OT: What is this Intel Parallel Studio?
Robert Baer
and what does it do?
Sylvia Else
> Go that CD in the mail with a mag and am curious as what is it for and
> what does it do?
Robert Baer
instructions and data out the same pins so cannot run in parallel in the
long term (am ignoring short term storage in CPU cache).
Sylvia Else
Hmm...
Well, I suppose if Intel can get people writing software that would run
faster on systems with multiple memory buses, then they'll be able to
justify pushing such systems into the general market.
They can also push the advantages of bigger caches.
But I'm sure Microsoft will quickly consume any performance increase
with yet more software bloat.
Sylvia.
Nobody
>>> Go that CD in the mail with a mag and am curious as what is it for
>>> and what does it do?
>>
>> http://software.intel.com/en-us/intel-parallel-studio-home/
> They have gotta be joking; the so-called multiple cores
What's with the "so-called"? The current generation of CPUs do have
multiple cores.
> must ship
> instructions and data out the same pins so cannot run in parallel in the
> long term (am ignoring short term storage in CPU cache).
That's only an issue if you're saturating the memory bandwidth, and
there are plenty of applications which don't.
Joel Koltner
news:pan.2009.11.21....@nowhere.com...
> That's only an issue if you're saturating the memory bandwidth, and
> there are plenty of applications which don't.
My impression that it's actually rather few of them!
Archimedes' Lever
So use a solid state drive as your virtual memory volume.
MeowSayTongue
<syl...@not.at.this.address> wrote:
>
>But I'm sure Microsoft will quickly consume any performance increase
>with yet more software bloat.
>
>Sylvia.
Yet another misdirected, follow the dumb flock dimwit.
Meow
Sylvia Else
It's seemed fair. Even ordinary operations take just as long as ever
despite the huge increase in bus, cpu and memory speeds.
Sylvia.
Sylvia Else
How would that help?
Sylvia.
Robert Baer
take advantage of that hardware, AND one _must_ also use software
written for the multiple cores.
This means that almost all of the private PC users have useless bloat
hardware and were cheated via improper marketing hype.
Joel Koltner
news:HpKdnXZT6I9LyprW...@posted.localnet...
> "So-called" refers to the fact that only server boards are able to take
> advantage of that hardware, AND one _must_ also use software written for the
> multiple cores.
Eh? Desktop and laptop PCs with multi-core CPUs seem to do just fine
scheduling threads across them under Win2K, XP, Vista, Win 7, Linux, the Mac
OS, etc.?
It's true that a lot of software isn't multi-threaded yet, although it's been
changing a lot this past year -- LTSpice now is, EM simulators from the likes
of Agilent and AWR are, AutoCAD and SolidWorks are, and of course many games
are. Even a simple dual-core system is nice in that if you have some old
non-multi-threaded number cruncher simulation running: Having another free
core around allows the Usenet reading/posting you do while waiting for it to
finish :-) to run quickly.
> This means that almost all of the private PC users have useless bloat
> hardware and were cheated via improper marketing hype.
I suppose so, but I'm guessing that after somewhere in the late '90s the
average PC buyer no longer had much of a clue as to what, exactly, they were
buying at the hardware level. PCs are like any other appliance today... they
sell more on "looks" (flashly graphical demos and -- particularly for
laptops -- mechanical design) tha anything related to the technical specs.
---Joel
Sylvia Else
> "Robert Baer" <rober...@localnet.com> wrote in message
> news:HpKdnXZT6I9LyprW...@posted.localnet...
>> "So-called" refers to the fact that only server boards are able to take
>> advantage of that hardware, AND one _must_ also use software written for the
>> multiple cores.
>
> Eh? Desktop and laptop PCs with multi-core CPUs seem to do just fine
> scheduling threads across them under Win2K, XP, Vista, Win 7, Linux, the Mac
> OS, etc.?
>
> It's true that a lot of software isn't multi-threaded yet, although it's been
> changing a lot this past year -- LTSpice now is, EM simulators from the likes
> of Agilent and AWR are, AutoCAD and SolidWorks are, and of course many games
> are. Even a simple dual-core system is nice in that if you have some old
> non-multi-threaded number cruncher simulation running: Having another free
> core around allows the Usenet reading/posting you do while waiting for it to
> finish :-) to run quickly.
That was always possible just by tweaking the process priorities.
Sylvia.
Paul Keinanen
<rober...@localnet.com> wrote:
> "So-called" refers to the fact that only server boards are able to
>take advantage of that hardware, AND one _must_ also use software
>written for the multiple cores.
> This means that almost all of the private PC users have useless bloat
>hardware and were cheated via improper marketing hype.
The nice thing in a dual core architecture is that the other core can
be used to run the OS and the user interface bells and whistles, while
the other core can be used for actual work.
