with respect to IRIX.
>
> * Continued support?
> Most definitely, for many years to come.
>
> * Continued implementation for new CPU's / systems?
> Yes, SGI has explained to us that this will happen, and
> that's good news.
>
> * Continued program development?
Read the disclaimer. These are my opinions, not SGI's, but hey, some
people on comp.sys.sgi.* are foolish enough to attach some credibility
to my opinions ;).
For some of the spots I'm interested in, the team that is implementing
the features on IRIX is still happily producing away, so the answer
would have to be yes. It appears to be really alive, not on suspended
animation, and news of its impending demise is greatly exaggerated.
While SGI is, I believe, going to push Linux to have as many features so
that as many people can choose between Linux and IRIX, I don't think SGI
is stupid enough to think it's going to satisfy current IRIX customers
totally (and the wish list of those who had UNICOS machines and use IRIX
machines now) for some time.
--
<standard disclaimer: these are my personal views, not SGI's>
Alexis Cousein a...@brussels.sgi.com
Systems Engineer SGI Belgium
Alexis Cousein
>
> All of this scares me. For most of this year I have been scraping and
> saving so that I could get my own SGI/Maya animation workstation. Now
> it seems that when my goal is finally in sight the whole picture is
> getting cloudy again.
>
> I don't even pretend to understand big business, but it certainly
> cannot be GOOD business to be giving off so many negative signals and
> alienating the core customers.
I think the press is giving signals that are more negative than the SGI
execs themselves...
david_d...@non-hp-roseville-om3.om.hp.com
fl_a...@thepentagon.com (I R A Darth Aggie) wrote:
> As if. I've seen 'em die in many different ways...fortunately, those
where
> under a university program, so we resurrected them...
Fascinating. University (Varsity) contracts don't cover hardware...
perhaps you where thinking of Sun systems.
My 4D/35 still chugs away, serving NFS mounts day in, and day out. I
run my SGI systems 24x7 and, unless I put a 20" monitor on top of my
INDY, they all will continue to do so.
- David
Kristoffer Lawson
: Remember when SGI did the NT move and everybody said: WHY NT!!!!
: PUT LINUX ON IT!!! Now they are putting Linux on it and they say:
: WHY LINUX!!!
Don't generalize too much. It's the loud ppl who get the attention.
I was one of the people questioning the NT business but I think Linux
is definitely a move - if not in the right direction then at least somewhere.
Yes, Linux is overhyped (it's not really technically advanced or is the
GPL concept new) but why not use that?
If all the features of IRIX can be implemented on Linux - with the nice
integrated desktop we're familiar with - then hey, I'm at least not going
to complain.
--
- ---------- = = ---------//--+
| / Kristoffer Lawson | www.fishpool.com
+-> | se...@fishpool.com | - - --+
|-- Fishpool Creations Ltd - / |
+-------- = - - - = --------- /~setok/
Kristoffer Lawson
:> I would say it is the opposite. The fact that there are many different
:> environments, many different competing vendors, many different APIs
:> (there is no real standard audio API f.ex.) is exactly the reason why
:> UNIX is failing to compete with Windows. Simply, a large, unified
:> environment is much easier to sell than a little bit of this and a little
:> bit of that. You may not like that but that's how things are.
: But that is *BORING*.
Very true. Ahh.. the good old days when people still had fights over
which was the best computer (had a Sinclair Spectrum myself) :-) Pity
computing is so boring these days.
Network
Its time for *MY* soapbox.
First off, the only thing I think the whole Mips/Linux port would
bring to SGI users is a GREAT OS with tons and tons of applications.
Think about all the time all these amature and non amature programmers
have spent building this OS from the ground. With no single direction.
Oh we want a what? Yea this seems to work lets add it in the kernel
at ground zero!
Lets not forget .... about the APPLICATIONS!
Lets see, there is at *LEAST* 600 "nextstep like" window managers. Oh,
yes and ALL 920 emulators that let you run PACMAN and boot operating
systems from machines that I buy from the thrift store. But hey, if you
got the SVGA lib, you can also run SASTEROIDS. The asteroid clone with
nuggets that look like meatballs. I'm shure the MIPS port will have
support for that.
