He was being a good scientist and trying to explain it in terms people with no understanding of anything technical, i.e. MS and Mac users, could understand.
Please! RTFA. You seem to forget that in congress someone pulled a switcheroo, not to mention the pressure the "if you dont give Bush a blank check you must be supporting the terrorists" shit this administration spreads put on all politicians.
DOJ CLAIM: Congress enacted the Patriot Act by overwhelming, bipartisan margins.
Congress voted overwhelmingly to pass the PATRIOT Act in October 2001. But Congress acted under intense time pressure and without serious debate and deliberation. The PATRIOT Act was signed into law a mere 5 weeks after the Administration's draft was first circulated - lightning speed for legislation. And on the House side, the version approved by the Judiciary Committee with some changes prompted by civil liberties concerns was replaced by a different version in the middle of the night, and a vote was taken just hours later - leaving members and their staff with literally not enough time to read what was in the lengthy bill. Any legislation adopted under these circumstances is likely to contain provisions that deserve to be revisited and corrected if appropriate.
I think the evidence points to the fact that Ashcroft has little no respect for this country's constitution or existing laws (remember that this was a big issue at the confirmation hearings, and is sadly proven true by his actions since then [not to mention he is a total wacko - spending $$$$$$ to cover up some nude statues is just one example]). So a bit of criticism is not only valid but (ironically) also patriotic if you're a US citizen.
I'm hoping this is bogus, but if its not, what search engine out there offers similar power and quality? Netscape search has sucked for years now. Lycos seemed to go down hill before that even...
SATA certainly works out of the box in Fedora test 3 (2.4.22 something or other kernel), the appear as your standard SCSI/dev/sd* devices. They might even work in older releases, but I haven't investigated that.
I believe the official release for the first non-beta Fedora in Nov 3.
Hopefully the more community-oriented nature of Fedora will work out well too...
This guy is way over hyping it - I've seen claims like this from people in the same space-based warfare branch of the military for several years now, but its always more of PR push to get funding than a realistic threat assessment.
Anyway, despite that, remember Reagan's "Star Wars" - despite the claims it was never, not would ever be, a good missile defense system. BUT it would be great as a first strike anti-satellite weapon system (and they knew this). Would have totally destabilized the world, as it would up the ante for a country to strike first and without warning.
M-x politics-mode-on
And as for next major conflict - aren't we in one? Shouldn't the US and its "willing" allies be worrying about sorting out the major conflicts they're already in? Afghanistan is falling apart already, and despite spending 1G$ a week the Iraq thing is not going well.
You seem to be ignoring the photoshop tests, you know, the application Apple makes a big deal about how fast it runs.
And the raid stuff makes little difference - as the table shows.
Granted, none of these are benchmarks I'm interested in, but its hardly one sided. The result is what you'd expect if you compare *official* SPECfp numbers for Opterons etc to the unofficial SPECfp numbers for IBM's (cough, I mean Apple, Steve Jobs made it himslef!!!) G5.
Yes, they do give the scalar (~unoptimized) results were the G5 does rather poorly. But I what I said was
..the vector tests use Altivec, so these results are what you can expect for an optmized (my typo!)application.
which is factually correct, as the Altivec tests are what you can expect from an optimized application. I was replying to the original./ post which incorrectly claimed
Even with un-optimized code and one processor, the G5 performance is impressive.
Someone please explain to me how 5177 MFLOPS and ~300 MFLOPS are even comparable.
As the Mathematica guy said, the competition is no longer high-end PCs, it's now $10,000 UNIX workstations...and the G5 is still faster than any of them.
Argh! Totally the wrong way to interpret the article.
Single CPU scalar benchmark
P4 2.66 Ghz --> 255 MFLOPS
G5 2.00 Ghz --> 254 MFLOPS
Then the guy basically switches on the part of his code (developed on a G4) that use Altivec (Apples SIMD instruction/pipeline), and wammo, reasonably good MFLOPS numbers (2755 MFLOPS) ensue (any modern processor can get over 1000 MFLOPS with suitably written code).
This guys code is parallelized (nice), so with the 2nd processor on he gets more done, hence 5177
MFLOPS.
