Benchmark results cited by Apple at the launch of its Power Mac G5 desktops yesterday have already come under fire for seeming to not only tweak the Mac test system to improve its performance beyond anything an ordinary user might experience, but crippling rival systems to deliver below-par average user performance. The tests described by Apple CEO Steve Jobs were conducted on the company's behalf ("under contract") by VeriTest. The benchmarks used are SPEC CPU 2000 integer and floating-point tests. Apple asked VeriTest to compare a pre-release a dual 2GHz Power Mac G5 with a Dell Precision 650 workstation based on twin 3.06GHz Intel Xeon CPUs and a Dell Dimension 8300 based on a 3GHz Pentium 4.
The Dell's were running Red Hat Linux 9.0, the G5 Mac OS X 10.2.7. The test software was compiled using GCC 3.3 and NAGware Fortran 95. VeriTest recorded SPECint base score of 800, 889 and 836 for the G5, 8300 and 650, respectively. The equivalent SPECfp base scores were 840, 693 and 646. So the G5 out-performs the other machines, yes?
Well, so says Apple, but a closer look at VeriTest's documentation, freely available from its web site, suggests otherwise. Certainly SPEC figures published on the SPEC web site do, as Register readers noted, along with readers at a number of web sites today. The corresponding SPECint and SPECfp base Dell-provided results for the 650 are 1089 and 1053. Equivalent figures for the Dimension 8300 are not available.
View: The full story
News source: The Reg
The Dell's were running Red Hat Linux 9.0, the G5 Mac OS X 10.2.7. The test software was compiled using GCC 3.3 and NAGware Fortran 95. VeriTest recorded SPECint base score of 800, 889 and 836 for the G5, 8300 and 650, respectively. The equivalent SPECfp base scores were 840, 693 and 646. So the G5 out-performs the other machines, yes?
Well, so says Apple, but a closer look at VeriTest's documentation, freely available from its web site, suggests otherwise. Certainly SPEC figures published on the SPEC web site do, as Register readers noted, along with readers at a number of web sites today. The corresponding SPECint and SPECfp base Dell-provided results for the 650 are 1089 and 1053. Equivalent figures for the Dimension 8300 are not available.
View: The full story News source: The Reg
AIMutation features:
- Plug-ins that add new and exciting features to AIM.
- Transparency on Windows 2000/XP.
- Make transparent windows opaque when they're active.
- Modify the Login Logo with an image of your choice.
- Use Login Logos from skins.
- Log IM conversations
- Add start and finish timestamps to logs.
- Group IM windows into a single tabbed interface.
- Flash IM windows when you receive messages.
- Fast and convenient tab switching through Ctrl+Tab keyboard shortcut.
- Set IM windows to be 'Always On Top'
- Set 'Always On Top' option automatically for all IM windows.
- Set transparency level automatically for all IM windows.
- Skin AIM with your own graphics!
- Hide the ad window on the buddy list.
- Hide the various buttons on the buddy list.
- Cloning.
- And more to come soon!
http://www.amdzone.com/articleview.cfm?art...eid=1296&page=2
i don't know where Apple got some of those numbers
Last edited by 30000 on 24 Jun 2003 - 17:21
How hard is it to understand?
But yet, they disabled HT in at least in 2 of the 3 benchmarks that the G5 wins (it was enabled in the benchmark that the G5 lost to both competitors in).
There's a word for that: "misleading"
Ok, who's gonna cheat next? Compaq? Dell? HP? lol
Its a fact of corporate marketing nothing more!
People in the real world (like independant benchmarking sites) convince me what is good and what isn't
Its also funny to note that while Apple were marketing stability last time round (ie: the Megahertz myth etc etc) now the seem to be focusing on speed themselves, ah ho hum.
Last edited by 11321 on 24 Jun 2003 - 17:35
But it's BS.
The G5 is far slower at 2Ghz than a P4 or Athlon at ~3Ghz or 3000+ rating. All benchmarks have shown that.