I would certainly hope the MIPS port will have the red-hat flavor of
looking like Windows 95!@#!@#!@##!@# So everyone feels RIGHT AT HOME!@#
All out seriously. Maybe the BSD folks can get some use out of the
Linsux source code, and give us OpenBSD or NetBSD on MIPS (they have had
Mips support for Decstation 5k style machines for as long as linux, if
not longer. Of course OpenBSD borrowed major parts from NetBSD). I
don't know if there is a huge point in it... but its a tad bit
more directed than Linux.
Linux is just hype. I just see tons and tons of Windows NT admin types
get all hyped up and go spastic. Then their high moment is shattered
when they learn they are going to have to type on the keyboard. (Not for
long the way Redhat is going). As it grows "stronger and stronger" and
supports more crazy stuff, it will eventually get just as bloated as
microsoft products. Hell, already with the 18 level transparent
Window managers... it gets sluggish on midrange pentiums.
Shure there is lots of software. But instead of porting Linux to Mips to
get the applications, why not work on getting the tools out to people so
they will develop more for IRIX? I'm not a programmer myself. I've heard
lots of programmers curse IRIX. I'm shure its lack of knowledge and of
course there are going to be some differences. Maybe its lack of TOOLS?
Most of your in-my-free-time Linux developers don't have money to buy
a proper compiler suite for IRIX. Linux has more support because its
cheap. (Most important software for Linux has BSD ports. I said ports, I
didn't say compiles right on both).
All I want is good audio tools! Thats all! Cubase clone, you know,
a Sound forge clone? Really kickin' CD-R utilities?
SILICON GRAPHICS INC. Thats right, I just called you SILICON GRAPHICS.
Pull a Sun! Release IRIX, the IDO, and MIPS-PRO for $10 each to home
users! There are lots of SGIs on EBAY (with inflated costs unless
ebay breaks... homeysan)! Thats alot of would-be linux using 15 year
olds getting Indys! You too could have the same exact
developer base as Linux! All to yourself! And your stock might go up
from all those sales! I bet you could make $5 extra on each order if
you sold posters too. (Seriously). Income is income!
There are TONS of people buying "legacy" SGI machines. As the big
companies turn away you have a whole new generation of people turning
in.
Free copies of major programs don't hurt either. Like that stunt
Oracle pulled to try to hurt Microsoft...
SGIs are hitting homes!
When I was younger I drooled over that $8k entry Indigo. Man one day.
10 years later...
My desktop BECAME an Indigo r3k entry. (Thanks Greg D!).... I achieved
that dream.
(then set my goals to a 4d/480).
(Thanks wolf)
(then set my goals to a Crimson).
(Thanks hillbilly auction rat in NC)
(now my goal is a challenge XL and to murder my roomates for complaning
about the power useage of my 2nd and 3rd goal).
(4d/480 with VGX still ISN'T TOO SHABBY!)
(Props to mashek on the ghetto indy, and Antipent on the 4d/420!)
On the Challenge XL. Any takers?
I try to mellow out and use everything. Every platform has something to
offer. Linux offers desktop users with lots of... wnidow managers. Irix
offers an out of the box OS that KILLS any other as far as utilities.
Look at Solaris out of the box. Look at Solaris out of the box with CDE.
Digital audio tools? bahaha. CDE looks like barf (I speak of 2.6).
My indigo from 1991 has features that are just coming around on the
desktop PCs as far as sound quality.
I'm totally stringing off course now. But hey, this will start
several new strings as people pick on their favorite part.
SGI should have evaluated all of free OSes. Give XFS to FreeBSD
and OpenBSD at least. They are operating systems good enough
for servers (as good as it gets on intel hardware), and open sourced.
They don't rely on media attention and books from Barnes and Nobles.
They rely on satisfied user base and a sane development process.
(Truth is, SGI could probably make millions selling a $500 purple
PC with the cube logo. But anyways)
Could you imagine a BeOS port to MIPS? That would be funky!
Oh. And sell Trusted IRIX cheap too. For $10.
MRNUTTY. YOUR IINNSSAANNEE!
- tm
email responses? Especially on that $800 Challenge XL
to "telmnstr" followed by a big, fat, @, then "757.org"
end the soapbox stance.
