Toms Hardware's New GPU Test Methods

There are lots of different ways to manipulate a benchmark to skew results. It's tough for consumers to really get a grasp on what is the best option for us. Tom's is attempting to resolve that issue with their new GPU benchmarking.

They now have a static test bed that will be used for benchmarking all GPU's. All GPU's will be used in the same rig, all benchmarks will be the exact same, all tools used will be the same. This will be good for us consumers who want true results for a GPU.

If you want to have a read and check some results the link is below.

http://www.tomshardware.com/reviews/graphics-card-benchmarks-charts-review,3154.html

They didn't do that in the past?

They didn't do that in the past?

KABCOOL

[QUOTE="KABCOOL"]

They didn't do that in the past?

whitey_rolls

[QUOTE="whitey_rolls"][QUOTE="KABCOOL"]

They didn't do that in the past?

KABCOOL

So you can just run the benchmarks that make your sponsors look better.

[QUOTE="KABCOOL"][QUOTE="whitey_rolls"] Nobody does - at least almost nobody, go check any website they are always changing the hardware they use or the games they use to benchmark etc.GummiRaccoon

So you can just run the benchmarks that make your sponsors look better.

Erm, i dont think its necessarily that.

A video card can be held back by a CPU and RAM's performance. In many cases, each time they test GPU's they have a different motherboard, memory, hard drive, and cpu configuration. If you had declared your standard configuration to be a system from 3 or 4 years ago, you'd have almost all the higher-end video cards perform identically because they are bottlenecked by the rest of the system - not a very accurate representation of the video cards.

If you keep upgrading the rest of the configuration, then you can see the limit of the video card's power since the rest of the system is allowing it to show its potential, but it doesnt serve as useful comparison tool.

There are lots of different ways to manipulate a benchmark to skew results. It's tough for consumers to really get a grasp on what is the best option for us. Tom's is attempting to resolve that issue with their new GPU benchmarking.

They now have a static test bed that will be used for benchmarking all GPU's. All GPU's will be used in the same rig, all benchmarks will be the exact same, all tools used will be the same. This will be good for us consumers who want true results for a GPU.

If you want to have a read and check some results the link is below.

http://www.tomshardware.com/reviews/graphics-card-benchmarks-charts-review,3154.html

whitey_rolls

I've been reading GPU reviews for a long time. I check the test system specs most of the time, and they are usually the same across all GPUs tested. The benchmark would be almost completely worthless otherwise.

[QUOTE="whitey_rolls"]

There are lots of different ways to manipulate a benchmark to skew results. It's tough for consumers to really get a grasp on what is the best option for us. Tom's is attempting to resolve that issue with their new GPU benchmarking.

They now have a static test bed that will be used for benchmarking all GPU's. All GPU's will be used in the same rig, all benchmarks will be the exact same, all tools used will be the same. This will be good for us consumers who want true results for a GPU.

If you want to have a read and check some results the link is below.

http://www.tomshardware.com/reviews/graphics-card-benchmarks-charts-review,3154.html

hartsickdiscipl

I've been reading GPU reviews for a long time. I check the test system specs most of the time, and they are usually the same across all GPUs tested. The benchmark would be almost completely worthless otherwise.

[QUOTE="hartsickdiscipl"]

[QUOTE="whitey_rolls"]

There are lots of different ways to manipulate a benchmark to skew results. It's tough for consumers to really get a grasp on what is the best option for us. Tom's is attempting to resolve that issue with their new GPU benchmarking.

They now have a static test bed that will be used for benchmarking all GPU's. All GPU's will be used in the same rig, all benchmarks will be the exact same, all tools used will be the same. This will be good for us consumers who want true results for a GPU.

If you want to have a read and check some results the link is below.

http://www.tomshardware.com/reviews/graphics-card-benchmarks-charts-review,3154.html

whitey_rolls

I've been reading GPU reviews for a long time. I check the test system specs most of the time, and they are usually the same across all GPUs tested. The benchmark would be almost completely worthless otherwise.

Most of the websites that I've looked at use the same CPU and motherboard for a couple of years. Quite a few sites are still using overclocked i7 920s and 930s, and have been since they came out. I'm not saying that there aren't cases where we end up with invalid results for comparison due to hardware changes, but most websites that I've frequented make an effort to keep a testbed system around for awhile for this very reason.