AMD 64 for the people, but no G5 in the tests

[manitoubalck]

AMD has now released the Athlon FX 64 (Sledge Hammer) clocked @ 2.2GHz and the Athlon 64 3200+ (Claw Hammer,) clocked @ 2.0GHz, their first real consumer based 64-bit Processors at an affordable price. As we know the Opteron has been out since April but that doesn’t really matter. Now the G5 has a true competitor. Tomshardware has done what appears to be the most extensive processor benchmark every undertaken entitled,
“A New World Order: the Athlon 64 FX and Athlon 64 vs. the P4 Extreme” can be viewed here ‘www6.tomshardware.com/cpu/20030923’

The results are the most comprehensive I have seen, and are a must view for anyone wishing to buy a high performance single proc computer in the near future. However the test neglected to include Macintosh’s best offerings in their tests. Why would they go to all this effort and not include the G5? Probably because there target audience is people who use windows and spend most of there time gameing, not mac uses tend not to look to benchmarks performed by others such as www.tomshardware.com but rater turn to Steve for comfort in what they believe.
That aside it still strikes me that they didn’t acquire G5 systems that are now freely available and include them their benchmark. And entitle it ‘A New World Order: the Athlon 64 FX and Athlon 64 vs. the P4 Extreme and the Apple G5,’ Telling the public the full uncut story about the 3 major players?

Remember Tomshardware has nothing to gain or lose form what they uncover. only chip makers do.

 

[tomf87]

Not only do they have a different class of people using G5's or PC's, but they would also need identical software that is not optimized for either platform.

So if they run some Apple or 3rd party product against the G5 and ran a different 3rd party product against the PC's, the numbers wouldn't really mean anything.

It's similar to testing two cars on two different roads. Say, take a 1980 station wagon on a speed strip and go as fast as you can go. Then take a Corvette down a road that you can't get over 25 mph because of debris in the road and very tight turns. Yes, the station wagon may look faster, but the tests are very different.

The data just wouldn't line up to say anything, so what's the point of inviting Apple to the competition.

 

[Ian123]

Whilst I agree with your comments it would still be fun to have some empirical data comparisons between bleeding edge Mac and PC systems using common applications available. If nothing else it would continue the endless debate as to which platform currently has blagging rights - always entertaining to read.

Interestingly AMD have touted their latest 64 bit offering as a killer games machine for the serious enthusiast. AMD sees the multibillion dollar games market as a source for increased revenue in years to come. Certainly 64 bit games will be a talking point next year in the quest for near photorealism in games playing. Those limited to playing 32 bit games will be missing out on a whole new experience.
Apple with the G5 and imminent Panther software is ideally poised to exploit this opportunity. The question is will they?

 

[williacs]

My problem with these tests is they didn't even test with 64bit OSes. Especially when Suse Linux is available in a 64bit version. They made a statement that: Not yet available: software for 64 bits but that really meant to them no windows software. When will that monopoly go away and give Linux and Apple a chance to show their stuff?

 

[macrumors12345]

"Why would they go to all this effort and not include the G5?"

Tom's Hardware only does benchmarks with Windows software (maybe occasionally some x86 Linux stuff). Needless to say, the G5 does not run Windows. Therefore, they do not benchmark it. They are a Windows benchmark site...that's why you'll only find x86 machines there (don't know how many different ways I can say this!).

 

[acj]

first fully 64 bit consumer app

You know what's funny? AMD is actually going to bundle a game with its new CPUs. Called American Army, it is 64 bit optimized, and it's a bootable game! No OS overhead. This is funny to me because it reminds me of the long ago PC gaming days when there were bootable games. In case you didn't know, that means you have to reboot your PC just to play the game!

 

[Independence]

Re: first fully 64 bit consumer app

Originally posted by acj
In case you didn't know, that means you have to reboot your PC just to play the game!

it could probably be started from within windows too.

ever hear of 3 dimensional thinking?

 

[manitoubalck]

Pc weren't the only ones with bootable games

I cant remember the name of the games exactly but I remember having to use 'startup' disks to play some game on my now non-working 'LC-475'

 

[slowtreme]

Tom's hardware is an extremely Intel and Nvidia biased site. The journalistic integrity of that site was lost they day Tom started charging for content and Intel bought huge banners for the site.