I R A Darth Aggie
<37BDCB...@krypton.engrs.infi.net> wrote:
+ Its time for *MY* soapbox.
Its a very pretty soapbox. Too bad it is 95% FUD. Hmmmm...just like the
microsquid crowd...
James
--
Consulting Minister for Consultants, DNRC
The Bill of Rights is paid in Responsibilities - Jean McGuire
To cure your perl CGI problems, please look at:
<url:http://www.perl.com/CPAN/doc/FAQs/cgi/idiots-guide.html>
Penio Penev
|> With all the intellectual property? I highly doubt it. BTW, is the
|> E2K sampling? What feature size do they use? Do they use copper?
|> SOI? Which fabs do they produce it in?
| The Russian team wants to do a full joint venture with a western
| firm which has the money. See the press release on the CIC
| at Berkeley.
So, is the E2K sampling?
'Cause Merced is, and I belive R14K is close to that stage.
And if E2K is not sampling, what E2K machines is SGI going to sell in
6 moths?
|> H1 and H2 were better that IA64 too, remember? Why were _they_
|> dropped?
| They were not *significantly* better. Besides, some of the people
| designing IA64 came from MIPS anyway. Lots of staff poaching in
| winter of 1996.
How "significantly better" is the E2K (vapor-ware, I presume) to
Merced (sampling)?
| 20% better isn't enough to justify the much higher cost of
| non-mass-produced CPUs.
Exactly. So abstract speed does not wil always.
|> Why, then, did SGI pay a fine recently for selling computers to the
|> Russians (and why were they buying it, if they have better ones)? ...
| If it's what I think you're referring to, it had to do with sales of
| supercomputers and nothing to do with CPU designs.
But, if the Russians had "vastly superior" CPUs, why buy a bunch of
R10Ks?
|> ... And
|> why, then, was /Serge moaing and groaning recently that when he was in
|> Russia, he could not use the latest-and-greatest machines to do
|> computational chemistry because of the embargo -- didn't he have
|> access to those better CPUs?
| No, he doesn't have access to those better CPUs, because the
| Russian team don't have a fab partner. Read the PR.
So, E2K is vaporwave. And how can we compare E2K (vaprware) to Merced
(sampling)?
| E2K is between 3 and 5 times faster than IA64 and runs at higher
| clock speeds anyway.
I did not head an answer to the questions -- on what feature size, do
they use copper or aluminium, do they use SOI?
'Cause, if I am free to disregard the answers to those, I'll make the
correct statement that the Pentium is 3 to 5 times faster than the
Pentium, and runs at higher clock speeds anyway.
| There's actually a number of technologies in Russia which have
| been far superior to western tech for many years.
Here we agree. but semiconductor technology is _not_ one of those.
An will not be, for the forseeable future.
| E2K runs all x86 code anyway btw.
The Pentium, too.
|> This questions you need to address to Her Majesty's Tax Office. SGI
|> has repeatedly stated that they make _exactly_ the same profit on
|> every piece of service and/or equipment, regardless of geographic
|> region.
| I don't believe this I'm afraid. Even without the UK VAT tax
| of 17.5% the price is still 500 UKP, ie. $800. Something stinks.
As somebody pointed out, that might be AWE vs. ASE issues.
| IRIX should be sold for the fab/stamp cost of the CD media, manuals,
| box and time, etc. The sw itself should be free IMO.
That's why SGI is moving to Linux, right :-)
[the price of O2]
|> Because they run IRIX, of course -- you get more value, you pay more
|> money.
| That's nonsense. Are you suggesting there's a $6000 markup just
| because IRIX is on the machine?
In a machine that is $7500 list, and can be had for $5K academic, I
don't see how can there be a $6000 markup.
| O2 is too expensive.
That's what the 320 is for, right?
| The hikes for R12K are far too high.
That's what the 540 is for, right?
| R7K should
| be used instead in O2 - it'd be *much* faster than R12K at the same
| clock speed (especially for fp) and much, much cheaper.
|> Isn't that a question for QED?
| Actually probably not.
Is the R7K sampling?
| They designed the chip, but I doubt they're
| at the top of the pack when it comes to SGI deciding whether or
| not to use it and who to fab it if it were (eg. NEC).