Apple's benchmarks didn't include AMD chips, and the Intel chips they used were outdated (533mhz bus variant), on an outdated platform, with SSE/SSE2 and HyperThreading DISABLED!!! But did they disable AltiVec to make it "fair"... they sure as hell didn't. In fact, they ran the P4/Xeon tests with a crippled compiler, and ran the G5 tests with a perfectly tuned AltiVec-designed compiler setup. They even tweaked prefetcher and memory access settings in ways that *couldn't work in the real world*.
What's worse, they ran their tests using a hacked, single-threaded library specifically to inflate their test scores. A library that defeats the purpose of the benchmark because it can't work in a real-world environment.
Their benchmarks are intentionally misleading and by a very, very large margin. It's one thing to leave out the Athlon and Opteron, knowing that they both make the G5's performance look anemic. But to include "fixed" scores and use that as the basis for claiming the "Fastest desktop PC in the world" is ludicrous. I'm surprised no one has taken legal action against Apple for this yet. But I guess you can't sue someone for assuming their consumers are dumb.
SPEC benchmarks are for testing specific parts of chips. When you're testing for integer performance you don't usually use SSE (or AltiVec for that matter) and HyperThreading shouldn't help all that much for a single process which should already be optimized to fully use the chip. Yes, optimized. Code such as this isn't usually just compiled from C, it's tweaked at the machine code level to fully use whichever chip it's run on. This tweaking is why you should never bother with benchmarks, they really don't mean anything, unless all you do is run benchmarks.
Forums about benchmarks/speed always remind me of when I was about five, listening to some of my friends talk about how great their fathers were. They went back and forth, making more and more ridiculous claims until one person said that his father could flap his arms and fly. I guess things never change.
Anyway, threedaysdwn, all your posts are quite venomous (legal action?), why?
Calling Intel's compiler "shoddy" is ridiculous. Many, and I mean many, developers rely on that compiler. It sees far more use on programs for the Windows platform than GCC does. GCC is known for its lack of x86 optimizations.
They used a hacked, single-threaded Malloc library that is completely unusable in the real-world. Apple altered the G5 platform by enabling "Memory Read Bypass" which allows the G5 to run the benchmark without really completing all the requested work.
The fact that they disabled SSE2 and HyperThreading, yet clearly identify the chip as a "Pentium 4 3.06Ghz with HT" is appauling.
Mac users above all should be furious, they're the ones being played!
GCC is a cross platform compiler and if it doesn't perform as well in the SPEC tests as the highly customized compilers then tough.
Ultimately the SPEC scores aren't nowhere near as important as real world scores (of which there are none at all), because each vendor optimises it to their own accord. The x86 platform being as experienced as it is has a higher level of optimisation.
Point is, it seems to be almost level pegging across the board in regards to power, with no-one "owning" the other and all performing amazingly well. Price/performance ratio is still squarely in the x86 court, especially with the 3.2GHz P4 outperforming the 1.8GHz G5 and being cheaper, but for me OS X offsets this price difference; for others it doesn't. Each to their own.
Roll on the Athlon 64 so we can bring 64bit computing really into the mainstream.
but honestly with their marketshare it probably won't be such a big hit to the MS/Intel side of things. it's not like many professionals let alone consumers are saying where the hell is my 64bit software.
Good point macroslover.
Most 64-bit platforms are "limited" to addressing 4 TeraBytes of memory.
The Power4 is so much higher because it is actually 4 chips in one. The PPC970 is a crippled single-core Power4 with AltiVec (Apple's crappy SIMD engine) thrown in.
It also has far smaller caches than the Power4. Considering the Power4 is by far the most expensive chip on there (costing at least more than a 4-way Opteron or 2-way Itanium 2), it damned well better hold its own.
do they honestly want users to believe that now with the revolutionary G5 (just like the 4 others before it), the mac now ACTUALLY outperforms a pc at 1/3 the price? of course not. pc outperforms mac anyday, the pc has tens of thousands more software available, hundreds more hardware companies competing and producing better cheaper products. oh.. and you can upgrade them as neccessary.
