This Must Upset AMD (also slight benchmark rant)

[Wreckage]

Has anyone noticed that all the early Radeon benchmarks are done using an Intel CPU?

I think AMD needs to ship out some AMD CPUs along with the ATI video cards.

[rant]
I know we all ask a lot out of review sites, but I would like to see cards benchmarked on more than one CPU to get an idea of how a card will perform along with a system that does not have a CPU\RAM\Motherboard combo that costs $2000 while using a $200 mid-range video card.

We have dozens of tech sites to read benchmarks from. They should try and use different systems and games so we have a reason to read more than 1 site. Tired of seeing cards benchmarked on an Intel Quad-Core and 3Dmark....yawn.
[/rant]

 

[swtethan]

cause quad core oc'ed to 3.4+ ghz will take the cpu out of most games to show you pure video performance, AMD cant match that at the moment, but yeah, intel LOL

 

[Zstream]

Originally posted by: swtethan
cause quad core oc'ed to 3.4+ ghz will take the cpu out of most games to show you pure video performance, AMD cant match that at the moment, but yeah, intel LOL

I think I would prefer more "real" cpu's tested. 6300's @ 3ghz or 4800+ etc... Not very many people have these other cpu's.

 

[swtethan]

Originally posted by: Zstream

Originally posted by: swtethan
cause quad core oc'ed to 3.4+ ghz will take the cpu out of most games to show you pure video performance, AMD cant match that at the moment, but yeah, intel LOL

I think I would prefer more "real" cpu's tested. 6300's @ 3ghz or 4800+ etc... Not very many people have these other cpu's.

All you really want to know when youre buying a video card: which one is better? You have a budget, all you need to know is what performs best in that price range, correct? CPU limitation sucks since it makes all the gpu's look like they perform the same (or close to), so why not save yourself $100 or $200 and get the lower card, correct? I mean its right there in the cpu limited chart.


http://www23.tomshardware.com/graphics_...elx=33&model1=706&model2=707&chart=289


http://www23.tomshardware.com/graphics_...elx=33&model1=706&model2=707&chart=293

 

[SPARTAN VI]

I think the point of testing with the fastest possible CPU is minimize any such bottleneck as much as possible. In any case, it doesn't really matter to me. Benchmarks are simply an objective comparison. If you throw a slower CPU into the mix, you run the risk of exposing a bottleneck and will just skew the results in the favor of slower cards.

I can tell you if I benchmarked my 8800GTS & 3700+ combo vs. my 8800GTS & 5600+ combo, you'll see a significant gap in Oblivion. I've had a huge performance improvement, specifically in crowded cities, when I switched to my 5600+

 

[Wreckage]

Originally posted by: SPARTAN VI
I think the point of testing with the fastest possible CPU is minimize any such bottleneck as much as possible. In any case, it doesn't really matter to me. Benchmarks are simply an objective comparison. If you throw a slower CPU into the mix, you run the risk of exposing a bottleneck and will just skew the results in the favor of slower cards.

I can tell you if I benchmarked my 8800GTS & 3700+ combo vs. my 8800GTS & 5600+ combo, you'll see a significant gap in Oblivion. I've had a huge performance improvement, specifically in crowded cities, when I switched to my 5600+

Well, in that case you are benchmarking the Video Card and the CPU. As clearly the CPU is helping the video card attain better performance.

Maybe they should test with a Celeron and take the CPU out of the equation.

 

[Matt2]

I agree with Spartan. You have to remove the CPU bottleneck in order to truly know which card is fastest.

Otherwise, you end up with multiple cards producing the exact same results. Just like when GPUs become bottlenecked by CPUs at low resolutions.

 

[BFG10K]

Has anyone noticed that all the early Radeon benchmarks are done using an Intel CPU?

Of course they are; it's currently the best tool for the job.

The card is the one being tested, not the CPU/platform.

If I was AMD I would be pissed off if people didn't use a Core 2 processor.

Maybe they should test with a Celeron and take the CPU out of the equation.

Putting a Celeron in there would do the exact opposite of what you're asking.

 

[imported_thefonz]

Does it even matter?

As long as they are using the same cpu at the same clock for the entire testing of the cards it becomes a controlled variable in the test with the video card being the manipulated.
The only valid complaint I have seen in this thread is that they should use more common cpu's such as a over clocked e6400 or 4800+ as most people are not interested in quad core or cannot afford it.

 

[soydios]

Originally posted by: Matt2
I agree with Spartan. You have to remove the CPU bottleneck in order to truly know which card is fastest.

Otherwise, you end up with multiple cards producing the exact same results. Just like when GPUs become bottlenecked by CPUs at low resolutions.

QFT

 

[bigsnyder]

It is helpfully to see how video cards scale with slower CPUs. At one time, the ATI cards were not as CPU limited as the Nvidia offerings,
and a few sites did some scaling articles when HL2 and Doom3 were first hitting the market. Again, I understand the purpose of using the
fastest CPU, but I want to see how a card behaves when paired with a mainstream.

C Snyder

 

[tuteja1986]

I don't think AMD would care if it was benchmarked with Intel CPU. They would how ever care if a high tech website like anandtech went and posted "X2900XTX doomed from start". I don't think ATI would be pricing the GPU as high as GTX or Ultra. They will price it at $300 to $400 which will compete in the GTS range. X2900XT can beat up a GTS and also ATI has enabled the professional drivers making its a very temptings offer for workstation user who don't want to buy an $2000 Nvidia workstation cards like FX5500.