[There is no information of what that _that_ is that is being
discussed below, so the point is moot.]
|> But that's the whole story with NVIDIA, right?
| NO.
| NVIDIA are doing the next mid-range stuff. Read the performer
| news bulletin by Allan Schaffer.
Penio Penev
| In article <pYHu3.45$Z4.3...@rockyd.rockefeller.edu>, Penio Penev
| <pe...@venezia.rockefeller.edu> wrote:
|> | Russian CPUs have been better than
|> | western designs for years.
|>
|> Why, then, did SGI pay a fine recently for selling computers to the
|> Russians (and why were they buying it, if they have better ones)?
| Just because you have a great CPU design, doesn't mean you've got the fab
| facilities to realise it.
Which makes my point -- comprating vaporware (E2K) to sampling silicon
(IA-64) is futile.
Doug Dingus
Network wrote:
(snip)
> SILICON GRAPHICS INC. Thats right, I just called you SILICON GRAPHICS.
> Pull a Sun! Release IRIX, the IDO, and MIPS-PRO for $10 each to home
> users! There are lots of SGIs on EBAY (with inflated costs unless
> ebay breaks... homeysan)! Thats alot of would-be linux using 15 year
> olds getting Indys! You too could have the same exact
> developer base as Linux! All to yourself! And your stock might go up
> from all those sales! I bet you could make $5 extra on each order if
> you sold posters too. (Seriously). Income is income!
>
> There are TONS of people buying "legacy" SGI machines. As the big
> companies turn away you have a whole new generation of people turning
> in.
I run training classes on a set of INDY R5k machines. In the 4 years
that those machines have been running I have never had any trouble that
was not my fault. Students come in and can't believe that they are
150Mhz machines! Many ask if they also can get used ones.
You are right on about new people turning in. They see the machine,
and its GUI, and they know just like we do that it will be fun to use.
There is power there that you just don't get with other operating
systems. How many other platforms can you type 'man audio', and get a
short education along with the options for the built in tools on your
particular machines. None! Any teenager with an interest in these sort
of things would kill for that type of enviroment.
I really kind of missed UNIX early on. Just didn't work in the right
places. (bummer all I saw was SCO and that was scary!) Linux
interestingly enough was the catalyst for my return. Each time I learn
to do something on a UNIX machine, I get the feeling that I am learning
to do it very well. On windows the feeling is more like, "How much do I
have to pay, and what do I click to get you do do it for me?".
Anyone that appreciates IRIX, or Linux for that matter wants to control
the computer. Make it do what they want. All the others are just
users.
>
> Free copies of major programs don't hurt either. Like that stunt
> Oracle pulled to try to hurt Microsoft...
>
> SGIs are hitting homes!
Yeah like mine!
(lost of SGI braggin deleted :) )
> end the soapbox stance.
Rant = 1
The only bitch I have about IRIX is common applications. It is a very
friendly well documented UNIX. Applications are where where SGI missed
the boat tho. Maybe even was not their fault. Every one that I know
that is running Office is trying to push their SGI machines out the
door, even though some of them have replaced their "Cheap" PC machines
twice during the lifetime of their SGI. Why is that? Office that's
why. Life sucks.
Rant = 0
Later,
Doug Dingus
App. Eng. / Trainer
Acuity Inc.
Portland, Oregon
SDRC AliasWavefront CADKEY Solid Edge
SGI Intergraph Unix NT Linux... Too many good toys, just not enough
time!
EVILjosh
| O2 is too expensive.
> That's what the 320 is for, right?
| The hikes for R12K are far too high.
> That's what the 540 is for, right?
Yeah but what if I DONT want 320/540?
I dont like Wintel. I prefer SGI hw
running IRIX.
This is not a clear cut solution.
EVILjosh
> : But that is *BORING*.
>
> Very true. Ahh.. the good old days when people still had fights over
> which was the best computer (had a Sinclair Spectrum myself) :-) Pity
> computing is so boring these days.
Well that's true for the environment as a whole but it is easy
to make your own fun.
