I like to run FRAPS for my video card benchmarks here at Legit Reviews as I think it lowers the risk of ‘driver cheating’ by the big video card companies. It is very easy for the said companies to optimize drivers for scripted benchmarks that they helped develop or have sponsored financially. One of the reasons I like to benchmark with FRAPS is that you that you actually get the mouse in your hand and experience the game play on that card. That is feedback that a time demo can’t give you no matter how fair it is. One other reason I like FRAPS is that you catch differences between video cards. One bug that has bothered me for months can be found in S.T.A.L.K.E.R. Clear Sky with V1.5.08 installed.
With an NVIDIA GeForce GTX 275 graphics card installed with Forceware Release 185.65 drivers you can see a tent cover, or military camo mesh netting for those that like to get technical.
When you load up the same scene on the AMD Radeon HD 4890 or any Radeon HD 4000 series card for that matter, you have the wire frame for the covering, but no mesh netting. I contacted AMD about this driver bug back in January 2009 and got this response:
“We have it running on R700 and GTX280 SLI (mocked up as GTX295) and I can’t see a difference between the two. Looking at the images we can see that the camo netting is missing on his system, but for me it’s fine.” – AMD PR
I sent AMD another e-mail about my specific drivers and never got a response, so I let the subject pass thinking it would be fixed in a newer driver build as this was back on CATALYST 8.12 drivers. Well here we are with CATALYST 9.3 drivers and CATALYST 9.4 beta drivers and the bug is still there. This morning when AMD announced the Radeon HD 4890 graphics card the press release contained this statement:
“The ATI Radeon HD 4890 graphics card continues AMD’s proud tradition of technology leadership with support for the latest DirectX 10.1 games, including Ubisoft’s Tom Clancy’s H.A.W.X., Electronic Art’s BattleForge and SEGA’s Stormrise released last month, as well as GSC Gameworld’s S.T.A.L.K.E.R: Clear Sky. When compared to DirectX 10 game play, DirectX 10.1 games have proven to deliver higher game performance and an improved visual experience.” – AMD Press Release on 4/2/2009
Call me silly, but I don’t see an improved visual experience in the screen shots pictured above as an object is actually missing from the game. If you didn’t know any better you would think that the mesh cover wasn’t supposed to be there, however it should be. It might be silly to you, but I want my mesh cover damn it! I want complete game immersion and I feel robbed without that camo mesh. I put a post up on my Facebook page asking if any of my friends could confirm the image quality issue that I was seeing and sure enough David Makin over at ir0Xor.com saw the same issue on a PC that he built this morning just to see if he could confirm what I was seeing. At least I know I’m not going crazy yet!
After some further research from the community it and on the test bench it appears that when running Enhanced Full Dynamic Lighting that the camo mesh is present on NVIDIA GeForce GTX 200 series graphics cards, but not on ATI Radeon 4000 series graphics cards. If the setting is changed to Enhanced Full Dynamic Lighting (DX10) on the ATI Radeon HD 4000 series of cards the camo mesh appears after the game is restarted. This is still interesting as it means that benchmarking ATI versus NVIDIA with the ‘Enhanced Full Dynamic Lighting’ setting doesn’t matter much as the cards will be rendering different objects in various parts of the game and this could impact the frame rate as the render path is not identical.
This is of course just one minor issue with one game title out of thousands available, but I thought it was interesting to point out that what video card you pick does impact your gaming experience. Sadly, no one has the time to sit around and play every single game on both NVIDIA and ATI graphics cards to make sure that all the objects are being rendered or if they appear with the same video settings. Video card reviews are good at showing performance, but in the real world performance numbers don’t add up to having a better gaming experience if objects in the game are not being rendered for some reason. Just to clarify again, both AMD and NVIDIA cards have this issue, but I just spent a full day benchmarking for the Radeon HD 4890 versus GeForce GTX 275 article and saw this issue once again.