Interesting find, Benchmark favs Intel purely based on CPUID...

[pete4r]

Interesting find, Benchmark favs Intel purely based on CPUID...

http://arstechnica.com/reviews/hardware/atom-nano-review.ars/6

 

[JDocs]

This is old news and has already been discussed here. It could be cheating or the developers could have decided to identify CPU functionality (ie SSE2 and MMX) from the ID in which case the Via Nano would under perform.

Other possibilities are that the developers simply knew more about the Intel CPU and where able to get more performance out of it.

 

[spuddyt]

and once again it is proved that 3dmark is a pile of crap and its much better to benchmark real world situations....

 

[Wisecracker]

It's not ""Old News"" when Ars Technica demonstrates what can only be intentional bias against AMD ...

Second, there's the issue of performance when Nano is identified as AuthenticAMD. If performance between the AMD and Intel CPUIDs was identical, there wouldn't really be a story here, but it isn't, and that's curious. Futuremark could plausibly argue that VIA's C3/C7 processors weren't exactly on the radar back in 2004-2005, but AMD and K8 certainly were, and K8 launched with full SSE and SSE2 support, with SSE3 added in 2005.

None of this constitutes proof of wrongdoing, but it flies in the face of Futuremark's neutrality claims. Bad code is a fact of life, but companies that write benchmarks for a living and sell those benchmarks as evaluation tools have a responsibility to ensure that their software delivers the neutral framework that it promises. Based on the information I've gathered thus far, it seems Futuremark may have created three paths—one for Intel, one for AMD, and one generic "other" path. There's nothing wrong with optimized code paths, but our results would seem to indicate that some paths are decidedly more optimized than others.

I don't blame the manufacturer for wanting to optimize performance in benchmarks. This has been the case since day one of the first benchmark ever run. If they want to take the chance on being exposed for their 'fudging' of a test it's on them to take the ultimate responsibility. This is the case not only in synthetic benchmarks but 'real world' testing.

That is not the case with the testing organization itself. People have some 'splainin' to do ....

 

[jimmysmitty]

I love how the OP tries to blame Intel when its more than likely Vantages screw up.

Either way PCMark/3DMark is in no way a good measure of a PCs real world performance. Its just a way for people to extend their e-peen and thats it.

 

[pete4r]

This is old news and has already been discussed here. It could be cheating or the developers could have decided to identify CPU functionality (ie SSE2 and MMX) from the ID in which case the Via Nano would under perform.

Other possibilities are that the developers simply knew more about the Intel CPU and where able to get more performance out of it.

Im sorry but not all of us checks this forum 24/7 like you do. And NANO reviews only recently published (2-3 weeks, this term for technology is pretty new to me, but by all mean, not "OLD" news)

It is either a Futuremark screw up, or Intel's "Anti Competitive play" i.e. "We are Intel, we give you $1bn for you to boost all performance by Intel CPUs and crap out the rest especially AMD" and the easiest way of doing this? - detect by CPUID name

Just using my imagination, nothing is fact above comments

 

[Wisecracker]

Im sorry but not all of us checks this forum 24/7 like you do. And NANO reviews only recently published (2-3 weeks, this term for technology is pretty new to me, but by all mean, not "OLD" news)

It is either a Futuremark screw up, or Intel's "Anti Competitive play" i.e. "We are Intel, we give you $1bn for you to boost all performance by Intel CPUs and crap out the rest especially AMD" and the easiest way of doing this? - detect by CPUID name

Just using my imagination, nothing is fact above comments

I cannot find any response from Futuremark concerning this issue. Anybody see anything?

There were some interesting comments to the Ars piece on the matter. One was a rather lame attempt that said something like 'maybe the Nano is more like the Intel chip because of the front side bus' without explaining why the score would increase 50% simply by changing the CPUID (or why the AMD ID scored 33% less than that) ...

 

[jimmysmitty]

^The AMD chip scoring less could be because of the FSB. If the CPUID tells it that the AMD chip has a IMC and that it uses HTT links it could have tried that and not had the memory bandwidth like most AMD chips do.

I would love to see a true CPU being able to do this because with UMPC type chips uts hard to say.

I can't think of any reason why it would perform better with a Intel CPUID nor why it wouldn't perform better with a AMD CPUID. It would have been nice though to see what CPUID they used for each chip. Was it C2D for Intel and K8 for AMD?

 

[radnor]

^The AMD chip scoring less could be because of the FSB. If the CPUID tells it that the AMD chip has a IMC and that it uses HTT links it could have tried that and not had the memory bandwidth like most AMD chips do.

I would love to see a true CPU being able to do this because with UMPC type chips uts hard to say.

I can't think of any reason why it would perform better with a Intel CPUID nor why it wouldn't perform better with a AMD CPUID. It would have been nice though to see what CPUID they used for each chip. Was it C2D for Intel and K8 for AMD?

Just changing a string tag should not cause this disparity on the SAME chip. I smell foul play here. Might it be by Futuremark, Intel, AMD, NVIDIA or all of them.

This feels like sand tossed to the eyes of enthusiasts.

Me not happy.

 

[B-Unit]

WTF does nVidia have to do with it? I cant discount Intel having there fingers in this, but the majority of the blame MUST fall on Futuremark. They are responsible for developing and marketing an unbiased way of comparing PC hardware. This is evidence that they dont.

 

[spongebob]

Just changing a string tag should not cause this disparity on the SAME chip. I smell foul play here. Might it be by Futuremark, Intel, AMD, NVIDIA or all of them.

This feels like sand tossed to the eyes of enthusiasts.

Me not happy.

There are many levels of CUPID. Changing the ID string (level 1 query) will result in misinterpretation of the extended CPUID levels IIRC, and may have some impact on the standard levels as well. If CPUID is used for the purpose of making full use of the chip's features and capabilites, changing the string completely buggers the test, IMO.

 

[Grimmy]

Thing is... if its purely based on CPUID, why only the memory?

The graph above covers all of PCMark 2005's test suites except for the memory benchmark. As you can see, everything here is as it should be; PCMark doesn't care if Nano identifies itself as GenuineIntel or CentaurHauls. Memory subsystem performance, on the other hand, looks a wee bit different.

In my opinion, its more based on poor coding. Heh, you might as well blame other software makers on game benchmarks for low scores on CPU's.

 

[radnor]

There are many levels of CUPID. Changing the ID string (level 1 query) will result in misinterpretation of the extended CPUID levels IIRC, and may have some impact on the standard levels as well. If CPUID is used for the purpose of making full use of the chip's features and capabilites, changing the string completely buggers the test, IMO.

Yes, and it will use different optimizations for each chip. The problem is there mate. The Chip shouldn't be faster with the Intel Tag. Should be slower or much slower. I don't disagree they make slight adjustments to improve each chip and use the strong points and capabilities for each chip.
But they don't over tweak it for some chips, and for others, they just pass the generic code.

I made a reference to Nvidia because Nvidia has been found in the past tinkering inside the drivers just to pump in Futuremark. And no, i am not talking this latest physx drivers.