But there isn't a intel or AMD comparison for the vector tests, cos his code doesn't seem to have been written to use their SIMD instruction sets.
The Intel and AMD pundits claim their SIMD architecture is superior to Apple's (e.g. see discussion in this article).
As the author of these benchmarks himself states Based on an extrapolation of current P4 results, the 2GHz G5 would lag newly announced 3.2GHz P4 systems in Jet3D scalar floating point performance by about 20%, but this kind of comparison is best deferred until G5-aware compiler tools become available (since a 20% performance gain is well within the potential of compiler optimization).
Finally, it is important to note that the current test does not factor machine cost or intended use into the picture, and that can have a large impact, especially in clustering applications.
I.e. a G5 is probably comparable to a top-end desktop 32bit PC for joe blogs applications once the small changes from G4-to-G5 specific optimizations take effect. A comparison of suitably SIMD-aware number crunching applications is still needed to evaluate absolute speed.
Processor power per unit dollar cost is another issue that the author hints at, which is probably not in Apple's favor.
You're absolutely correct -- it does give the false impression that G4/G5's have something "special" not available on Intel or AMD platforms (Like "Bluetooth" or "Airport" eh! Ahh, when will concern for the truth ever concern a marketing department?).
Some of the articles (AMDzone?) is saw discussing the G5 were of the opinion that the SIMD (Single Instruction/Multiple Data) designs on intel and amd chips was superior to Apple's.
The code used was developed on a G4, and the vector tests use Altivec, so these results are what you can expect for an optmized application.
The apples to apples (no pun intended) comparisons made to old P4s (not the vector plots) are not all that impressive, as other posts have pointed out. If anything, Apples latest and greatest is slower than an old P4.
What we need is a comparison using a code that is written to use both Altivec (on G4/5s) and sse/sse2/3dnow! (on pentiums/athlons) enhancements. Anyone got something like that?
In other news, norhtec.com anounce the world's fastest coffee-cup-sized computer (your coffee cup isn't 4.75 x 6.25 x 1.9! How do you get any work done?).
Comparison of SPEC CPU2000 benchmarks against suitably optimized(*) Apple G5 demonstrates mATX computers are infinitely faster than Apples latest offering at both integer and floating point computations.
(*) G5 test optimized by switching it off. "Its faster that way", claim benchmarking company.
Well, thanks for actually reading my post.
I base decisions on SPEC because in my 9 or so years
of watching SPEC cpu benchmarks (92, 95 and now the 2000 variety) I have found that the relative performance I get for my apps (grid-based astrophysical hydrodynamics with typical run times of up to a few weeks on a current top range machine, but also a variety of other floating point intensive data analysis) on various different machines (alphas, suns, linux etc) scales as expected based on the SPEC values, i.e.
to within a few percent, when you optimize appropriately (and I'm more than happy to use gcc). Its a fairer benchmark than most, and if you look what the FP tests are comprised of you'll see they're real scientific number crunching of a variety of types. If I want to buy something new and fast to work with, then a detailed inspection of the SPEC results
**that are on the official spec web site** (including the motherboard used, performance on specific aspects of the SPEC tests closer to my type of computing etc) is the way to go.
IMHO, unless you're trapped with only being able to run apps on one OS you should always base your purchase (if you're after number crunching power) on the highest SPEC FP_base/cpu_dollar_cost ratio you can find. Right now I estimate thats best for a 2.8 Ghz P4 (faster than that you pay too much extra for the top of the range).
As for this G5 stuff, the way I feel is...
1) Remember the G4 all the hype? Its not the first
time Apple has stretched the truth. People are right to be skeptical of *all* extraordinary claims, be it by Intel, AMD or Apple.
2) Forget the minute details -- the fundamental point is that the SPEC results apple quoted for P4s and Xeons are not in line with the tons of existing spec measurements with those systems. Its deliberate spin - yup, others do it to, but cheating is cheating.
3) Apple did use a version of gcc optimized for the G5 - you'd be naive to believe they didn't experiment until they found the compiler that gave the best results.
4) I don't know about mathematica, but I'm sure you will agree that Apple has a history of comparing apple-optimized photoshop against non-pentium-optomized photoshop.
