Originally Posted by macatak
Yes i agree, it's not a true 'apples to apples' comparison like my comparison between my previous 3200+(sktA) and my current 3500+(skt939) where all other hardware/settings were kept exactly the same and only the CPU and Motherboard were changed.
I relaxed my ram to 2.5-3-3-10 and scaled up to a flat 2.60 on my processor with 4xHT 11xMulti 237 FSB 1T at 1:1 and the result with "quality" driver image settings was 75.17. These were the only changed variables.
Does that mean CPU doesn't impact Far Cry? It suggests in this Bench that moving from 3200-3500, or from a 4000-FX 55 may very well not. I would suggest, if possible, when someone makes an action/fps bench for Far Cry they have your character get set to "God" and leave the aggro active. That way you could still walk through the bench and have the added benefit of testing the effect of many AI creatures following you about and creating havoc with their weapons/activity. Aka, much more reflective of real gameplay. X bit labs revealed the flaw of this sort of bench with Doom 3 measurements.
Later, I will run this bench with my processor at variable speeds and the bench at a res value of 800x600 and 1024x768. I do not play those settings and it would still be interesting to test. I have seen Far Cry tests before and it is much better to have a medium speed CPU and high medium graphics card over a high end CPU and medium/low graphics card. Such can be said about many games. A decent CPU and good to very good graphics card is best if you must choose which will be higher end.
These sort of benches are interesting, and they may not reflect real world performance when you involve mass mobs, special effects and AI. Compare 3dmark05 scores between graphics cards and in game play measurements of Far Cry with AA/AF at 1280x1024 in any review that has them. So this bench, and ones like it, are a good measurement of how your rig will do in game when you are walking around not fighting.
All your investigations and time spent are appreciated.