For the sake of keeping the post tidy, I'll post some generic graphs and provide a link to a .zip of each benchmark result. Test Setup Stock Q9450 @ 2.66GHz ( 333x8 )1.25v Two HD4870 512Mb 825/1100 Catalyst 9.6 4Gb G.SKILL DDR3 1800MHz 8-8-8-21 @ 1333MHz 5-5-5-14 Vista Ultimate 64-bit Overclocked Q9450 @ 3.7GHz ( 463x8 ) 1.2875v Two HD4870 512Mb 825/1100 Catalyst 9.6 4Gb G.SKILL DDR3 1800MHz 8-8-8-21 @ 1543MHz 6-6-6-16 Vista Ultimate 64-bit The results FarCry2 - All settings maxed at 1920x1080. I was unable to do a test with 8xAA due to my cards' 512Mb buffer. As you can see, there's little difference between 2.66GHz and 3.7GHz with CrossFire disabled, which leads to imply that FarCry 2 isn't overly CPU bound. Left4Dead - All settings maxed at 1920x1080, film grain off. Another example of a GPU bound game. The only test that yielded any significant gain with overclocking was with CrossFire enabled. Resident Evil 5 - All settings maxed at 1920x1080, HDR at 32-bit Much like Left4Dead, the only result yielding a significant gain while overclocked was with CrossFire enabled. F.E.A.R. - All settings maxed at 1920x1080, soft shadows enabled GTA4 - All settings on high at 1920x1080, View Distance 32, Detail Distance 70, Vehicle Density 51, Shadow Density 10 As you can see, F.E.A.R. received little gain between both CrossFire and the overclocked speeds. GTA4 on the other hand, is known to be highly CPU bound, hence the reason for such a small difference between CrossFire being disabled and enabled. Track Mania Nations United - All settings maxed at 1920x1080, motion blur enabled. Another CPU bound game, but with the age of the graphic engine, it's not too surprising. The performance difference between 4xAA and 8xAA is so little, it's almost laughable. The .zip containing each screen cap http://g.imagehost.org/download/0661/Benchmarks The HTML files for FC2 http://g.imagehost.org/download/0075/Results c:
FC2 3Ghz FC2 4Ghz Fear 1920X1080 in game options maxed out with forced 16XAA 3GHZ 4GHZ RE5 Benchmark 1920X1080 DX10 Maxed out ingame settings with 4XAA 16XAF 3GHZ 4GHZ More to come later.
Like I said in the other thread, I will be posting results showing the effects of overclocking using two different video cards and two different resolutions. The clocks speeds being tested for my i7 will be 2.66GHz(stock), 3.8GHz, 4.0GHz and maybe even 4.2GHz. I might consider also adding results with stock GPU vs overclocked GPU. Hopefully this will be all be done by tomorrow and I'll post my results up as soon as possible.
3.6 there still a bottleneck cause most games don't use quads if i can use a fast (4.2+) dualcore on the other hand.
Yes but there are some that get a healthy boost from Quad core and with 45nm Quads clocking as high as they do there is little need to own a dual core cpu anymore. DiRT Mass Effect Bioshock UE3 Lostplanet Ghostbusters RE5 FlightsimX Grid FC2 Assasins Creed GTA4 And there are others not listed here, plus the list is growing.
Looks like I need to delay my testing til further notice gentlemen. I ran into some problems with the 285 and I'm currently working on fixing them.
Thanks Lavans. I don't know what went wrong with it. I usually keep the card at stock OCX clocks for gaming and all the other stuff I do and I only clock it to the speeds listed while doing benchmarks. So I started my benchmark session and had gotten through RE5, Crysis, Crysis Warhead, FEAR and SF4. I decided I'd also throw in Devil May Cry 4 as I enjoyed that game and it has a nice benchmark. So about halfway through the benchmark my temperatures just kept climbing and climbing. This is the first time I had ever seen my card reach 80 degrees Celsius. Running all the other benchmarks prior to the DMC4 one my card had never really gone past 65 degrees Celsius. Now the temperature wouldn't have bothered me much but as soon as it started climbing higher than that my computer had just turned off. So I restarted my computer and wasn't thinking much of it but now the temperatures on the card are just much higher than they were before. The fan is still working as I can hear it so I'm just waiting on my local store to get more AS5 in stock so I can change the paste on the card. Once all that's done I'll be finishing up the benchmarks hopefully. I haven't tested anything since then so I hope nothing was damaged on my card. All I know is even if I play simple games the card burns up to 80 degrees and higher.
im still cpu limited at 4.3ghz :/ lol but less so at 4.5ghz so i think about 4.6 -.7 will be where im not limited
Funny thing is I had done all that just before doing all my testing. Anyways, I picked myself up some Arctic Cooling MX-2 today. As soon as I got home I opened the case up and changed the paste on the GPU. However while taking off the heatsink and looking at the paste that was already there I noticed that the central area of the chip and heatsink looked like there was no paste on it. Perhaps the paste had already gone over it's use period and burned up? That's just speculation though. Works like a charm now though. Temperatures are actually much lower than what they used to be with the stock paste. I'm going to be checking to see if my overclock is still stable or if anything got fubared in the process. I'll report back soon. Edit: It seems that the core speed overclock I had on the card has become unstable. Both the shader and memory speeds are still fine where they are but the core clock can no longer be set to 756 without causing lockups. I guess I'll set it to around 740 and redo all the tests I had done.
That's not too surprising since you're running a dual-core CPU with two relatively powerful GPUs. ^_^;
^^ yeh i know but still i get higer oc and fps than the same priced quad core cpu so its all win as i only game and web brows on my pc
uberfail thread name and subject, it should be named CPU bottleneck not GPU, seriously UBER FAIL NAME AND SUBJECT IS FAILING LIKE UBER FAIL
Just wanted to thank Lavans for the information as well as all the others who contributed! Thank you!:thumbup:
