The specs for the G5 machines that were accidentally posted at the Apple store a few days ago were correct. Steve Jobs just announced G5 Powermacs at the WWDC conference. He’s calling it the “world’s fastest personal computer.” They just finished doing a demo in which a Dual 2 GHz G5 vastly outperforms a High-end Dual Xeon. Read more for preliminary specs. Prices will be $2000-3000. Oh, and the Panther OSX update was announced, but we already knew about that.Update: Official specs are up at the Apple Store. Also see Apple’s G5 Benchmark Page.
Note: this is a fast and dirty transcription from the conference floor:
Single/Dual processors. Up to 8GB Memory. Can talk to memory @ 6.4GB/s. 4x Superdrive in each. Geforce fx5200 in lower, radeon 9600 pro in higher mode.
System – G5 System Controller – dedicated bandwidth to each subsystem. IBM is fabricating it. 6x faster than G4, 8GB bandwidth, 1processor doesn’t slow down the other one. 400mhz DDR memory, AGP 8x Pro, 133mhz PCI-X slots. Hypertransport. Serial ATA. 1.5GB/s bandwidth. independent interfaces to each drive. Rest of the i/o – high performance. optical digital i/o and analog i/o, fw800 ,usb 2.0
Massively parallel. Up to 250 inflight instructions. — can be processed at the same time. The G4 can do 16. Floating point “monster”. Two fully symetric integer units. massive branch prediction logic.
Fastest front sidebus – ever. designed for dual processor systems.
Blurry photo of case (digital pic taken of big screen at keynote speech)
“yup it is effectively gone. Market share gap will be slowly closed next (closed not ended). G4 mini tower for $1299 will do that.
PC days of glory are closing fast. but you know what they say. nothing ever lasts. sorry pc fans”
ROFLMAO !!! What kind of dope have you been smoking ? For that same price I can get a PC that is waaaaay better then a freaking 1.25GHz Mac and which actually comes with a freaking monitor ! You know whats really funny though is how the hi-end G4 still cost more then a hi-end G5 ! $3,274.00 vs $2,999.00 !
“he GLIB shipped with RedHat is only a 80386 library;”
-*nope, RedHat ships with a i386 glibc and an i686 glibc. The i386 glibc is compiled with CFLAGS=”-O2 -mcpu=i386 -march=i686″, while the i686 glibc is compiled with -mcpu=i686. So anyway you put it, the glibc is i686 tuned, just like the rest of the disitribution. If the installer detects an i686 CPU it will of course install the i686 glibc.
With that said, I conclude that i686 optimized is enough for any Intel CPU.
-*by the way, I’ve compile on Slakcware Linux MPlayer with full optimizations for my athlon 2200+, then on my laptop for my Celeron @ 2Ghz (wich is acdtually an older P4 renamed Celeron), and then with -march=i386 -mcpu=i686. Couldn’t notice any diference. The x86 assembler is almost the same, and no multimedia app can take full advantage of the x86 multimedia exetinsions like 3DNow! or MMX+, or MMX2 or SSE or SSE2 and so on, because of the way they are designed. The best way to take advantage of them is to implement them into a math library and make usable functions out of them, otherwise they are pretty useless.
Just my 2 cents.
“My system is lightning fast for games, multimedia, and anything else I throw at it. This is what the Mac is competing with”
Apple is competing against $650 dollar build your own computers? What are you smoking because it’s better than the crack Steve Jobs is on.
Sorry. The average Mac user isn’t a computer hobbiest who likes to tinker with their toys. The average Mac user wants a computer that just works and they’re willing to pay for it. Let me ask you this? Do you build your own car? Do you make your own repairs? Do you build your own TV? Do you build your own toaster? Why is it computer geeks expect everybody to build their own computer?
From everything everyone has posted it certainly appears that Apple is pulling numbers for Dell systems out of their ass.
I can’t believe they are doing this again. Haven’t they learned their lesson with this in the past?
I don’t think I’ll be linking to Apple’s web site for performance comparisons. For now let’s all settle down, relax for a few days, and wait for some 3rd party application benchmarks to roll in…
In my opinion, the use of the GCC compiler on RedHat for the Pentium 4 completely invalidates their results.
Now, GCC is a fine compiler. It may even be equivalently optimized for both x86 and PPC. In anything but games, most people won’t notice a 20-30% change in performance on their computers, so GCC is definitely good enough. GCC may not even be the problem here (they are also using a different operating system and I don’t know if the system specs are the same). If they were going for a truly apples-to-apples comparison, why didn’t they use Yellow Dog Linux on their G5?
The point that Apple is trying to make here is that their machines are faster than the machines you buy from Dell. The VAST majority of machines coming from Dell run Windows XP, NOT RedHat 9.0, and most applications on XP are NOT compiled with GCC. These benchmarks are NOT representative of a high end Wintel box that you buy from Dell.
The TRUTH is that the scores posted on http://www.spec.com for a Dell 650 3.0GHz beat the scores posted by Apple for their G5 (tie on fp rate base). The ONLY reason they used RedHat 9.0 with GCC rather than Windows XP with Intel or MSVC compiler is that it made the Pentium 4 look bad. Vincent, please come up with another reason.
The G5 (970) is a great processor and certainly makes Apple competitive again. This misleading benchmarking is completely unnecessary! Jobs has, once again, proven himself completely untrustworthy.
“You know whats really funny though is how the hi-end G4 still cost more then a hi-end G5 ! $3,274.00 vs $2,999.00 ! ”
This means that the price of the high end machine went down, whats wrong or funny with this?
At these prices, I hope a monitor is included.
Compare the prices in the UK and USA…
The top model –
2999 dollars in USA
2299 pounds in UK (3,833.52 USD)
…
I feel your pain. I have started with Apple products sice the original Apple II+. Is the price abuse Apple users have to endure really worth it? For example, I just purchased the following components for only $355 + tax:
– P4 2.4G (B version with 533Mhz System Bus)
– Intel Motherboard with ethernet built in
– 256MB of 266Mhz RAM
The nice thing about x86 systems is that I didn’t have to buy a whole new machine. All I had to do was toss out the old 1Gz CPU+MB+RAM and install the new parts with just a simple pain old philips head screw driver.
The cost of a monitor is about $699 EXTRA !! So add $699 dollars and take a real look at how competitive this so called prices are !
Spec ‘tweaking’ aside, two things would make me pause before buying a G5:
1) With Apple each time a new mod is released you have to buy an entirely new system. Ie. when the 3Ghz G5 comes out you won’t simple be able to upgrade. That sux bigtime.
2) Canadians (and apparently Europeans and Australians) are getting royally f*cked by Apple on the exhange rates. Again, as most of the components are made in Taiwan this is totally unacceptible.
Overall though Steve managed to pull off a great show…no one else comes close.
Next we’re going to see dumb PC aftermarket companies making window kits for the side of this thing…
I think the best looking part of the system is inside…the case sucks, but looking inside is sweet sweet.
But if I see someone with a blue neon light and a window kit I am going to cry.
@Europeans & others
In the US, tax is not included in the price, whereas in Europe (& perhaps Australia), the VAT is included in the price. Apple charges us Americans the local state’s sales tax during the ordering process.
Quote: By Behrang Saeedzadeh — It seems that Apple is not going to support Panther with a Journaled File System… End Quote
You are mistaken about this. I have a journaled file system now with OS X 10.2. I use journalling for all of my disks. I can’t imagine that jfs won’t continue into Panther.
Regards,
Mark Wilson
The G5 (970) is a great processor and certainly makes Apple competitive again. This misleading benchmarking is completely unnecessary! Jobs has, once again, proven himself completely untrustworthy.
Agreed. Apple should be pushing application benchmarks anyway, which make *much* more sense to end-users, rather than making up their own SPEC scores.
I think everyone here can agree that Apple lied on the benchmarks again. I contend the best solution for this discussion is to simply drop the issue for the time being and wait for some 3rd-party application benchmarks…
I have to agree with both of you guys. We need real world benchmarks that do not bend the truth just a little like the orginal poster said Steve did. Let’s compare a new G5 wiith a x86 Gentoo based system or XP based system and let the real benchmarking begin ! As always though this is another fine example of Steve Job lying once again and blowing smoke up the Apple cult follewers arse ! As they say “Theres a sucker born everyday !”
Jack, maybe in Europe, but in Canada tax is NOT included. I have to pay 7% GST on top of things, and if I’m one of the unfortuneate souls who live outside of Alberta I’d have to pay provincial tax (10 – 15%) on top of that.
The live head-to-head demos of Photoshop, Logic, Mathematica and Lexology must have been staged by the guys who faked the moon landing.
And before anyone says it: yeah, of course those programs were chosen because they’ll show off the PPC970’s strengths. This doesn’t mean that those aren’t programs that are used by professionals in the real world on a daily basis.
But, yeah, I’m sure a P4 with Nvidia’s newest card of the week will spank the G5 in that all-important Quake frame rate test. And after all, isn’t that what’s really important?
$2999.00 US for not monitor, no mention of a Powerbook or XServe … come on Apple
I need a laptop and you haven’t givem me anything and even if I wanted to buy a desktop … I have to wait until AUGUST !!!
It seems that Panther won’t you the old HFS+ filesystem.
Searching files on the entire disk is very fast, it even display the results
as you type juste like in iTunes.
This is possible thanks to Fast UFS, the new filesystem for OSX. This filesystem should be is solid (based on UFS) and have some BFs features.
Thanks Steve, you’ve fixed most of the Jaguar issues (filesystem, finder, networking, …)
Jack, maybe in Europe, but in Canada tax is NOT included.
You’ll notice I didn’t mention Canada: I specified Europe & “perhaps” Australia. 😛
Mark, what do you meean that you have jfs on your disks? OS X 10.2.x does not come with jfs, only server versions. I bought iMac 17″ recently and I am amazed except that I don’t have jfs.
Mark, what do you meean that you have jfs on your disks? OS X 10.2.x does not come with jfs, only server versions. I bought iMac 17″ recently and I am amazed except that I don’t have jfs.
Journaling support is included in the standard version but must be enabled from the command line.
[ITALIC] But, yeah, I’m sure a P4 with Nvidia’s newest card of the week will spank the G5 in that all-important Quake frame rate test. And after all, isn’t that what’s really important? [/ITALIC]
Um…….no
What I think Apple is trying to do with release is go towards the business side not consumer side.
“I think everyone here can agree that Apple lied on the benchmarks again.”
Wait, you understand the specs of the machine tested, you know the OS tested, and the compiler. Are you not accepting the validity and impartiality of VeriTest?
Because as far as I can tell, these aren’t made up. You aren’t satisfied with the choices Apple made, but that’s your problem.
People keep saying it should be Wintel, but I was down at the keynote this morning: it is apparent Apple is making a push at the UNIX, scientific market. (It makes sense to use RH and GCC then.)
People who are asking for Athlon64 or Opteron benchmarks are on crack… Why should Apple say: “we’re slightly slower than computers that won’t be out for a month or two and which we aren’t competing against.”
So let’s repeat: we’ve got valid benches for SPECint and fp for the new Apple machines that slightly beat equivalent Dell hardware running Red Hat with GCC compiled apps. Doesn’t that give you a sense of how fast they are? Even if they are slightly slower than if it was Windows using ICC. (Why should Apple give Microsoft any press after dropping IE.)
We got to see what these machines running Photoshop, Mathematica, Logic, et al. will do. That is: kick ass. Adobe says so, Wolfram says so. Two benchmarks for a statistic that doesn’t even come close to belying the performance of any complete system at all task for all software doesn’t change that. So please, this “make sh!t up” comment is ridiculous. They are valid test results. You know the parameters for these tests. Educate yourself rather than sound like a whiner.
So please, this “make sh!t up” comment is ridiculous. They are valid test results. You know the parameters for these tests. Educate yourself rather than sound like a whiner.
Well “Anonymous”, I have only been posting for *months* about how PPC970 will outperform x86 offerings at the time of its launch. I’ve also been saying for months that PPC970 systems will launch at WWDC and have been giving details of the architecture.
Anyone who frequents OSnews will know me as a (somewhat rabid) supporter of Apple.
And you’re telling me to educate myself? That’s quote a comment for an anonymous coward
Given all of this, I’ll readily admit that Apple lied on these benchmarks. The purpose of SPEC is for vendors to submit scores and compare them. By not using vendor-supplied scores for x86 systems, Apple is giving information that is not correct, hence a “lie”
I’m trying for a degree of impartiality here, as many accuse me of “blindly following Apple”
I suggest you do the same…
I think I’ll stick with my Mac SE/30………wonder if it will run Linux?
Now that Apple is using IBM’s chip, what are the possibilities that IBM will license OSX for the P Series of Workstations and Servers?
But where is the lie?
Where is the made up part?
You know exactly the parameters of what Apple used in the test. Does this give you something to compare to?
So what’s made up? Where is the lie?
What about their benchmarking can’t you comprehend?
“The purpose of SPEC is for vendors to submit scores and compare them.”
So can you look at Apple’s valid test results and compare them to the ones at Spec? For them to have lied, they must have faked their own results for you to not make a valid comparison.
But where is the lie?
Please go on http://www.spec.org and find the numbers posted for the x86 systems.
The only numbers for the given systems do not match the ones Apple has posted. This is intentionally misleading.