I overheard a conversation last night where some people were
discussing their paying US$650.00 for MS ORifice2000. As I listened,
I knew that this discussion comprised about 70% of what computer world
is to them. Not for me.
Find your own ground or if you're a ChemBros fan, dig yer own hole.
Jerry Jorgenson
> | O2 is too expensive.
>
> > That's what the 320 is for, right?
>
> | The hikes for R12K are far too high.
>
> > That's what the 540 is for, right?
>
> Yeah but what if I DONT want 320/540?
> I dont like Wintel. I prefer SGI hw
> running IRIX.
>
> Josh...
> _/_/_/ _/ _/ _/ _/
Hear, Hear!! I agree completly, who would want a Wintel solution?
Jerry
--
Jerry Jorgenson
je...@j3iss.com
http://www.j3iss.com/
-----------== Posted via Newsfeeds.Com, Uncensored Usenet News ==----------
http://www.newsfeeds.com The Largest Usenet Servers in the World!
------== Over 73,000 Newsgroups - Including Dedicated Binaries Servers ==-----
Penio Penev
| | O2 is too expensive.
|> That's what the 320 is for, right?
| | The hikes for R12K are far too high.
|> That's what the 540 is for, right?
| Yeah but what if I DONT want 320/540?
| I dont like Wintel. I prefer SGI hw
| running IRIX.
Hence, you pay more, for the better product (the one running IRIX). QED.
EVILjosh
> Hence, you pay more, for the better product (the one running IRIX). QED.
Fine by me. You get what you pay for.
Yours,
Josh...
_/_/_/ _/ _/ _/ _/
_/_/ _/ _/ _/ _/
_/_/_/ _/ _/ _/_/
EVILjosh
> Remember when SGI did the NT move and everybody said: WHY NT!!!!
> PUT LINUX ON IT!!! Now they are putting Linux on it and they say:
> WHY LINUX!!!
Different product line. What they do now directly effects IRIX
in the future as IRIX enters this new products frey.
Arthur Hagen
> Merced isn't sampling yet. It has taped out, and they expect first
> silicon back from the fab in a couple of weeks. Which may or may not
> work well enough for engineering samples.
Note that Intel usually spends a year from the prototype release until
the production release. It's interesting that HP, who helped Intel
develop Merced, recommends waiting for McKinley instead of going for
Merced. (McKinley is planned to be released during the second half of
2001.) And the MIPS 64-bit CPU's have been around for quite some
time...
Regards,
--
*Art
Andrew Maizels
> Which makes my point -- comprating vaporware (E2K) to sampling silicon
> (IA-64) is futile.
Merced isn't sampling yet. It has taped out, and they expect first
silicon back from the fab in a couple of weeks. Which may or may not
work well enough for engineering samples.
Andrew.
--
Biting weevils by moonlight,
\\\\\ Eating bugs by daylight,
\\\\\\\__. Never running from a food fight,
____\\\\\\\'/_______She is the one named Sailor Hedgehog!
mr.pink
Hey Josh, just a thought here, but wouldnt the vw line actually become
more attractive (os-wise) with a decent linux implementation running on
it instead of NT?
at the very least it would be UNIX system again...
Y.s.,
Menno
Richard Masoner
Art wrote:
>> 2) ...it is much easier to
>> get IRIX technical goodies into the Linux universe, than it is to
>> get them into the NT universe....
> Out of curiosity, how is it easier to port an IRIX goodie to Linux than
> a Linux goodie to IRIX?
Read Mashey's statement again, Art.
Richard Masoner
Arthur Hagen
Stop misquoting, mr. Masoner. You're making a fool out of yourself,
since everyone easily can go back and look at the original text.
The full text I replied to (and which I quoted and you mangled) was:
> 2) Linux has some complementary virtues, and it is much easier to
> get IRIX technical goodies into the Linux universe, than it is to
> get them into the NT universe, or to get the applications from
> either of those places onto IRIX at any reasonable cost.
Excluding NT from the above statement, I read it as saying it's easier
to port goodies from IRIX to Linux than from Linux to IRIX at any
reasonable cost. Thus my question.
--
*Art
mr.pink
> Lumis.com benchmarks make the o2 look very bad in the 3d modelling and
> while the
> o2 r1000 rendering scores aren't bad, they lose on price/performance. It
> would
> be interesting to know if the o2 had the 30 to 50 million dollars in
> sales like the visual pc.