GAY MARRIAGE IS LEGAL HERE IN CANADA,but I am not sure about a Mac
Hey, look! We're the fastest at loading Photoshop, and Quarx, therefore we've got the fastest computer in the WOOOOORRRRRRLLLLLD.
Sigh... when will they learn.
As a video editing student I can promise you that Apple has long lost the video editing market to much bigger and better players.
Sure, iMovie might be great for home videos, but for *real* editing, you use a PC. You buy any of Discreet's editing packages (Fire, Inferno, etc.), they come on a PC.
Everything from SGI is basically a PC. Avid's best software is now on PCs only, and when they do offer cross-platform support, performance ALWAYS favors the PC.
The G5 isn't going to change that. Sure, it's a nice step up for Mac users. But no self-respecting video editor is going to switch to a Mac "because they saw the G5 video."
Who needs Dual-PowerPC 2GHz processors and this much ram to run IMOVIE???
I get offensive with Apple really quick when I see zealots come out to protect it so blindly. I like Windows, but I bash Microsoft left and right - especially when it comes to the IT side of things. I do the same to Linux, Unix, Sun, Cisco, Intel, AMD, Cyrix, VIA, Nvidia, ATi, etc..etc..etc....
We have one Apple server where I work. We only have it because we have a graphics teacher who wants nothing but macs in his department. So, for remote storage, we purchased a Mac G4 Server with an external drive. It's cost was enough to purchase two PC servers of equivalent speed and storage. I doubt we'll be running out to purchase a new G5 computer - server OR personal computer. Not when we could buy one he** of a nice 1.8ghz Opteron server system for less.
--> pirce of crap BTW
Sad day it is.
DOWN WITH APPLE
And the 486 was not a 16-bit chip. The 386 was the first 32-bit chip.
The switch to 32-bit that everyone remembers was really the switch to 32-bit Protected Mode operation. It was the ability to address more than 640k of "conventional memory." It was the ability to run the kernel, the rest of the OS, and individual programs in their own protected memory spaces. DOS 3.5 was the first OS to start doing this.
Moving to 64-bit offers a few advantages, but they mostly, if not completely, apply to high-end workstation and server use. A desktop user going from a 32-bit Athlon to a 64-bit Athlon with an otherwise identical design will notice no difference. Even with a 64-bit OS, the difference will be completely negligible. It won't be more stable, and it won't be noticeably faster.
The reason Opteron (and other future Hammer chips) perform so well is not because they can execute x86-64 code. It's because they have on-chip memory controllers, an incredibly balanced pipeline that is always fed and always makes use of its 800mhz (and soon higher) system bus. They have big, smart, fast caches. They have great branch prediction and comparatively light penatlies for mispredicts. It's simply a better chip than its predecessors.
It works wonders in SMP configurations because it is the first solution to offer each chip its own completely independent bus, using HyperTransport. Sound familiar? The G5 works in a similar way.
The move to 32-bit registers "back in the day" was also improtant. But nowadays most chips (P4 and Athlon included) already have 128-bit SIMD registers!
It's certainly true that there's no disadvantage to moving to 64-bit platforms. There will be *some*, if unnoticeable improvement. It's the natural evolution of processor design, but it's not *needed* on desktops to any degree. That's why Intel is sticking to its 32-bit guns for a couple more years. They figure, and rightly so, that desktop users won't be needing several GBs of RAM anytime soon.
Until then, they'll scale the Prescott P4 (or P5 as it's apparently being called now) to 5Ghz and beyond, offering competitive performance for sure.
Apple fans tell me when's the last time you got 5 or 6 hours of battery life on your ibook/powerbook?? Apple claims it can do it, but i've never seen it done. i had an ibook with the battery fully charged and nothing running (like IE, dvd etc...) just the operating system and i let it sit there and it only got 2 hours. now if it does that just sitting there, i HIGHLY doubt it can get 6 hours if you actually do some work on it.
Apple can claim whatever they want, i'll believe somebody not paid by Apple anyday.
Commenting has either been disabled on this article or you are not logged in. Click here to login or register, its free!
Note: Anonymous commenting is disabled in order to keep the quality of responses to a high standard.