If you care to continue arguing the semantics of the word “lie” go right ahead, but the point remains the numbers given on http://www.apple.com/powermac/performance/ are intentionally misleading.
Apple has been guilty before of just short of outright lying over it’s computers’ capabilities before: anyone else remember the hype over the G4 being a “legitimate” supercomputer?
As I recall that ad campaign came out of the fact that it was the first Gigaflops machine and that was the level the government set for export controls for “supercomputers”. We don’t want those machines getting in the hands of the enemy!!!
1. probably already answered but….check out the apple store. Yay. A fast full system for $3k
2. no idea
3. (the real reason I responded to your note)
I like to think of it more as: What cool things will I be able to do with OS X down the line because of the speed of these systems?
There is a huge difference. Your views on this issue are out of sync with Apple’s. Your view is: I wish Apple used vendor-posted data and said they were in third or fourth place. Whatever. They built these tests because they are wooing the UNIX market right now. They want to compare these PMs with workstations and servers from SGI, HP, IBM, Dell. They don’t care about your intentions.
“This is intentionally misleading.”
What part were you confused about? You know the reason for the difference is the different OS and compiler, right. You know that OS X is closer in comparison to Linux don’t you? You know that the compiler Apple uses is GCC 3.3, right?
What part were you mislead by? Sounds like you actually do know how to read and do know what the reality of the situation is. So where is the misleading and the lying?
Or rather why are you so frothy over the issue? Doesn’t seem to me like you’ve been mislead or lied to.
Well, after I posted the eweek link the well dried up, eh?
I found this article in Slashdot that reads as follows:
“Posted by pudge on Monday June 23, @02:07PM
from the sir-my-mac-is-on-sir dept.
Steve Jobs announced the end of Jaguar, and the newness of Panther, today at his WWDC keynote address. Panther is to be available as a preview release now, and by the end of the year retail, for $129….” read more here:
http://apple.slashdot.org/apple/03/06/23/188254.shtml?tid=179&tid=1…
In other words, you get suckered with a machine for about $3000 US dollars and the OS in it is 32bit MacOSX!!! The 64bit MacOS is “promised” for the end of the year.
Lets hope they deliver, still is pretty shameless to put so much hype into nice hardware, and put it in the market with emulated 32bit crap. Only Apple can find so many suckers!
Now if Apple didn’t control both the OS and the Hardware it would be a different story, but they are use to claiming hardware mighty and suckering people in the OS side, this is why I no longer buy Apple. Been there, done that.
The point of SPEC is that the benchmark is the standard way of comparing processors from different architectures. What happens is that every vendor use the best system and the best compiler with the optimal compiler flags to produce the highest possible score. Then you use these scores to compare the processors to get an idea of which processor is faster. It’s not perfect but it’s a rough estimate.
The system that Apple used on their performance page was already submitted to spec but the score was too high for Apple. So Apple had a third party perform the spec test using a sub-optimal software configuration and got a lower score. For all I know they might even have created debug builds. Is this a lie or not? I don’t know, but it is certainly misleading.
Your an idiot, anyone who purchases a 970 will receive Panther. Panther is a 64 bit OS. There is no 32 bit emulation, it runs native.
Bye, bye.
You should have done the benchmarks against a dual opteron 1.8GHz PC. And what about some standard benchmarks like SPECint and SPECfp? Oh I know – there the G5 doesn’t have any chance against an dual Opteron or Xeon.
Actually it creamed the dual Xeons. Didn’t you watch the show? As for the Opteron, is the updated version of XP even out yet that can run on the Opterons?
“What happens is that every vendor use the best system and the best compiler with the optimal compiler flags to produce the highest possible score. Then you use these scores to compare the processors to get an idea of which processor is faster.”
No, what happens is the vendor submits results for every product they sell. IBM releases SPEC scores for different OSes and compilers using the same machines. Obviously, the way you describe it does not produce an objective or even approximate benchmark of processors if it is optimized for a particular OS or compiler… Come on. The purpose of the SPEC benches is to benchmark the complete system: processor, mobo, OS, ABI.
However, the way you tell it, points out what some of us know and also why the idea of using ICC and Windows is just as arbitrary as using Red Hat and GCC.
If they compared to a Intel proc with Windows compiled with ICC that wouldn’t test the performance of the 970, because who knows how Windows would run on a PPC… And vice versa… The question is: who is Apple targeting right now? You may see an consumer push… They may say some things about Windows…. But right now, they are definitely targeting UNIX/Linux vendors and app developers. So… use GCC and RH.
Look, folks. Yes, companies will choose benchmarks to make themselves look the best, and tweak as best they can within those limits. Does that mean it’s disingenuous for Apple to use GCC when the tests on SPEC’s own site have more than likely been done with ICC? You can argue that. But the problem is that with two different compiler chains, you’re testing the compilers, not just the hardware.
Realistically, people use GCC for nearly everything on PowerPC desktops (OS X and Linux), and GCC has a far greater ‘market share’ than ICC does on x86, too. And, GCC is cross-platform, which makes it as close to ideal as possible for a test like this. You can argue that at that point you’re testing the compiler’s optimization, sure–but by the same token, you can’t have it both ways: if you want to test against ICC on x86, test against CodeWarrior on the G5, or better yet, test against some compiler IBM’s specifically written to take advantage of the chip. (We’ll wait here while they write it.) As someone else pointed out, it’s dubious to claim that GCC is far more optimized for the PowerPC than it is for the x86 family, where its history is much longer (and on which companies, like Cygnus and Red Hat [and Be, for that matter], have employed engineers full-time on compiler development).
Conversely, if you’re arguing that GCC on x86 isn’t “real world” in the way it is under OS X, then don’t use ICC on x86, either–use Visual C/C++. It’s by far the most widely-used compiler on the platform.
Either way, though, asserting that the only fair comparison is between GCC on the PowerPC 970 and ICC on the Pentium 4 is silly.
Suckered?
By Anonymous (IP: —.broadviewnet.net) – Posted on 2003-06-23 23:45:39
Your an idiot, anyone who purchases a 970 will receive Panther. Panther is a 64 bit OS. There is no 32 bit emulation, it runs native.
Bye, bye.
Lets see G5 start selling in August and Panther will be available for retail at the end of the year, months after G5 are on sale. Still shameless, still Apple!
At times is hard to see who is the bigest crook Gates or Jobs.
I’ve been holding off buying any hardware for quite a while now. The G5 tips the balance for me. I’m off to buy a Mac and I can honestly say, I’ve never thought it would happen.
As long as Apple used the same gcc settings for each run and each different machine I don’t see the problem with the SPEC numbers. And by using a possibly different set of compiler settings it is very easy to see why they would get different numbers from what appear on spec.org. It’s sad that some people question the integrity of VeriTest but I guess some people are blinded by hate….