> I doubt it.
well, i have a l33t vapour idea: put some geometry acceleration chips on
a 64bit pci card, write a nifty hack for irix, and voila octane
performance!
now, all i need as 40m USD from some rich company that wants to produce
& distribute it. i think i'll consult my E2K friends in russia about
that issue ;)
> >
> > O2 is too expensive. The hikes for R12K are far too high. R7K should
> > be used instead in O2 - it'd be *much* faster than R12K at the same
> > clock speed (especially for fp) and much, much cheaper.
> >
hear hear! the r7k has an on-die 2nd level cache memory of 256kb.
remember what the 128kb on-die cache did for the celeron? Although the
R7000 does not have the (semi)advanced design of the r12k, it sure beats
the living daylights out of the r5k (13 specint/15 specfp?).
check out ian's site (http://www.futuretech.vuurwerk.nl) for more info
on why the r10/12k in an O2 isnt as powerful (fp-wise) as in the
octane/o2k/onyx2.
> It would be interesting to hear from someone in the know how a chip that
> cost 140.00 in mass
> quantities gets sold for 2,250. SGI will give a credit of 750.00 to
> trade in
from what i understood did QED not only want to sell the R7k for a
140.00 in mass quantities, rather they'd be manufactured on that
cutting-edge copperwire technology. 1.21 gigahertz any1?
> my old R5000 that didn't even cost that much for SGI to buy. The markup
> on this chip
> is very high. Even with the price of the cache memory, this still seems
> way too high.
> The new r5200 didn't require any patches or os revision to work, so I
> wonder
> what took SGI so long to upgrade from the R5000?
one reason could be that the r5a never did appear. we can only guess as
to why sgi/mips killed that project.
obtw, the rm5200 actually does require a patch! it's only supported
since 6.5.3 i believe.
>
> I want to upgrade my R5000 as I am sure many others do, so hopefully
yup, me too
> with this
> restructuring SGI will be able to lower their costs. If they can make
> complete
> linux boxes excluding graphics using the same mips chips for around
> 1,000. Least,
> SGI could sell us just the chip for 1,000. See the liux cobalt raq
> servers.
>
yes well if you cant wait, buy the rm5200 upgrade, desolder the cpu,
solder the R7K onto the PCB, make the necessary hardwire adjustments etc
etc. QED has a very informative pdf on their website about designing a
pcb that would house either the r5271 or r7k.
only thing left to do is getting the os-side done ;)
> QED was shipping the R5271 at lower speeds for awhile before SGI went
> with the R5200.
> At least that is what their press release said.
>
true. they also designed the cpu for use in routers and the like, why it
should end up in my o2 actually puzzles me given its performance.
Yours,
Menno
John R. Mashey
|> Note that Intel usually spends a year from the prototype release until
|> the production release. It's interesting that HP, who helped Intel
|> develop Merced, recommends waiting for McKinley instead of going for
|> Merced. (McKinley is planned to be released during the second half of
|> 2001.) And the MIPS 64-bit CPU's have been around for quite some
|> time...
1) A standard interval is 12-18 montsh from tapeout of a brand-new micro
until you see production systems.
2) Both HP and SGI are doing what any sane systems company does in teh
middle of a transition: offer a long overlap period when both CPUs
are available, so that customers can convert when they feel like it,
which varies tremendously. (Also, both HP and SGI have struggled to
explain this overlap, and it never works very well in linear words,
it's a lot better on graphical chrts where you can see roadmaps.)
3) HP saying *its* customers may wait for McKinley. Note that the
early IA-64s will first appeal to technical-compute customers, who:
a) Often buy early things.
b) Like floating-point
c) Often have a small amount of portable code to move.
