View Single Post
  #25 (permalink)  
Old May 1, 2008, 08:21 AM
SKYMTL's Avatar
SKYMTL SKYMTL is offline
HardwareCanuck Review Editor
 
Join Date: Feb 2007
Location: Montreal
Posts: 11,768
Default

I am glad this is becoming a good discussion on the subject since this is becoming quite hotly debated.

While I know that 3DMark06 is CPU limited, you have to admit that it STILL shows the differences between graphics cards quite well even at default resolutions. Just look at any of the reviews we have done on the site: performance differences between single GPUs still stay pretty consistent. What will happen with the next round of uber cards is anyone's guess but I am guessing we will still not see the full effect of CPU bottlenecking. Many of you don't like 3dMark06 but I happen to like it to a certain extent since its scores are pretty consistent from one card to another.

I know many people come down into two camps; those of you who like the consistency of canned benchmarks and those who SWEAR by the true in-game benchmarking. That is why we do a little of both here. This is what I might do with the GPU reviews:

- Expand the 3DMark06 test by not including the Shader Mark scores and adding benchmarks at 1600x1200 and 2560x1600

- Add 3DMark Vantage test with the Performance & Entry tests for sub-$400 cards and Performance and High tests for over $400 cards

- Possibly add one more synthetic benchmark like Lightsmark or then OpenGL Fur rendering benchmark

- Add another in-game benchmark. I am not sure which one yet but I don't want one where I will have to go running around benching with FRAPS. I want repeatable tests so that means timedemo benching. Any ideas are welcome.

I think that may have a little bit for everyone.
Reply With Quote