Generally the numbers on THG don't match similar tests on any number of sites like anandtech, hardocp, etc.

When THG has an article on video encoding, LAN parties, Games, new toys, and other subjects, they are generally informative and good to read. Reviews and benchmarks on THG can not be trusted however.

 

[DrBoar]

When IBM release the IBM 970 Linux blade servers I am sure that they will be extensively tested against other Linux hardware. This will give us a good test of how well the the G5 fares, perhaps more server applications than Jedi Knight 2 but what the heck

 

[manitoubalck]

http://www.xbitlabs.com/articles/cpu/display/athlon64-fx51.html Here is the web address of the x-bits labs test of the Athlon FX.

http://www.anandtech.com/cpu/showdoc.html?i=1884 And here is the URL of the anandtech test.

The point of my article was not to plug tomshardware but to question why they didn't include the G5 when it would obviously be a fair competitor. You will also note that the G5 is not included in the tests of the testing houses you have named.

Another reply stated that tomshardware are a windows benchmarking site, while this is more than likely true, all the game titles and video compression test, etc. could be run under the mac OS, It would also put the OS’s under scrutiny as to how well they are optimised for their hardware configurations.
I’m just trying to find our the true story from a true 3rd party about what’s on the market.

 

[Jonathan Amend]

There are many reasons why the G5 cannot be properly benchmarked against an Athlon 64 or a Pentium 4. Tomshardware and other benchmarking sites put lots of effort into making sure that the hardware is equal with the exception of whatever is being tested. If you wanted to test an Athlon 64 against a Pentium 4, all you would have to do is switch the chip and motherboard (and people even complain about motherboards skewing the scores since they affect the overal performace). The software also has to be identical to keep the scores fair. Macs obviously can't run Windows and Linux is offten optimized towards one specific chip. Sure, you could take Suse 8.1 for AMD64 and Suse 8.1 for PPC970 and run the respective versions of UT2K3 on them but it's well known that the PPC970 version of Suse isn't quite as far ahead as the AMD64 version and even the AMD64 version still isn't ripe. Macs also need special graphics cards and completely different sound cards (the mothatboard's onboard sound is disabled since it varies from board to board). Idealy you could just switch the part being benchmarked and run the same tests but the G5 makes that impossible.

Also, Tomshardware did test Windows XP 64-bit edition for the AMD which you can see if you watch their 10th video but it was only a beta and the scores were around 10% lower than that of normal Windows XP.

 

[aethier]

Originally posted by Jonathan Amend

Also, Tomshardware did test Windows XP 64-bit edition for the AMD which you can see if you watch their 10th video but it was only a beta and the scores were around 10% lower than that of normal Windows XP.

On another note, there is also no 64 bit version of Mac OS X at this time...


aethier

 

[manitoubalck]

Originally posted by Jonathan Amend
Sure, you could take Suse 8.1 for AMD64 and Suse 8.1 for PPC970 and run the respective versions of UT2K3 on them but it's well known that the PPC970 version of Suse isn't quite as far ahead as the AMD64 version and even the AMD64 version still isn't ripe. Macs also need special graphics cards and completely different sound cards (the mothatboard's onboard sound is disabled since it varies from board to board). Idealy you could just switch the part being benchmarked and run the same tests but the G5 makes that impossible.

That's the whole point Macintosh’s best against the best of the competition with all the trimmings for all the systems. Pitch the G5 using After FX, against the Athlon FX using After FX, get them to render the same sequence, etc…

People are after “Bang for their Buck” If the G5, P4 or the Athlon FX/Opteron can’t cut it in a real world benchmark using publicly available drivers/optimises and most importantly hardware then the results will speak for themselves. It makes no sense to complain that the G5 isn’t up to this or that it’s not optimised for that. What it boils down to is what you get when you pay the dollars. Apple claim to have the world’s fastest personal computer, and I dispute it. Remember it’s only a claim; it’s yet to be proven.