” In other words, you get suckered with a machine for about $3000 US dollars and the OS in it is 32bit MacOSX!!! The 64bit MacOS is “promised” for the end of the year. ”
Its perfectly fine that Apple ships fast hardware as soon as it is available and its actually an advantage that the 64bit hardware runs the 32bit software with no problems.
” Lets hope they deliver, still is pretty shameless to put so much hype into nice hardware, and put it in the market with emulated 32bit crap. Only Apple can find so many suckers! ”
Apples track record for delivering is pretty good, not perfect but good. There is no emulation either, do some homework. As far as suckers at least we’re well off than the average PC white box nerd.
” Now if Apple didn’t control both the OS and the Hardware it would be a different story, but they are use to claiming hardware mighty and suckering people in the OS side, this is why I no longer buy Apple. Been there, done that.”
Yeah it would be great if Apple didn’t control their hardware and software, maybe they should put you in charge. That would be perfect.
I really like the spec’s on the new G5’s, its nice to see that Apple is not dumping old technology on us poor users for top $$. The G5 looks as if its a damn good chip, but would it be faster than the new P5’s that are due soon or even the Operton/Athlon 64’s I somehow don’t think so. The new road map from AMD states that by 2004, the Athlon64’s are gonna be equiv to 4.3Ghz P4’s. Not sure on the FSB, but from everything out their – Current dual opertons just smoke Dual Xeons and itanium’s and thats running in 32bit mode.
Talking about money – to get one of those G5’s in the UK I’d be looking at paying out anything from £1500- £3000. Well, my new PC is just going to cost around £500 – the specs you may ask :
Shuttle NForce 2 ultra (400FSB);AMD Athlon 2500XP; 1GB PC2700 RAM (2*512MB); IBM deskstar 120GB (8MB buffer)and Sony DVD/CDRW; Integrated Nvidia GFX + ATi 8500 (128MB)
I’m looking for an ATI 9500 Pro (128MB)as the graphics card but I haven;t priced that in yet but i should be able to pick one up for about £160.
Harj
: o )>
PS…I hope Panther doesn’t run dog slow on the G5 like OSX10.2 does on the current G4 but it does look better than jaguar.
“At times is hard to see who is the bigest crook Gates or Jobs.”
Whose suckered more people?
Which is one of the links on this page: http://www.apple.com/powermac/performance/
It shows SPEC 2000 benchmarks vs Redhat, and Photoshop, audio and gaming benchmarks – some of which are on Win XP.
I would have liked to have seen several Win XP systems in all benchmarking categories as THAT is Apple’s competition NOT Redhat Linux.
I had a good laugh after reading those. They are exactly what you would expect, BUT
What an earth is the point of comparing a 32 bit app on a 32bit OS running on a IA-32 processor against a 64 bit app on a 64 bit OS running on a 64 bit chip?
The correct comparison is G5 v Itanium 2.
In fact G5 is going to have similar problems to Itanium getting into desktop systems, i.e. large die size, high power dissipation. To get this fantastic performance requires large memory (> 4GB)which currently drinks a LOT of power. Large power supplies and extensive cooling fans make for very expensive desktops.
Only native 64 bit code will produce these fantastic results for BOTH architectures. To be fair Itanium needs native softwhere even more due to its VLIW nature.
All in all though, G5 and Itanium 2 are still server/expensive workstation chips only at the moment.
Now the real problem will be all those 32 bit software applications that don’t immediately get 64 bit, native versions in good time. You can expect many ISV companies to sit and wait for the installed base to grow large enough to form a decent market.
Back to Performance benchmarks:
It is being ‘economical with the truth’ of Steve Jobs not to post G5 v Itanium research on Apples website, pretending that PIV is Wintel’s (and Lintel’s)only contender at that “weight class.”
Fare enough to post both PIV v G5 AND Itanium 2 v G5, but really Mr Jobs, just PIV v G5?
Pull the other one.
I’ll leave it to you all to debate these numbers…
Actually the base spec numbers only apply to single processor systems so I don’t know why you bothered to list “Dual 2.0Ghz G5”. There is a separate spec test for dual processors.
” PS…I hope Panther doesn’t run dog slow on the G5 like OSX10.2 does on the current G4 but it does look better than jaguar.”
Run MacOSX on a slow system and it will be slow, run it on a fast system and its fast. How is this so hard to understand? G4s and fast G3 run MacOSX fine.
“No, what happens is the vendor submits results for every product they sell. IBM releases SPEC scores for different OSes and compilers using the same machines. Obviously, the way you describe it does not produce an objective or even approximate benchmark of processors if it is optimized for a particular OS or compiler… Come on. The purpose of the SPEC benches is to benchmark the complete system: processor, mobo, OS, ABI.”
Yes, the vendors submit lots of results (which you can see on http://www.spec.org), but when you use these results to compare processors you take the best system. When I see something about the “SPEC score” for a P4 X.Y Ghz I assume it is the highest score obtained with that processor. Stating that they compiled with gcc 3.3 doesn’t tell us very much but I suppose the tests are documented somewhere.
The SPECs given by Apple are correct for the configuration noted in the Detailed SPEC Documentation supplied; thus, Apple didn’t lie (beceides Apple didn’t do the testing, they had a independant company do it).
Yeah, Apple didn’t lie – they had their lawyers pay VeriTest to do it for them!!
Seriously, though. This whole discussion about SPEC numbers is why all anyone should care about are application benchmarks.
Of course then the whining will start about how Adobe is working closely with Apple to make their systems look better……*brace for it*
“Wait, you understand the specs of the machine tested, you know the OS tested, and the compiler.”
The only reason I understand this is because I looked at VeriTest’s report. The vast majority of people out there will only look at the bar graphs on Apple’s website. Funny, I didn’t see RedHat or GCC mentioned on Apple’s sight…
“Are you not accepting the validity and impartiality of VeriTest?”
No, I’m not. VeriTest was “commisioned” (aka hired) by Apple. I trust that VeriTest did not lie in their report. However, this is NOT and independent analysis. From the report: “At Apple’s request, we configured the Dell Dimension 8300 and Dell Precision 650 using RedHat Linux Professions, GCC version 3.3…” Apple was obviously in control of some parts of the test, hence it was NOT impartial.
“Even if they are slightly slower than if it was Windows using ICC.”
If the numbers were slightly lower, I wouldn’t be upset, but they are dramatically lower.
836 vs. 1089 (23% lower)
16.7 vs. 21.7 (23% lower)
646 vs. 1053 (39% lower)
11.1 vs. 15.7 (29% lower)
If the proper values from the spec website were used, the G5 would have appeared slower than the Dell 650.
“But where is the lie?”
There isn’t a “lie”. The facts Apple put on their website are correct. The problem is that what Apple put on their website is also intentionally misleading.
“it is apparent Apple is making a push at the UNIX, scientific market. (It makes sense to use RH and GCC then.)”
Maybe, but this is a major stretch. Apple’s whole switch campaign is targeting Windows users.
Bascule, thanks. I believe in your impartiality. While I may sound like a rabit anti-Apple person, I’m not trying to be. I just get pissed when I see intentionally deceptive advertising. Jobs has a history of making false claims based on factually correct benchmarks (when you read the fine print).
its completely backwards compatable with 32 bit apps. From the apple site:
The PowerPC architecture was designed from the get-go for both 32- and 64-bit processing. That means the Power Mac G5 can run everything you run today, with no performance penalty. Contrast that with the competition, where switching to 64-bit computing will mean costly expenditures for 64-bit software or running a 32-bit operating system in a slow emulation mode. That’s right, current 32-bit Mac OS programs — including Mac OS X itself, the Classic environment and existing applications — run natively at processor speed, with no interruptions to your workflow and no additional investment in software. In fact, Mac OS X applications that take advantage of the PowerPC G4 Velocity Engine will immediately run faster on the G5. Of course, Mac OS X applications tuned for the G5 will run even faster.
What I would like to see is how the 1.6 GHz model stacks up against the current dual processor Macs. I know there are no real 64 bit apps and Panther isn’t out anyway, but just hardware wise, that would give me a better handle on this than Mac vs. PC benchmarks.
For everyone arguing about which compiler used I have this thought:
If the choice of compiler for a system can alter SPEC numbers by over 30% on similar systems (1164 vs 889) then doesn’t that totally invalidate the whole notion of SPEC benchmarks? It seems like people are saying that the compiler used is much more important than the actual hardware.
Yeah, nice try, idiot! They DID run SPECint and SPECfp ON STAGE today at the keynote, and yes the G5 somehow beat your precious dual Xeon.
If the choice of compiler for a system can alter SPEC numbers by over 30% on similar systems (1164 vs 889) then doesn’t that totally invalidate the whole notion of SPEC benchmarks?
It certainly does if you generate your own SPEC numbers for a competator’s platform using an inferior compiler, which is precisely what Apple did.
A compiler is a necessary evil for a cross-platform benchmark. However, keep in mind the quality of compilers available for a given platform is important as well.
This is why vendors should not generate SPEC numbers for their competators. Fairness in the benchmark is achieved by letting each vendor optimize their systems and compilers to achieve the best numbers possible.
I’m writing this on my new $500 2.088 Ghz AMD ShuttleX PC.
I’m amazed at how much computers have changed over the years. I grew up on a 1 Mhz Commodore C-128. That computer lasted me almost ten years. Then I finally migrated to the PC at 100 Mhz 486 DX4. But that was my mom’s system, my first computer was a 486 DX2 66 Mhz box that ran Linux with a trident video card that wasn’t supported under X.
Anyway, to make a long story short, I migrated to a Cyrix @ 120 Mhz and got a nice 4+ MB video card and Monster 3D and spaceorb and quake, and then to the Cyrix M2, P2 400 Mhz with G400 TV card and the AMD K6-450 and Athlon Slot A 500 and Thunderbird 1000 Mhz. And then came the socket A 1133 and 1200, Duron 800, Duron 1.2, Dual MP-2000+, XP 1700+ shuttle, XP 2400+ RAID server replacement, and now the XP 2600+ shuttle desktop.
And now the PPC 970 and soon Athlon64 are being released. Have I died and gone to CPU heaven? Computers cost me less than ever before and they’re better than I could have ever dreamed, possibly even better than the Alpha.
“when you use these results to compare processors you take the best system.”
No, that’s what you do. I take the one which is for the system and compiler I would use. Which is why there is nothing deceptive about these benchmarks. All benchmarks require reading the fine print.
“If the choice of compiler for a system can alter SPEC numbers by over 30% on similar systems (1164 vs 889) then doesn’t that totally invalidate the whole notion of SPEC benchmarks?”
Yes, if you think SPEC rates hardware only. THose who use it appropriately know it’s testing a complete system. But, yes, in general, I agree with you that they aren’t very valid anyway.
“The vast majority of people out there will only look at the bar graphs on Apple’s website.”
Actually looks like the vast majority never understood SPEC in the first place and are claiming that Apple is lying or deceiving them.
“If the proper values from the spec website were used, the G5 would have appeared slower than the Dell 650.”
Proper? What the hell is proper? If you just want a hardware benchmark, Apple could have run YDL or Darwin and maybe then the G5 would be fastest. As is clear from where this discussion is going Apple doesn’t have to bench every possible config nor should they be expected to. Your insistence that the way you think and what you want to be proper is improper.
“Maybe, but this is a major stretch. Apple’s whole switch campaign is targeting Windows users.”
And ‘Switch’ has been deprecated on the site, and I haven’t seen a commercial in 3 months. But I know Apple is wooing the scientific market, the CAD market, and this is a developers show. And you’re ignoring the fact that RH/GCC is closer to X/GCC than Win/ICC. It’s an appropriate and targeted comparison.
“The correct comparison is G5 v Itanium 2.”
That’s foolish; the Itanium is only a server product that needs to have apps developed for it. Apple has the advantage of native 32 bit support, a chip that costs the same as a consumer desktop chip, and a few key apps already available in 64 bit mode and the rest of the apps will run. Apple is targeting this machine at the high-, mid-, and low-range… Itanium isn’t going to do that. Itanium is going to be more expensive and useless to the consumer for years and years to come.
“All in all though, G5 and Itanium 2 are still server/expensive workstation chips only at the moment.”
Show me a Itanic box I can put on my desk for two grand.
whats the deal? Upgrade to G5?
Dual 1Ghz FSBs, PCI-X, Serial ATA with dual independant channels, AGP 8X Pro(for more juice), FireWire 800 and 400 and USB 2, Hypertransport interconnects, 8GB Dual Channel DDR 400 ram, digital optical audio in/out, internal bluetooth and 802.11g (BTO).
If there is another motherboard on this planet that can claim all of this, let me know.
plus the 13 automated fans with different temperature zones- soooo cool (and still quiet)
“where switching to 64-bit computing will mean costly expenditures for 64-bit software or running a 32-bit operating system in a slow emulation mode”
This is true of the Itanium, but not really true of the Opteron. The Opteron doesn’t have an “emulation” mode, but it does have a compatibility mode (still native). The only penalty with the Opteron’s 32-bit mode is that it won’t take advantage of the enhancements made in x86-64 (added registers, improved floating-point instructions).
The whole 64-bit thing, while great, is really being overstated. The 64-bit mode for the 970 will provide absolutely no benefit unless you need more than 2GB of memory (there may be 32-bit issues with 4GB, not sure). This could be important to some professional users (3D etc.), but means very little to the home user. On the Opteron there is a benefit to running in 64-bit mode because AMD added some registers and cleaned up some x86 ugliness. The PowerPC already had enough registers and none of the x86 ugliness, so there wasn’t really anything to improve.
I think we should all be moving to 64-bit. I hate that Intel isn’t moving the Pentium 4 to 64-bit. In 2-3 years, I think home users will probably hit the 32-bit wall.
It certainly does if you generate your own SPEC numbers for a competator’s platform using an inferior compiler, which is precisely what Apple did.
And if “good” compilers are that different then SPEC numbers are totally irrelevant.
Fairness in the benchmark is achieved by letting each vendor optimize their systems and compilers to achieve the best numbers possible.
Is that fair? Sure, I guess so. Is it meaningful in any way? No. Frankly, it just encourages the kind of cheating we saw from NVIDIA recently – optimizing the compiler to help out the benchmark in ways that won’t help out real world apps.
Congratulations on beginning the 64 BIt Era Apple! http://www.64bit.org/index.html#64BitEraBegins
haha, this is so funny. If Apple had gotten higher scores used a ppc ultra-optimized compiler and the Intel compiler for the benchmarks, then the same people would whine that the comparison is not correct, that it wasnt fair, etc etc. Either way, they will always say something negative.
There can never be a truly perfect benchmark to compare different systems. You just HAVE to look at the fine print. Apple is just marketing their product, just like Intel, AMD, Sun, IBM, Sgi do, ALL THE TIME claiming they have the most powerful ultra fastest stuff based on xxx benchmark.
If the most valid SPEC numbers are derived from the using the best compiler available, it might be interesting to find the SPEC numbers derived by using Code Warrior instead of GCC. The fact that Code Warrior is about 40% faster compiling than GCC in MacOS X may be testament to it’s maturity and optimization vs current GCC(or maybe not).
Sigh,
1. Alpha Linux is 64 BIT all the way through. If there was some 32 bit os in there I might be able to get em86 to work agin.
2. First 64 bit desktop, Alpha
3. First 64 bit laptop, Alpha (tadpole technologies)
4. Why in hell did Apple Computer Inc. use ide drives? These are workstations and should get SCSI. (SCSI is a 5 year warnity for running 24x7x365, IDE is 2 years at 8x5x365)
5. And not to make a fully tricked out box sigh…
6. Does the MAC have ECC with scrubbing?
7. Now the delima, My new system a dual Apple or a quad Opteron??? Close in price Opteron has better game support and linux, Apple has better native UI (OSX), and Darwin is a good OS. sigh… Hmm I wonder if BeOS will run on a quad opteron…
Summary,
Nice move, Good industrial design (for the second product sweep, I wonder who was responsible for that shift?) Still some rough edges and not server ready, 75 percent rating.
Leslie Donaldson
“Dual 1Ghz FSBs, PCI-X, Serial ATA with dual independant channels, AGP 8X Pro(for more juice), FireWire 800 and 400 and USB 2, Hypertransport interconnects, 8GB Dual Channel DDR 400 ram, digital optical audio in/out, internal bluetooth and 802.11g (BTO).”
Except for Firewire 800 all the above are already readily available in the x86 world. DDRII (533 MHz), 1066 FSB and PCI Express will be available in X86 negating any current Apple advantages. Expect cheaper/faster Apple killing AMD64 boxes by Xmas.
I just wish they could make them affordable, the prices are just way too high. Anyone know if they’re going to offer any real price drops on the G4’s now?
1. Highest recorded 2P and 4P SPECweb® 99 scores, breaking for the first time the 10,000 barrier on a 4P system
2. Highest recorded 2P and 4P SPECint®_rate2000 scores
3. Best scaling 1P to 2P to 4P SPECfp®_rate2000 scores
4. Highest recorded 4P SPECweb 99_SSL scores
5. Highest recorded 1P, 2P, and 4P SPECjbb® 2000 scores
http://www.theinquirer.net/?article=10086
“”Fairness in the benchmark is achieved by letting each vendor optimize their systems and compilers to achieve the best numbers possible.”
Is that fair? Sure, I guess so. Is it meaningful in any way? No. Frankly, it just encourages the kind of cheating we saw from NVIDIA recently – optimizing the compiler to help out the benchmark in ways that won’t help out real world apps.”
You are correct that it does encourage the kind of cheating we saw from NVIDIA recently, but to say that it isn’t meaningful in any way is wrong as well. It shows the maximum performance that can be obtained from a certain processor running certain test programs. It doesn’t say much, but it’s an accepted way of compairing systems. Apple obviously thought it was meaningful enough to use it in their marketing.
People buy Macs for the experience, not the garage laboratory PC-you-build-yourself prices.
I have 2 PCs and 2 Macs. I use the PCs for BeOS and Windows 2k. I use my Macs for everything I can without having to go back to Windows. Why? Because it works, and it’s not so intrusive.
Money is no object when I want something that works. Again, maybe that’s why I don’t drive a Yugo either! :^)
Congradulations Apple on the new line of hardware. I enjoyed your real-world tests of Adobe Photoshop, and I will be purchasing 4 new Dual G5-2GHz workstations for work next week.
i would have thought this news would generate more comments…
I take it there is news from Apple today? Hmmm, didn’t interrupt those wonderful Lycoris “TV” Commercials..
f u c k all that shit, fuck all the threads going back and forth, I haven’t even read any of it because it doesn’t matter, i’m just going to buy an apple g5 dual 2ghz.
The benchmarks were using off the shelf software NOT specially tuned version of Photoshop. Photoshop is a 32bit application, and if it can run faster on a PowerPC clocked at a lower speed than its Windows counterpart, what is wrong with that? all that demonstrates is that the CPU is VERY efficient per-clock cycle.
In terms of 64bit-ish-ness, something Intel doesn’t understand, everything doesn’t need to be 64bit. The problem with Intanium is that you are forced to make your software 64bit.
As for the future of Mac, I am looking VERY forward to the future of the Mac platform. Hopefully they can get their act together in Australia and really start pushing their products in terms of marketing.
Thank goodness, I finally found another person like me I wants quality. I’ve got a Dell XPS T550, yes, that was before the Intel world became buggy and crap, and the 8xx chipsets were released on the public. The MB I have has a BX440II chipset, rock solid, Matrox G550, again, rock solid. If I priced up all the additions and changes I could have most likely bought a Mac back when I got my PC.
Fast forward 3 years, and now I am really focused on buying a Mac. I want a UNIX based operating system with the applications I want. MacOS delivers it. I can run Office X, 4D Relational Database etc etc. I’ve just signed up for the early notification regarding the availability of the PowerMac G5, hopefully it will arrive shortly in Australia. I assume that Apple Australia want to build up some local inventory so that demand can be satisfied 😉
Even though the new Macs beat the fastest available PC’s in the SPEC tests the SPEC tests don’t take into account Altevec in the 970’s. Hopefully this might give Intel a kick to start producing processors that produce more results than just a high gigahertz rating. It will be interesting to see if the PC world can come up with anything at a similar price point to the new Apples by the time they come out in August.
