Intel Benchmark scamming results?

[Ticktanium2038]

I read on a russian site that through Intel's influence that bechmarks are skewed toward favoring Intel CPU's. This is very similar to the NVIDIA optimised driver issues that are going around.

Bapco's sysmark seems to be the big offender they are targeting.

The suspicious came about when the Mobile P4 somehow outscored the regular P4 using a nearly 50% lower clock rate.

So is there any way of disabling the CPU detection in the benchmarking utilities and verifying if this is true? I assume this is what the russian people accusing Intel of doing are attempting. Or making an AMD register as an Intel CPU and see if the benchmark increases magically. This would ultimately show that benchmarks are being skewed toward a paticular CPU.

It really doesnt make sense that games and applications show a much different story than simulated benchmarks. So there must be something behind the benchmark story.

 

[DAPUNISHER]

The Bapco/Intel scandal is acient history, this snipit is over a year old and the controversy started well before that, where have you been?

The fact remains though that the P4 is faster than even the 64 in some situations and when Multitasking comes into play so synthetic measurements be damned, the P4C is a great CPU

BTW, you are starting off like the anti-stevejst He was a troll who was recently banned for starting flame wars against AMD now you show up and in 2 posts are trying to do the same to Intel, steve? is that you?

 

[Jeff7181]

Most synthetic CPU benchmarks that give you an arbitrary number will favor clock speed. A better indication of how the processor performs would be to run a benchmark like UT2k3's Botmatch... or... time it to see how long it takes to complete a working unit in a Distributed Computing application... or... run a Quake 3 benchmark... or encode some divx, or mp3's from wav files already on your hard drive.

 

[Ticktanium2038]

This is actually new news that runs on the heals of the recent centrino chipset. Surfaced about a week ago.

This is not the year old issue resurfacing.

Seems the benchmarks arent configured to detect centrino and properly guage the CPU which is causing anomolies in its performance ratings. This has caused someone in russia to investigate how a centrino blew away a P4 3200 in some benchmarks.

If I were investing in Intel I would want some answers.

 

[Wingznut]

Originally posted by: Ticktanium2038
The suspicious came about when the Mobile P4 somehow outscored the regular P4 using a nearly 50% lower clock rate.

First off, a Pentium-M (aka Banias, aka Centrino) is not a Mobile P4 and is a totally different architecture than a P4. Thus, they will perform very differently.

So... I don't really understand what you are trying to say. Do you have a link to the article in question?

 

[Ticktanium2038]

Tuesday 09 September 2003
http://www.theinquirer.net/?article=11465

Ahmetov claims: "SM ICC always had been the test which was used to indicate Pentium 4's advantage over Athlon XP. Anandtech had not paid much attention to this results, but they are very interesting. Don't look for the Pentium M at the top of the results graph. Instead wonder that 1600 MHz P-M scores less than 1400 MHz P4! Concern that in every other synthetic and real life tests P-M trashed its older brother, including Content Creation suites in SYSMark 2001 and PC Magazine's Winstone MCC:"

How much says its still happening today?

How about we turn off CPU recognition in the software and find out?

 

[Wingznut]

theinquirer referenced both the Russian article, and this one from AnandTech. I've tried, but I cannnot decipher the Russian article, and therefore really don't know what it's comparing or the methodology used.

But looking over the AT article, the reasons are not only obvious, but clearly explained by Anand.

Internet Content Creation SYSMark 2002 is much more optimized for the Pentium 4's architecture, which is why we find the two Pentium-M processors at the bottom of this chart. Overall content creation performance should be somewhere in between what we see here and what we saw in Content Creation Winstone 2003. One thing is for sure though, the Pentium-M at 1.6GHz is significantly faster than the Pentium 4-M at 1.6GHz.

Btw, I'm not sure how familiar you are with theinquirer... But they aren't the most accurate or reliable source on the 'net.

 

[Ticktanium2038]

Something still doesnt add up when benchmarks show a different story from applications.

Still I would like to see someone remove/disable the cpu detection from the benchmark utilities to determine if their is foul play afoot.

But I think everyone is too busy examining the NVIDIA/ATI driver issues with DX9.

I will stick to application/gaming performance and throw out any Benchmarking program analysis. Benchmarks dont get my work done or add any FPS to my games and dont appear to be based upon any reality situtation.

I would trust LOADSIMS on real world applications over a company developed benchmarking utility to tell me what is best.

 

[DivideBYZero]

This is older than your Mother, Tick.

Next time you read a CPU review, count how many synthetic benchmarks they do compared to 'real world' tests. Any site worth it's salt will put up more real world examples than simple sythectic benchies.

You will also note the anti-synthetic sentiment in this community, too.

This is not news.

 

[oldfart]

Originally posted by: DivideBYZero
You will also note the anti-synthetic sentiment in this community, too.

I resemble that remark!

 

[DivideBYZero]

Originally posted by: oldfart

Originally posted by: DivideBYZero
You will also note the anti-synthetic sentiment in this community, too.

I resemble that remark!

Keep 'em real!