They will not appeal so much, at the beginning, to enterprise data managmeent
customers, who:
a) Seldom buy early things.
b) Usually don't care much about floating-point [Wall Stree "rocket science"
types mostly act like technical-compute customers].
c) Needs tons of programs
Both HP and SGI have both kinds of customers, but a much bigger percentage
of SGI's customers are of the "innovator/early-adoptor/early-majority"
technical flavor than are HP's, which means that Merceds are a better
match for SGI's customer base. Compilers (at least SGI's) are already
well-matched to the IA_64 floating-point architecture, whereas every
compiler person I've talked to (not just at SGI) believes that the
integer-side features of IA_64 are fine, but will take a while longer
to get up the learning curve. (i.e., this is the predicated-logic feature,
good for helping gnarly/branchy OS, networking, and DBMS code.)
4) SO, anyway, both of us are overlapping, SGI just has a better match with
customers, and perhaps, with compilers, to take advantage early of IA-64.
--
-john mashey EMAIL: ma...@sgi.com DDD: 650-933-3090 FAX: 650-933-4392
USPS: SGI 1600 Amphitheatre Pkwy, ms 005, Mountain View, CA 94043-1351
David Hinds
:
: > 2) Linux has some complementary virtues, and it is much easier to
: > get IRIX technical goodies into the Linux universe, than it is to
: > get them into the NT universe, or to get the applications from
: > either of those places onto IRIX at any reasonable cost.
:
: Excluding NT from the above statement, I read it as saying it's easier
: to port goodies from IRIX to Linux than from Linux to IRIX at any
: reasonable cost. Thus my question.
John Mashey did post a reply to this specific question.
The Linux goodies that people are interested in are *applications*.
Some of these are open source and are mostly already ported to IRIX;
most of the commercial ones are not open source, and SGI is not going
to have much success convincing other vendors to do IRIX ports.
There are lots of IRIX OS goodies that people are interested in.
Since SGI owns these things, they can choose to release this code, or
port it, or whatever, on their own. They don't have to convince
anyone else that it is a good idea.
-- Dave Hinds
mr.pink
> yes well if you cant wait, buy the rm5200 upgrade, desolder the cpu,
> solder the R7K onto the PCB, make the necessary hardwire adjustments etc
> etc. QED has a very informative pdf on their website about designing a
> pcb that would house either the r5271 or r7k.
see
http://www.qedinc.com/downloads/qan-101_rm5270-71-rm7000_land_pattern.pdf
and
http://www.qedinc.com/downloads/qan-108_rm7000-migration.pdf
for more info.
this just goes to show that QED has given it a lot of thought already;
the actual execution of this implementation would be child's play for
SGI's hardware-dev department imo.
Menno
John R. Mashey
|> The full text I replied to (and which I quoted and you mangled) was:
|>
|> > 2) Linux has some complementary virtues, and it is much easier to
|> > get IRIX technical goodies into the Linux universe, than it is to
|> > get them into the NT universe, or to get the applications from
|> > either of those places onto IRIX at any reasonable cost.
|>
|> Excluding NT from the above statement, I read it as saying it's easier
|> to port goodies from IRIX to Linux than from Linux to IRIX at any
|> reasonable cost. Thus my question.
And I posted shortly thereafter explaining that assymetry of my
statement got lost in the translation. This has nothing to do with
porting, as I'd expect that:
1) Most Linux user source codes port relatively easily to IRIX [and many have]
2) Many applications in the industry can be compiled either place.
3) Some applications (like that use OpenGL seriously, or any of the
IRIX APIs not found on Linux) do not currently port to Linux usefully.
(People are busy adding some of the more popular APIs that IRIX shares
with other commerical UNIX systems.) See, for general info: http://www.sgi.com/developers/technology/irix.html
and for specific open source work:
http://oss.sgi.com
To try again, more precisely, albeit with more words in [].
it is much easier to get IRIX technical goodies into the Linux universe
[by Linuxizing (?, if that's the right verb) them, and then offering them to
the community as open source, or offering them as binary applications/drivers
that run on/with Linux, either free or for money], that it is to
get them into the NT universe, or to get the applications [usually in
binary form] from either of those places onto IRIX at any reasonable cost.]