I think you should try OS X on a PPC970-based system (namely one with a 64-bit kernel and userland compiled with the 64-bit PPC ABI) before making that judgement.
I very well can’t at this point, now can I?
Yes yes, I’ve heard it all before… wait for Panther, and try it on a new system. Apple has just released new hardware and will soon be releasing new software to fix some complaints…
True, but Intel and AMD won’t be sitting still either. The holy grail of a fast Mac always seems right around the corner.
Well, troll all you want, OS X on a 1GHz G4 does not have responsiveness issues in my opinion…
It does in my opinion, so I could just as easily regard your position as a troll… I have a dual 1.25 GHz G4, and I notice a considerable difference in responsiveness compared to my single processor 1.4 GHz P-4 running WinXP.
Some of the posters are from those who want a new G5 BD! I don’t blame them. So do I. I am going to have to save up my pennies for a while.
However, we are hearing the expected from the FUD merchants, Trolls, AMD Wimps (Sorry, the Opteron is nothing more than a hopped up K7 using kludges to emulate a true 64 bit processor…) and the usual claque of Wintel apologists… already dismissing test results that they find inconvenient.
Fact:
The tests *WERE* done by an indepentent lab, *NOT* by Apple. Too bad the results did not support your wishful fantasies. Oh yes, and this is with SPEC2000 benchmarks which are notoriously x86 biased.
Fact
Adobe ran their own PhotoShop actions tests and the results were spectacular for the G5. Sorry that you had to ignore those real world tests because they didn’t prove your favorite 40 year old obsolete x86 technology to be up to the task.
Fact:
A dual 2 GHz G5 is more than a thousand US dollars *CHEAPER* than the nearest equivalent Dell crap box which got stomped int the cross-platform tests….
Ah yes… the Trolls are going to have to work overtime to dismiss the inconvenient facts this time.
I wonder how fired teh guy that accetally put the graphic up for the G5 is?
This whole discussion about SPEC numbers is why all anyone should care about are application benchmarks.
Ah, couldn’t have said it better myself. Who the f**k cares about Spec, really? What I want to know is: how will Application X and Application Y run on the two systems? How great is the multitasking and networking? How about the Quake frame test? The list goes on. Real-world performance is what I actually experience in my day-to-day operations; it’s not likely that I’ll ever think to myself “this scored an 86? It seems like a measly 79!”
A similarly configured G5 is only $2000 more than the PC box I built only two months ago. Damn my bad luck!
By the way, how do you keep your Mac from electrocuting you in your underwater pineapple? 😉
Steve Jobs is a marketing god, he is not much of an engineer! He is using the engineering tools for marketing purposes (which is a norm for the tech industry nowadays! Intel does it, Sun does it, IBM does it, AMD does it, you get the point) … which we engineers loath; and that’s why we have trade journals! The keynote is a marketing event (no matter what you say, if it was webcast everywhere, it was all about marketing!) So, he was talking to the market, not the engineers or the IT personel; to him, it was OK to employ “standard’ marketing stratigies … I think that’s why the SPEC results. Doesn’t make it right, but there is no point arguing about it, I think! If you are in the IT industry, you should have gotten use to such BS, we have been numbed about these for years … and most accountants, or so called IT managers never make decisions base on these numbers anyway, there are too much other business and political factors! So, I guess these market schemems are all about the image of the companies, you are not supposed to look at the numbers too closely, the important message was ‘posed to be “Mac has the fastest desktop this week” …
One interesting observation, there was no MS representative on stage today … hummm, wonder why!
G5… Food for maclots trolls (evidently…)
“… (Sorry, the Opteron is nothing more than a hopped up K7 using kludges to emulate a true 64 bit processor…)”
Shows how little you know.
“The tests *WERE* done by an indepentent lab, *NOT* by Apple.”
When Microsoft pays an “independent” company to prove that their products are the best do you trust them? I don’t, I verify.
“Too bad the results did not support your wishful fantasies.”
Too bad they don’t show the truth.
“Oh yes, and this is with SPEC2000 benchmarks which are notoriously x86 biased.”
More proof that you are truely clueless. One word: Alpha
“Adobe ran their own PhotoShop actions tests and the results were spectacular for the G5.”
Yeah… But with Apples/Jobs traditional skewing of truth how can this be trusted? I remember that a dual 1,25GHz G4 is faster than a P4 3GHz according to Apple…
“Sorry that you had to ignore those real world tests because they didn’t prove your favorite 40 year old obsolete x86 technology to be up to the task.”
40 years? You really should learn to count (and try to learn what obsolete means at the same time).
“A dual 2 GHz G5 is more than a thousand US dollars *CHEAPER* than the nearest equivalent Dell crap box which got stomped int the cross-platform tests….”
It is truely amazing how difficult it is for macloths to use the Dell website…
Actually, if you read the whitepaper, they used the benchmarks which uses Red Hat Linux 9.0, not Windows XP, for the SPEC. That’s why the numbers are different. Not fabricated.
macedition.com/partingshot/partingshot_20010413.pdf
Fact: I don’t buy a computer to run SPEC2000; I buy one to get real work done.
The fact remains: OS X is not as snappy as WinXP, even when WinXP is running on much slower hardware. If I had to guess why, I’d say it’s all the layers between the user and the processor: Aqua, Carbon, Cocoa, BSD, Mach, etc…
I’d love to have a dual 2 GHz G5 (and I probably will get one), but the fact remains that OS X on the G5 will probably seem less responsive to me than WinXP does on my middle-of-the-line P4.
One interesting observation, there was no MS representative on stage today … hummm, wonder why!
Wake up and smell the camphor, knucklehead. Does MS invite Apple reps to their parties? I think not. Now go watch the rest of your Flintstones episode…you might miss something important to you.
Hey Gates suckers! Apple users can LEARN from you how to suck. First lesson: they have to BUY a windows ‘monster’…
The Mac is becoming a UNIX workstation, and I know plenty of people stuck on Suns that are extremely happy about this.
All those benchmarks were for other unix developers, they want a real platform that offers performance, UNIX compatibility/portability, and lets them get rid of that POS cheap PC they have in their cube next to their $15k sun workstation so they can do email.
The new Powermacs do that, cheaply.
Hell, I think I will finally use the developer discount and order the dual 2gighz come august/sept, itll only cost 4000 USD with a 20″ cinema display, APP, new 30gb ipod, and 2gb of ram.
