Are Benchmarks Valid?

[Cricket]

This is taken from an article over at The Inquirer: "THERE'S AN INTERESTING report on Ace's Hardware today called 3DMark03 Performance Factors. Brian Neal reports that while testing, Ace's came across benchmark results that showed a 350MHz Pentium II with PC100 memory and a Radeon 9700 Pro knocked the socks off a 2.8GHz Pentium 4 with dual channel PC2100 memory and a Radeon 9600 in three out of the four 3DMark03 benchmarks.

So does that mean that 3DMark03 ignores practically everything in a system than video cards?

We won't spoil it by reporting the results, but suggest you get over to this page forthwith and see what the whole article has to say."

Cricket

[David14259]

Very Interesting.... thanks for the link cricket.

[sleepypost]

No. Three out of what four tests??? Probably all graphics tests (duh). When it comes to pure graphics, the 9700 PRO SHOULD be beating the 9600. But there are a TON of benchmarks in which the P4 would destroy the P2. Like disk seek, FLOPS, addition, division and access times.

So no, benchmarkmarks are still a reliable way to test ovewrall system performance and stability.

[Ryan2318]

yes they are reliable to a point. when i had my ti 4200 in my old comp with a 900 mghz athlon i was getting like 3-4000 scores even though the pictures looked the same becaue of the video cards, but once i built a new system witha p4 2.4 the scores went up to like 13-14,000. thats on 3dmark2001 btw, 03 wasn't out when i had my old comp. so yeah it doesn't just test the video card.

[Mac Medic]

No, Benchmarks are not valid, even the ones that say Apple rules all

[Blakhart]

Well, I think it also shows that a p2 300 is still a nice piece of gear. And if one wants to know the real world performance, one should bench their own apps and games, not rely on a prefabricated synthetic bench.

[sleepypost]

Originally posted by oem_guy_2002 No, Benchmarks are not valid, even the ones that say Apple rules all

If you are for some reason talking about the "new" G5, that thing is a *piece* for sure. So apple sticks a server processor in it and calls it innovative? Whatever... Hey, I'll bet it rocks a 32 mb graphics card though....

AND, i'll bet it benchmarks 32 bit apps lower than a P4..

[TimPoet]

The answer is, ladies and gentlemen...

3D Mark 2001 is a good benchmark, 2003 is skewed by what ATI and Radeon did to make their last generation cards score higher.

[lil Jimmie]

I view benchmarks as a way I can see any increase/decrease in what I have done to my system either software or hardware and not as a gauge to compare to someone else's system.

[morriswindgate]

Benchmarks are as worthless as a 50LB bag of cow manure in your living room. Not only do manufactures manipulate them to make it seem they have the best product, but the review sites use benchmarks to make crap sound like sugar.
Would you buy an Automobile, say a Dawoo, based upon the type of stuff you see from a review site's benchmarks that go something like> After boreing and stroking the motor, increasing fuel delivery, and bullet proofing the transaxle we were able to get 0 to 60 times in the low fours. This is deffinitly a super performer at a budget price. But if you ran down to get one based upon this you would find yourself in a slow piece of cheap junk.
And unfortunatly you see a lot of people who buy based upon these benchmark reviews, instead of looking at real world problems people are having with the retail off the shelf parts.

[Brown Hornet]

i agree with last two posts all the way

[Texuspete00]

Still wish 3d mark 03 was good as 01. You could change fsb, mem timings, whatever, and it would always show in your score. Now you can OC a GHz and have it barely show. It IS now a graphics benchmark. I thought 01 was to show you how well a pc could play games. Not so any more.