What video card did Tomshardware used for CPU Charts Q3/2008 ?

[mafi]

What video card did Tomshardware used for CPU Charts Q3/2008 ?
I see a huge gap between High End Middle and Low End CPUs even at 1680x1050


http://www.tomshardware.com/charts/desktop-cpu-charts-q3-2008/benchmarks,31.html

 

[jimmysmitty]

I'm going to guess a 4870X2 at least. Thats the only GPU I know that can get such insane FPS in Crysis.

 

[snarfies1]

Your guess was incorrect, but thanks for playing!
See http://www.tomshardware.com/reviews/amd-intel-pc-processor,2031-5.html for the testbed info (which I think should be linked to the charts themselves to avoid these sorts of questions).

 

[mafi]

GTX 280 at low detail... The worst joke I have ever seen.

 

[zenmaster]

GTX 280 at low detail... The worst joke I have ever seen.

I think the joke is on you and you don't know why.
When Testing CPU performance, the key is to remove bottlenecks of other components.

As a result, when testing CPU performance , you get the strongest GPU possible and set the lowest possible graphic settings. The reason if that the graphical rendering is not done by the CPU.

The point is to find out what the CPU can do.

When testing the GPU, get the strongest CPU possible at the higher speeds so the limits you see are caused by the GPU.

Then you cross reference your CPU charts and GPU charts to try and determine the best CPU and GPU for you.

This way you need 1 setup for each CPU for the CPU test.
You need 1 GPU for each GPU setup.

Let's assum there are 20 CPUs and 30 GPUs.
That would be 50 setups.
If you tested each CPU with each GPU, you are talking about 600 setups.

This is something that frequently needs to be explained to folks.

 

[mafi]

I think the joke is on you and you don't know why.
When Testing CPU performance, the key is to remove bottlenecks of other components.

As a result, when testing CPU performance , you get the strongest GPU possible and set the lowest possible graphic settings. The reason if that the graphical rendering is not done by the CPU.

The point is to find out what the CPU can do.

When testing the GPU, get the strongest CPU possible at the higher speeds so the limits you see are caused by the GPU.

Then you cross reference your CPU charts and GPU charts to try and determine the best CPU and GPU for you.

This way you need 1 setup for each CPU for the CPU test.
You need 1 GPU for each GPU setup.

Let's assum there are 20 CPUs and 30 GPUs.
That would be 50 setups.
If you tested each CPU with each GPU, you are talking about 600 setups.

This is something that frequently needs to be explained to folks.

"When Testing CPU performance, the key is to remove bottlenecks of other components..................................
....................................................................................................................................................."
I agree with you and I already knew that.But WHO CARES how the performance of GTX 280 is influenced by the CPU at low details? Those results are only theoretical but not practical because no one plays at low details.