You people have misunderstood if you think that this chip uses an emulation scheme to handle 32 bit PowerPC code on the 64 bit PowerPC 970…
But that is almost to be expected since many of you are PC users more familiar with Intel, and how Itanium uses a completely different set of instructions than x86… the only way to run x86 code on an Itanium would be emulation…
The G5 (970) was designed from the ground up to run 32 bit code natively. Indeed, when the ISA for PowerPC was put together in the early 90s, there was always the intention for it to grow into a 64 bit architecture that has no problem running 32 bit code…
So 32 bit code that already exists today for Mac OS X will run natively. No emulation at all.
Not every instruction that runs through that processor has to be 64 bit to unlock impressive performance increases. Significant architecture improvements over the G4 count as well… you peope seem to completely forget that.
So what if 10.2.7 (code-named smeagol) is inherently a 32 bit OS? it doesn’t have to be 64 bit to run natively.
To me the highlight of the Keynote movie was when Steve showed a video in which we first saw a Panther and then a Longhorn-cow accompanied with some great music. I think this is some what symbolic for the Apple- and MS crowd, even in the world we have far more cows then panthers!
Another note about emulation: let’s imagine it is 1986: “I am NOT going to buy a 386! What do they think! I bring a very expensive 32-bit processor, and then there is only DOS, which, just like with Apple, will run in an SOFTWARE EMULATED 16-bit mode!”
The same story is true about the Opteron/Athlon64 processors, they can still natively run 32-bit code.
When transforming from the 68xxx series to the PowerPC, they needed emulation because the instruction sets of the processors were completely different. Additionally, the PowerPC did not even bring extra bits! So that is another reason why this whole comment about emulation is pointless.
However, it might well be possible that OS 10.2 does not take advantage of the G5, it is possible that it is just 32-bits. But that doesn’t really matter too, because in that case the G5 is still a more advanced, faster G4.
About real-life situations: Apple has done some tests with Photoshop and such and published the results on their website. Of course these are not wholly reliable, but in each case Apple does not only provide SPEC-results.
Hello
1. The user interface is snappier on X.
A. Well that is a really good statement. Apples UI, OSX is heavily dependent on the slowest part of the pipe between the cpu and the video frame buffer. In this case it goes through the textureing unit on the video card. The current XP doesn’t do that. XP is an graphics map though so I would say the OSX UI is going to be slower on identical hardware. The reason is you must rebuild the texture for each frame. The texture is then mapped to the screen. This will change with longhorn when XP goes to the “lets beat the video card down” model.
2. Spec marks.
Children, children. Spec marks are best used by marketers and people proving a bet. I can remember compiler loops and other toys for them. They give you a feel for the ability of the CPU on that platform to perform math. Never take someone’s word always check official results.
3. 64 bit this and 64 bit that.
THe Alpha paved the way for all this. It turns out the best solution is an hybrid 64 bit and 32 bit operands. The ev4 (If I remember correctly) only had 64 bit instructions. It was a pure 64 bit chip. To bad to concateinate a string would kill the barrel rollers on the chip. Also 64 bit means alignment problems with structures and memory refrences. ( Thats what unalign is for onthe Alpha)
4. Apple seems to have put out an good consumer and atypical product. They are not workstation grade yet. (substandard parts). OSX is a good UI and darwin seems to be a good OS so I hope Apple Computer does well and Steve Jobs gets screwed. Sorry I hate Steve Jobs after he F*cked up NeXT computer inc. Stupid SOB didn’t “like floppies” fin SOB. Great idea man stupid stupisd bussiness man. If you could combine Steve jobs visions with bill gates bussiness sense,
5. If you are going to argue 64 bit this and emulator this please learn about it first (And no the G5 isn’t emulating it’s a lot like an ARM chipset. ).
Leslie D.
If you take results from SPEC database you get something like this (it’s dutch):
http://www.tweakers.net/nieuws/27615
You can find database here:
http://www.aceshardware.com/SPECmine/index.jsp
Come on guys, the 970 is a 64 bit processor, you can not compare a 64 bit processor to a 32 bit processor. Would it harm you in any way to appreciate the new PowerMac G5. The fact is it’s a great improvement from the G4, and I think everyone one would agree on that. Even if the PowerMac G5’s was 1000 Ghz with 500 ghz of front end bus for the same price we have now, there would some people who would be far from being satisfied with it. Some people just hate Apple and their opinions would never change no matter what. They would beleive some siupid pseudo-geek sites on the benchmark test than Apple, Intel or AMD itself. I am pretty sure the guys running the test on those websites know more than the hardware engineers at IBM, Apple, Intel, and AMD. Fact is a fact, you like it or not have to accept it. Why is there always a debate when Apple releases some thing?
“Why is there always a debate when Apple releases some thing?”
because they are misleading people maybe?
“I am pretty sure the guys running the test on those websites know more than the hardware engineers at IBM, Apple, Intel, and AMD. Fact is a fact, you like it or not have to accept it”.
test is not run by Apple but by Veritest..bit more of bashing:
http://www.haxial.com/spls-soapbox/apple-powermac-G5/
I “hate” Apple *only* because of their way too over-hype-d and overpriced hardware, and (their=Steve’s) pompous behavior…But I guess I am bit subjective on that one..and beside, I like reactions of Apple users…I have always smile on my face, when I write things like this…cause I know some mac-zealot out there is going to have a bad day 😉 ..but, again, that’s their problem…
Yeah, that’s what they pay me $60K a year to do, run specmarks on various machines all day, 5 days a week, 49 weeks a year (3 weeks paid vacation + holidays). The higher the specmarks I produce, the better my chances for promotion. I’m really trying hard to shoot for “Chief SPECmark Test Manager” so wish me luck!
Not.
Funny seeing all these people getting all huffy about their precious SPECmarks. Last I check, Photoshop doesn’t use them, Mathematica doesn’t use them, Final Cut Pro and thousands of other apps don’t use them either.
I find it strange that so many people seem to be trying to debunk Apple’s SPECmark numbers, yet no one seems to be making a peep about the G5 finishing 100 Photoshop actions on a 360 MB PS file twice as fast as a dual Xeon, or the G5 doing solving higher order mathematical equations twice as fast in Mathematica, or the fact that a G5 can play a 1000 voices simultaneously in Logic with only 25% CPU utilization. Heck, the dual Xeon couldn’t even play the Matrix Reloaded musical arrangement in Logic without skipping and choking.
If you want speed, let’s talk stuff that people actually use for, like, you know…real work. The fact is, the new G5s are bandwidth monsters and that fact alone will make 10 times the difference in the real world with real apps with real users than some artificial benchmarks. That’s what we saw in all the different application bake-off and when the G5s get into the hands of the general public, we’ll see even more confirmation of that.
All this heavy breathing about SPECs is rather silly, imo.