G. Douglas Davidson
In article <37C092F0...@NOSPAM.fibre.a2000.nl>, "mr.pink" <m.b...@NOSPAM.fibre.a2000.nl> writes:
> NNTP-Posting-Host: node127e.a2000.nl
> Mime-Version: 1.0
> Content-Type: text/plain; charset=us-ascii
> Content-Transfer-Encoding: 7bit
> X-Trace: weber.a2000.nl 935366953 21158 62.108.18.126 (23 Aug 1999 00:09:13 GMT)
> X-Complaints-To: ab...@a2000.nl
> NNTP-Posting-Date: 23 Aug 1999 00:09:13 GMT
> X-Mailer: Mozilla 4.61 [en] (WinNT; I)
> X-Accept-Language: en
> Lines: 9
> Xref: speed.city-net.com comp.sys.sgi.misc:53132
This is the box I'd like to pick up. If SGI can provide a system that is
tuned to run Linux well with all of the cool graphics stuff I'd be psyched.
The GLX stuff along with the port of XFS goes some way toward this. My IRIX
box is _still_ more responsive than my vastly overpower Linux box under load.
The realtime stuff, like mp3's and movies, is much less chunky under IRIX.
So any help SGI could provide to smooth out the Linux kernel would be cool.
I'd even tolerate the (in my opinion) less elegant interfaces (KDE and Gnome)
until they can be cleaned up.
Basically, if they could just take everything that is cool about my O2 and
stick it into their VW Linux package, I'd pick one up quick.
--
-----
G Douglas Davidson | CityNet, Inc.
dou...@city-net.com | Pittsburgh, PA
voice: 412.481.5406 | fax: 412.431.1315
Doug Dingus
"G. Douglas Davidson" wrote:
(snip)
>
> This is the box I'd like to pick up. If SGI can provide a system that is
> tuned to run Linux well with all of the cool graphics stuff I'd be psyched.
> The GLX stuff along with the port of XFS goes some way toward this. My IRIX
> box is _still_ more responsive than my vastly overpower Linux box under load.
> The realtime stuff, like mp3's and movies, is much less chunky under IRIX.
> So any help SGI could provide to smooth out the Linux kernel would be cool.
> I'd even tolerate the (in my opinion) less elegant interfaces (KDE and Gnome)
> until they can be cleaned up.
>
> Basically, if they could just take everything that is cool about my O2 and
> stick it into their VW Linux package, I'd pick one up quick.
I'll second that. Right now someone would have to pry my O2 away from
my cold dead fingers. It is the machine where all the real work gets
done.
--
Douglas V. Mashek
I have on my desk a 200MHz R5000 O2, and right next to it is a Dual P3 running at
504MHz w/112MHz bus, 16MB TNT AGB card, running RedHat 6.0. I use the O2 more for
a desktop, and the Linux box more for a server. When it comes right down to it,
the O2 still beats the linux box for I/O, and it even let's me move the mouse when
I'm loading it down;^) The only 2 points of usablility on the Linux box that I
like is the 1600x1200x32bit graphics, and it's uncanny abilility to gzip two
streams at once really fast. Well, one more thing, games. It would be cool if
there were some games that really used the graphics horsepower of SGIs. Games are
very cool way to show off the graphics power of an SGI to a PC Head that only
knows graphics by how cool games look. From what I can tell, there are a lot of
people in that category.
Just my ramblings, no disclaimer needed ;^)
Doug
Doug Dingus wrote:
--
Mashek Consulting Corp. email: do...@mashek.com
SGI Reselling and Consulting www: http://www.mashek.com
16329 Kenyon St. NE phone: 612-434-3945
Ham Lake, MN 55304
Beau Vrolyk
Scott Elyard wrote in message ...
>In article <pYHu3.45$Z4.3...@rockyd.rockefeller.edu>, Penio Penev
><pe...@venezia.rockefeller.edu> wrote:
>
>> | Russian CPUs have been better than
>> | western designs for years.
>>
>> Why, then, did SGI pay a fine recently for selling computers to the
>> Russians (and why were they buying it, if they have better ones)?
>
Everyone,
Let me add a few points of fact. SGI was investigated for having shipped
computers to Russia, as was IBM. Following that investigation SGI was
absolved of any guilt, please see our press release of Feb. 18, 1997. IBM,
on the other hand, is reported to have paid $8m in penalties. Sorry to pour
water on the fire by providing facts.
Beau
John R. "Beau" Vrolyk
Sr. Vice President
Product Group
SGI