5) AMDZONE has an insightful analysis of the whole G5 PR-thing, not just the SPEC values. See
http://www.amdzone.com/articleview.cfm?articleid=1 296
An an astrophysicist myself, I certainly won't be spending tax-payer dollars on a G5. You get more floating point bang-for-buck out of x86-like architectures with Linux at the moment - their are tons of SPEC CPU2000 FP scores available for P4's and athlons will beat the pants off the Veritest G5 scores (ignore the fudged veritest P4 scores). Go to 64bit and that'll still be true.
Same with single or dual processor.
Lots of astrophysicists use Macs notebooks instead of Windows, primarily because they haven't cottoned onto the fact that can do presentations (the main reason astronomers now get notebooks) with OpenOffice, or they're irrationally afraid
of installing and administering Linux. OOo works better than MS office in my experience.
Macs are nice - better than MS Windows (although not for games), but for the cost conscious heavy duty number cruncher you're better off with x86 or opteron under Linux. Besides - not all astronomical software has been ported yet!
Slashdot Mirror
He was being a good scientist and trying to explain it in terms people with no understanding of anything technical, i.e. MS and Mac users, could understand.
DOJ CLAIM: Congress enacted the Patriot Act by overwhelming, bipartisan margins.
Congress voted overwhelmingly to pass the PATRIOT Act in October 2001. But Congress acted under intense time pressure and without serious debate and deliberation. The PATRIOT Act was signed into law a mere 5 weeks after the Administration's draft was first circulated - lightning speed for legislation. And on the House side, the version approved by the Judiciary Committee with some changes prompted by civil liberties concerns was replaced by a different version in the middle of the night, and a vote was taken just hours later - leaving members and their staff with literally not enough time to read what was in the lengthy bill. Any legislation adopted under these circumstances is likely to contain provisions that deserve to be revisited and corrected if appropriate.
I think the evidence points to the fact that Ashcroft has little no respect for this country's constitution or existing laws (remember that this was a big issue at the confirmation hearings, and is sadly proven true by his actions since then [not to mention he is a total wacko - spending $$$$$$ to cover up some nude statues is just one example]). So a bit of criticism is not only valid but (ironically) also patriotic if you're a US citizen.
I'm hoping this is bogus, but if its not, what search engine out there offers similar power and quality? Netscape search has sucked for years now. Lycos seemed to go down hill before that even...
I believe the official release for the first non-beta Fedora in Nov 3.
Hopefully the more community-oriented nature of Fedora will work out well too...
Something actually useful on ./, except its hidden away in the science section instead of the front page.
Anyway, despite that, remember Reagan's "Star Wars" - despite the claims it was never, not would ever be, a good missile defense system. BUT it would be great as a first strike anti-satellite weapon system (and they knew this). Would have totally destabilized the world, as it would up the ante for a country to strike first and without warning.
M-x politics-mode-on
And as for next major conflict - aren't we in one? Shouldn't the US and its "willing" allies be worrying about sorting out the major conflicts they're already in? Afghanistan is falling apart already, and despite spending 1G$ a week the Iraq thing is not going well.
You seem to be ignoring the photoshop tests, you know, the application Apple makes a big deal about how fast it runs. And the raid stuff makes little difference - as the table shows. Granted, none of these are benchmarks I'm interested in, but its hardly one sided. The result is what you'd expect if you compare *official* SPECfp numbers for Opterons etc to the unofficial SPECfp numbers for IBM's (cough, I mean Apple, Steve Jobs made it himslef!!!) G5.
which is factually correct, as the Altivec tests are what you can expect from an optimized application. I was replying to the original ./ post which incorrectly claimed
.Single CPU scalar benchmark
P4 2.66 Ghz --> 255 MFLOPS
G5 2.00 Ghz --> 254 MFLOPS
Then the guy basically switches on the part of his code (developed on a G4) that use Altivec (Apples SIMD instruction/pipeline), and wammo, reasonably good MFLOPS numbers (2755 MFLOPS) ensue (any modern processor can get over 1000 MFLOPS with suitably written code).