In some way, this arrangement resembles a time sharing systems I used
in the 1970's, one processor with 32 KW memory was running the OS,
while the other core with the same amount of memory served dozens of
simultaneous time share users.
Anyway, I do not understand what is the point with four or more core
processors on desktop, with only one core doing some useful work,
while 3+ cores will be used just to run the OS. I do not promise to
eat my hat, since no doubt Microsoft is going to find some "useful"
bells and whistles to run on any additional cores :-).
Paul
Martin Brown
> Nobody wrote:
>> On Fri, 20 Nov 2009 15:57:35 -0800, Robert Baer wrote:
>>
>>>>> Go that CD in the mail with a mag and am curious as what is it
>>>>> for and what does it do?
>>>> http://software.intel.com/en-us/intel-parallel-studio-home/
>>> They have gotta be joking; the so-called multiple cores
Intel multiple cores and before that hyperthreading have been around for
a while now as have the operating systems to cater for them. Plenty of
consumer software is not multithreading, but a lot of seriously compute
bound stuff for image or signal processing is fully multithreaded and
will use all CPUs if you allow it too. Some amateur chess engines use
all available cores if you let them. Makes the machine slow...
>>
>> What's with the "so-called"? The current generation of CPUs do have
>> multiple cores.
>>
>>> must ship instructions and data out the same pins so cannot run in
>>> parallel in the long term (am ignoring short term storage in CPU cache).
>>
>> That's only an issue if you're saturating the memory bandwidth, and
>> there are plenty of applications which don't.
>>
> "So-called" refers to the fact that only server boards are able to
> take advantage of that hardware, AND one _must_ also use software
> written for the multiple cores.
Obviously you need to use software that can use multiple cores to gain
the performance advantage. But there is plenty of that about if you do
not live in the stone age.
> This means that almost all of the private PC users have useless bloat
> hardware and were cheated via improper marketing hype.
I am sure that plenty of end users have multicore CPUs and graphics
engines that are only really ever stretched when they run 3D games.
However, PC computer games is a very big domestic market segment.
Regards,
Martin Brown
krw
It won't, but did you really expect AlwaysWrong to be of any help?
MeowSayTongue
<syl...@not.at.this.address> wrote:
>MeowSayTongue wrote:
>> On Sat, 21 Nov 2009 11:11:20 +1100, Sylvia Else
>> <syl...@not.at.this.address> wrote:
>>
>>> But I'm sure Microsoft will quickly consume any performance increase
>>> with yet more software bloat.
>>>
>>> Sylvia.
>>
>>
>> Yet another misdirected, follow the dumb flock dimwit.
>>
>> Meow
>
>It's seemed fair. Even ordinary operations take just as long as ever
>despite the huge increase in bus, cpu and memory speeds.
>
>Sylvia.
The things they incorporate are things that are asked for, so adding
the code to accommodate those 'asked for' things would hardly qualify as
bloat.
The problem here is the paradigm for incorporation of an operational
element that *they* decide whether you need or not, and leave you no
selection 'switches' by which to cycle that choice in or out of your
personal set-up.
In other words, modularized, and selectable.
Yes, much like Linux is.
Since the Windows boys cannot make it that way, idiots call their
paradigm 'bloat'.
To me, that means that YOU did not examine the requirements analysis.
However Linux is far more operator labor intensive, and you must also
already know that only the top 25% of users could be comfortable on a
Linux box. Granted, now the distros are getting more and more
pushbutton, the mental and technical knowledge to use Linux based "OSes"
is higher than the "any monkey can use it" paradigm an OS like Windows
*has* to attempt to follow.
So Windows wins for several reasons in the consumer realm. Mainly
because most folks are nowhere near the intelligence or technical prowess
levels required to immerse themselves in Linux based worlds. They would
rather have fat, dumb, and happy, and call support if they need help.
Linux is gaining ground in business and embedded and network management
industries. More so even than most know or believe.
Kernels can be small. An OS, however, can grow to be very large,
especially over long time periods. All one needs do is examine history
to see that.
Making it all work together is the task.
So the question one needs ask one's self is "Does that goal get
reached."
Meow
CellShocked
Dual core unit here, now over four years old. My second. The first ran
for three years before I got this one, so I have been on dual CPU or core
for over seven years now, and I have been on Vista and Windows 7 since
they offered beta ISOs.
They DO make a difference. I would never go back, and even the MAME
(as an example) emulators of late would likely puke on an old single core
machine just by their more modern video implementations alone.
They barely run on a PS3 with Linux, due to the handcuffing thing.
CellShocked
wrote:
>The nice thing in a dual core architecture is that the other core can
>be used to run the OS and the user interface bells and whistles, while
>the other core can be used for actual work.
Exactly!
