Intel Benchmarks in Serious Question ?

[xroyal]

Interesting article re Intel benchmarks:

http://www.vanshardware.com/reviews/2002/08/020822_AthlonXP2600/020822_AthlonXP2600.htm

Sorry, cut and paste didn't create link.

 

[Mikki]

That's some crazy stuff, nice info . Of course that affects when you're comparing AMD vs Intel, but it has no bearing on what I use benches for-->which is gauging how the changes, tweaks, and experiments are affecting my system.

 

[Oreo]

I'm not surprised at all, infact I thought this is what Intel did to Sysmark. Seeing as the P4 magically got 20% fast in Sysmark 2002 compared to 2001 (vs Athlon) it was obviuos that they skewed the numbers in their favor. Helping developers making optimized code is one thing (a good thing) but this is just dirty business tricks. Hope they get sued

 

[THUGSROOK]

they all pulled this crap at one point or another.

Nvidia did it not too long ago too.

 

[xroyal]

Oreo: Lawsuit?: Here's one - who knows how many benchmarks will be dragged in?:

http://www.pcworld.com/news/article/0,aid,104075,00.asp

Anyone linkify?

 

[THUGSROOK]

hey!

they said the new Crest toothpaste was new and improved! i personally dont think so and i like the old stuff better!

im gonna sue cause im an ID10T and i believe everything i read!

 

[gururu]

funny business. but its not like sysmark is the only benchmark used; heck, i've never even paid much attention to results obtained using it.

 

[Mikki]

Crest is new and improved? No wonder I don't like it anymore....

 

[xroyal]

THUGSROOK: Can't blame you for being prejudiced with your fine Intel CPU. Personally, I could give a hoot less for benchmarks of any kind. What works for a user just plain works (without breaking one's HDD in pursuit of "who can pixx highest up the pole"). I don't endorse frivolous lawsuits, a too popular pasttime in these United States. I do endorse fair, honest comparisons so that, ultimately, the man it the street (vs members here who know better) gets a fair shake.

 

[THUGSROOK]

a person is smart ~ but ppl are stupid

 

[Rand]

Pretty good discussion here about the relative merits of SysMark 2002. Includes a few links to comments, by Dean Kent, and AMD's own .pdf discussing the potenially... flawed modifications to SysMark 2002.
Anand has also made an appearance in the thread.

 

[Duvie]

I do my own testing, and I pay little attention to sysmark2002...There are enough other benchmarks that show p4 performance is not a trick.....

 

[SupermanCK]

Linked

 

[Kowan]

Go check out the front page of HardOCP about Kyle's view on the copy and paste vanshardware did.

 

[MajinVegeta]

First Hammer will bench the same as the P4 due to the addition of the SSE2 op codes being included. AMD could have included SSE2 in Tbred and it is covered under the current cross licencing agreement so they had no reason other than to cut cost not to include it.

So AMD has made it a point to bash Intel for thier there op code MMX, SSE, and Now SSE2, BUT AMD has adopted the same technoloy into there core and then supports it after bashing Intel for it.

Intel never complaied when SysMark 1999 added in the improvemnet of AMD 3DNow! and is SysMark2001 for 3DNow! Pro.

So why does AMD have its panties up in a bunch?

 

[BlvdKing]

I didn't think it was about SSE2. I thought it was about BAPCO purposely leaving out tests that favor the Athlon and replacing them with tests that favor the P4. That way the P4 gets the better overall score. Also, the test itself does not represent common computer usage, or if it does BAPCO does not explain why it represents common computer usage.

At least that's what I got from all of this.

I never pay attention to BAPCO benchmarks anyway. Everyone knows they are scewed.

 

[SSXeon5]

Originally posted by: MajinVegeta
First Hammer will bench the same as the P4 due to the addition of the SSE2 op codes being included. AMD could have included SSE2 in Tbred and it is covered under the current cross licencing agreement so they had no reason other than to cut cost not to include it.

So AMD has made it a point to bash Intel for thier there op code MMX, SSE, and Now SSE2, BUT AMD has adopted the same technoloy into there core and then supports it after bashing Intel for it.

Intel never complaied when SysMark 1999 added in the improvemnet of AMD 3DNow! and is SysMark2001 for 3DNow! Pro.

So why does AMD have its panties up in a bunch?

Two words: JERRY SANDERS

SSXeon

 

[MajinVegeta]

BlvdKing:

exactly the test Bapco included are 128 bit double presion FP operations that can take advantage of Intels SSE2.

Thus if AMD included SSE2 it wouldnt matter, AMD just wants to cry and whine like they always wrogned then when the include Intels op coded people still think the benches are off so the performance must be greater than the new benches show. So the Support the "better performing" AMD chip.

If you think Intel skews it AMD makes it total BS.

I bought my P4 for the reason of SSE2 and this is the first PC that has gotten truely faster as I have added new software, or updated to newer versions of software to it and upgraded to WinXP.

This prove that AMD want to force it user to upgrade the whole platform for SSE2. and then PNI.

 

[dexvx]

You usually opt to ignore all the synthetic benchmarks (but for reference, I'd still look at things that give a meaningful score such as SiSoft CPU bench/memory bench 3dMark fillrate/polygon count). The fact is that more programs are going to be SSE2 optimized and therefore the P4 will perform better. Not just synthetic programs, but real programs. It is more evident if you look at the history of the P4. The first P4s, 1.5 1.4 and 1.3 barely outpaced a 1.0Ghz p3. The 2.0 barely outpaced a 1.53 Ghz XP (4:3 ratio). Now a 2.53 is on par with about a 2.2 Ghz XP (7:6 ratio) and with SSE2 optimizations, it leads considerably.

 

[TheWart]

yea, I always took those benchmarks that claimed to simulate 'real world' performance with a grain of salt anyways

 

[JeremiahTheGreat]

exactly the test Bapco included are 128 bit double presion FP operations that can take advantage of Intels SSE2.

i think the whole point was that these 128bit double presion FP operations were overused in sysmark 2002, and it was not proportional to what was used in 'real life' terms.

It would be fine if they removed the AXP favouring benchmarks if it was true that these are no longer relevant to real life usage, but no.. they removed (or reduced the weighing) of these benchmarks that are most often used and replaced it with high bandwidth using P4 favouring benchmarks that rarely anyone in main stream computing use.. totally pathetic.

 

[CrazySaint]

Originally posted by: THUGSROOK
a person is smart ~ but ppl are stupid

lol I forgot that line. That was a great movie.