This guys code is parallelized (nice), so with the 2nd processor on he gets more done, hence 5177 MFLOPS.
But there isn't a intel or AMD comparison for the vector tests, cos his code doesn't seem to have been written to use their SIMD instruction sets. The Intel and AMD pundits claim their SIMD architecture is superior to Apple's (e.g. see discussion in this article).
As the author of these benchmarks himself states
Based on an extrapolation of current P4 results, the 2GHz G5 would lag newly announced 3.2GHz P4 systems in Jet3D scalar floating point performance by about 20%, but this kind of comparison is best deferred until G5-aware compiler tools become available (since a 20% performance gain is well within the potential of compiler optimization).
Finally, it is important to note that the current test does not factor machine cost or intended use into the picture, and that can have a large impact, especially in clustering applications.
I.e. a G5 is probably comparable to a top-end desktop 32bit PC for joe blogs applications once the small changes from G4-to-G5 specific optimizations take effect. A comparison of suitably SIMD-aware number crunching applications is still needed to evaluate absolute speed. Processor power per unit dollar cost is another issue that the author hints at, which is probably not in Apple's favor.
Some of the articles (AMDzone?) is saw discussing the G5 were of the opinion that the SIMD (Single Instruction/Multiple Data) designs on intel and amd chips was superior to Apple's.
The apples to apples (no pun intended) comparisons made to old P4s (not the vector plots) are not all that impressive, as other posts have pointed out. If anything, Apples latest and greatest is slower than an old P4.
What we need is a comparison using a code that is written to use both Altivec (on G4/5s) and sse/sse2/3dnow! (on pentiums/athlons) enhancements. Anyone got something like that?
Comparison of SPEC CPU2000 benchmarks against suitably optimized(*) Apple G5 demonstrates mATX computers are infinitely faster than Apples latest offering at both integer and floating point computations.
(*) G5 test optimized by switching it off. "Its faster that way", claim benchmarking company.
grid_zones_completed_per_sec(machine2) = ( SPEC_FP(machine2)/SPEC_FP(machine1) ) * grid_zones_completed_per_sec(machine1)
to within a few percent, when you optimize appropriately (and I'm more than happy to use gcc). Its a fairer benchmark than most, and if you look what the FP tests are comprised of you'll see they're real scientific number crunching of a variety of types. If I want to buy something new and fast to work with, then a detailed inspection of the SPEC results **that are on the official spec web site** (including the motherboard used, performance on specific aspects of the SPEC tests closer to my type of computing etc) is the way to go.
IMHO, unless you're trapped with only being able to run apps on one OS you should always base your purchase (if you're after number crunching power) on the highest SPEC FP_base/cpu_dollar_cost ratio you can find. Right now I estimate thats best for a 2.8 Ghz P4 (faster than that you pay too much extra for the top of the range).
As for this G5 stuff, the way I feel is...1 296
1) Remember the G4 all the hype? Its not the first time Apple has stretched the truth. People are right to be skeptical of *all* extraordinary claims, be it by Intel, AMD or Apple.
2) Forget the minute details -- the fundamental point is that the SPEC results apple quoted for P4s and Xeons are not in line with the tons of existing spec measurements with those systems. Its deliberate spin - yup, others do it to, but cheating is cheating.
3) Apple did use a version of gcc optimized for the G5 - you'd be naive to believe they didn't experiment until they found the compiler that gave the best results.
4) I don't know about mathematica, but I'm sure you will agree that Apple has a history of comparing apple-optimized photoshop against non-pentium-optomized photoshop.
5) AMDZONE has an insightful analysis of the whole G5 PR-thing, not just the SPEC values. See
http://www.amdzone.com/articleview.cfm?articleid=
Buy whats best for you, not the hype...
Lots of astrophysicists use Macs notebooks instead of Windows, primarily because they haven't cottoned onto the fact that can do presentations (the main reason astronomers now get notebooks) with OpenOffice, or they're irrationally afraid of installing and administering Linux. OOo works better than MS office in my experience.
Macs are nice - better than MS Windows (although not for games), but for the cost conscious heavy duty number cruncher you're better off with x86 or opteron under Linux. Besides - not all astronomical software has been ported yet!