CellShocked
wrote:
> I do not promise to
>eat my hat, since no doubt Microsoft is going to find some "useful"
>bells and whistles to run on any additional cores :-).
Applications (games are applications) already do. Same can be said for
distributed computing apps like BOINC
CellShocked
<|||newspam|||@nezumi.demon.co.uk> wrote:
>Obviously you need to use software that can use multiple cores to gain
>the performance advantage. But there is plenty of that about if you do
>not live in the stone age.
You can bet that the oil tycoons are using advanced software engineers
to write their advanced programs.
CellShocked
<|||newspam|||@nezumi.demon.co.uk> wrote:
>I am sure that plenty of end users have multicore CPUs and graphics
>engines that are only really ever stretched when they run 3D games.
>However, PC computer games is a very big domestic market segment.
For the most part, right on the money.
One has to think, however, that there are a lot of kids out there that
are just as smart as you were when you were their age. And that tax their
PC pretty good, and that likely even have PCs that make your pale, due to
the family's wealth level differences.
This is despite the decline in the educational averages we are seeing
in the US these days. Some kids just get endowed with a knack for some
things.
Today though...
The kids don't even know how to pull their pants up these days, and the
really sad thing is that it has been happening long enough that their are
adult aged idiots out there that do not know how to wear their pants up
past their ass cracks too! Scary.
Joel Koltner
message news:1dggg5tfvuo6d2sme...@4ax.com...
> One has to think, however, that there are a lot of kids out there that
> are just as smart as you were when you were their age. And that tax their
I was a bit surprised the first time I was using a terminal emulator I wrote
to discover that scrolling through a text buffer became noticeably slower when
the machine was receiving RS-232 characters continuously (it was bit-banged
serial @ 2400bps, so used a fair amount of CPU cycles per character receiver).
Suddenly 1MHz (a 6502) no longer seemed "infinitely" fast...
> This is despite the decline in the educational averages we are seeing
> in the US these days.
I've seen ads on Craigslist where teens are selling their their advanced cell
phones (e.g., those running Windows Mobile or Androis or similar) because
they're "too complicated to use."
The grammar and spelling you see on Craigslist is generally a sad commentary
on society today.
---Joel
Jamie
Autism , retardation! They all have their excuses..
JosephKK
It would make the swap / virtual memory operate at near memory speeds,
instead of the 100,000 times slower disk speeds. Of course just using
more RAM is often cheaper and more effective (eliminate the need for
swap entirely and remove it).
Martin Brown
> On Sat, 21 Nov 2009 13:04:04 +1100, Sylvia Else
> <syl...@not.at.this.address> wrote:
>
>> Archimedes' Lever wrote:
>>> On Fri, 20 Nov 2009 17:00:26 -0800, "Joel Koltner"
>>> <zapwireD...@yahoo.com> wrote:
>>>
>>>> "Nobody" <nob...@nowhere.com> wrote in message
>>>> news:pan.2009.11.21....@nowhere.com...
>>>>> That's only an issue if you're saturating the memory bandwidth, and
>>>>> there are plenty of applications which don't.
>>>> My impression that it's actually rather few of them!
>>>>
>>> So use a solid state drive as your virtual memory volume.
>> How would that help?
>
> It would make the swap / virtual memory operate at near memory speeds,
> instead of the 100,000 times slower disk speeds. Of course just using
> more RAM is often cheaper and more effective (eliminate the need for
> swap entirely and remove it).
I am afraid it wouldn't help that much even with a RAID0 array of SSD.
The seek time is very much better than a physical platter drive but you
are still limited by the external bus transfer rate which is typically
for sustained activity on an SSD about 250MBps read and half that for
writes. The latter hurts in a virtual memory swap file application.
They are magic for random lookups of multigigabyte datasets though.
Regards,
Martin Brown
Sylvia Else
That's a rather different issue. The one under discussion here relates
to how much faster a dual-core processor can run than a single-core
processor when all the non-cached memory accesses still use the same
memory bus.
Improving disk transfer times is always desirable, but it doesn't relate
in any particular way to the number of processors on the die.
Sylvia.
JosephKK
Yes, and how much do you get out of a swap volume (not normally
RAIDable)? 50 MB/s? (sequential read, and how much of that occurs)
and do remember that random (to the disk) reads, which may be
frequent, incur head movement overhead of milliseconds. In SSD this
is microseconds.
Just reinforces my preference for more RAM instead.
Of course there is the write lifetime issue for SSD as well, if used
as swap, life could be on the order of months.
JosephKK
<syl...@not.at.this.address> wrote:
Parallel Studio is basically a marketing move. Home and most office
applications do not profit from multicore CPUs. I surely do not find
noticeably better response from MSWord (or competitors) on dual core
machines than i get on single core machines. I only see modest to
moderate differences in the simulation phase with LTSpice. Most
computer use is actually USER bound these days.