PCSX2 review : CPU power DOES matter. Introduction Many of you have probably wondered how much faster will PCSX2 run with overclocked or faster CPU and with a dual core CPU. Will the performance be nearly same or will you see a significant performance increase? That's questions that come into my mind and being a person familiar with overclocking and the fact that I've recently ordered an AMD64 Opteron 165 1.8GHz dual core which is pretty equal to my AMD64 "San Diego" 3700+ 2.2GHz other than the dual core vs single core technology and different stock speeds. I thought what the heck, I'll do a little review for you guys & girls to show how good PCSX2 scale with different CPU speeds and single vs dual core with MTGS and Dual Core Mode enabled. I originally meant to include my old AMD64 "San Diego" 3700+ 2.2GHz CPU as comparision but then this ZeroGS plugin gets released and it simply wouldn't be fair to compare the current GSdx9 plugin to ZeroGS v0.95.2 so Opteron 165 will have to do both modes. Test setup CPU1 : AMD64 Opteron 165 1.8GHz 2x1MB L2 cache Motherboard : DFI Lanparty nForce4 SLI-DR GPU : Nvidia 6800GT 256MB PCI-E 350/1000MHz @ 440/1155MHz RAM : OCZ Gold VX PC4000 2x512MB @ 2-2-2-5 PSU : OCZ PowerStream 520W Sound : Creative SoundBlaster Audigy OS : Windows XP Pro SP2; AMD64 dual core driver, MS hotfix and dual core optimizer installed Cooling : Big Typhoon 120mm CPU aircooler and 8x case fans CPU speeds: Opteron 165 @ 1.8GHz single core mode (stock speed) Opteron 165 @ 1.8GHz dual core mode (stock speed) Opteron 165 @ 2.2GHz single core mode (~22% overclocked) Opteron 165 @ 2.2GHz dual core mode (~22% overclocked) Opteron 165 @ 2.8GHz single core mode (~56% overclocked) Opteron 165 @ 2.8GHz dual core mode (~56% overclocked) PCSX2 config PCSX2 Build : v0.9.1 VM build CPU1 (Opty 165): EERec, VU1rec, VU0rec enabled CPU2 (Opty 165) : EERec, VU1rec, VU0rec, MTGS, Dual Core Mode enabled GFX: ZeroGS SSE2 0.95.2; 1024x768, interlace 1, FFX Hack (only in FFX) Sound : P.E.Op.S. SPU2 DSound 1.4.0; fastest settings CD/DVD ROM : Linuzappz Iso CDVD 0.5.0 Test method 3 games are used to measure the FPS rate, Final Fantasy X, Grandia III and Phantom Brave. I thought these 3 games are among my best suited games for this review as all 3 of them works with MTGS and Dual Core Mode. I use 4 screenshots per game (I could use more but that would flood with screenshots as it already is and I hope you can excuse me from using ImageShack to get this many pictures attached and to get the pictures better organized ) from both higher and lower fps rate areas, should be enough to calculate a quite exact performance increase value. I try to take the screenshots in as exact same position as possible snapping the highest fps rate or with as exact/same polygon count as possible. In some pics it may vary with 1 - 2 FPS or so max but that's not enough to make this comparision biased. CPU speeds used in this review reflect the stock speeds of both the slowest (Venice 3000+ & Opteron 165), average (San Diego 3700+ & X2 4400+ or Opteron 175) and fastest (AMD FX57 & FX62). EDIT : I pretty soon realised displaying ALL screenshots of all speeds would flood this review with 72! game screenshots so I decided to only include both the slowest and fastest mode (1.8GHz single core mode and 2.8GHz dual core mode) so you've got at least some screenshots to compare with so that you can judge so I haven't cheated. Test1 : Final Fantasy X Opteron 165 single core mode @ 1.8GHz (stock speed) : Opteron 165 dual core mode @ 2.8GHz (~56% overclocked) : Results : Based on the screenshots Opteron 165 2.8GHz dual core mode is 22% faster on average than 2.2GHz dual core mode, 33% faster than 2.8GHz single core mode, 53% faster than 1.8GHz dual core mode, 69% faster than 2.2GHz single core mode and 105% faster than 1.8GHz single core mode. In average dual core mode is 35% faster than single core mode. Test2 : Grandia III Opteron 165 single core mode @ 1.8GHz (stock speed) : Opteron 165 dual core mode @ 2.8GHz (~56% overclocked) : Results : Opteron 165 2.8GHz dual core mode is 17% faster than 2.8GHz single core mode, 22% faster than 2.2GHz dual core mode, 44% faster than 2.2GHz single core mode, 48% faster than 1.8GHz dual core mode and 77% faster than 1.8GHz single core mode. Dual core mode is 18% faster than single core mode, so it's quite good also for single core users. Test3 : Phantom Brave Opteron 165 single core mode @ 1.8GHz (stock speed) : Opteron 165 dual core mode @ 2.8GHz (~56% overclocked) : Results : Opteron 165 2.8GHz dual core mode is 34% faster than 2.2GHz dual core mode, 59% faster than 1.8GHz dual core mode, 76% faster than 2.8GHz single core mode, 130% faster than 2.2GHz single core mode and 175% faster! than 1.8GHz single core mode. Dual core mode is 74% faster than single core mode in average, offering by far the biggest speed increase with dual core mode of all these 3 games. ** EDIT ** Here's an example of that overclocked graphics cards doesn't show any performance gain with lowest quality settings in ZeroGS v0.95.2, but with AA4x and bilinear filtering enabled a faster graphics card provides a smaller or even no performance loss at all depending on your CPU speed (the more powerful CPU you have the more powerful GFX card you will need in order to not get FPS drops with AA4x & BF). Traxxx with a powerful X1900XTX graphics card has already proved us that he doesn't get any performance loss at all with AA4x and bilinear filtering and since I got better speed with my overclocked graphic card with max quality settings, it means faster graphics card doesn't gain any speed with lowest quality settings but can help significantly when playing with max quality settings in ZeroGS v0.95.2. Conclusion Does faster/overclocked CPU help significantly? = YES! Does dual core CPUs offer a a lot faster speed at same CPU clock speed? = YES! PCSX2 already scales well performance wise with faster CPU speeds and offers HUGE performance boost with a dual core CPU running in MTGS and Dual Core Mode. Pretty soon there will only be dual core CPUs (or quad cores) available in stores and to very affordable prices so I think it was a smart move by PCSX2 team to include MTGS and Dual Core Mode when they did as it offers very nice FPS rates (PC game manufacturers take should also take note!) and I hope compability will be improved in future so more games can take advantage of it. What suprised me also was how big performance boost it added only by overclocking the single core/dual core CPU, in PC games you see a much lower performance increase with overclocking only the CPU like that which once again shows PCSX2 is VERY CPU dependant as the performance increase was pretty linear or same as the overclocked CPU amount! Take note, I DON'T recommend any1 just start overclocking and I don't take any response for people damaging their computer because they got interested doing so because of this thread. There's always a chance of risking your hardware's life when overclocking, altough when you know what you're doing and stay within reasonable margins the chance is very minimal. I personally have overclocked for a while and always will as it has become like a hobby for me with any comp I own from now on and have not had any hardware failures and I'm not too worried I will ever see one. Anyways the intention of this thread was to show you how well PCSX2 scales with more cpu power and how much dual core CPUs does improve the performance in MTGS and Dual Core Mode vs single core at same speed. I wanna congratulate the PCSX2 team once again for doing the awesome work they're doing, they are prolly the most amazing team in the world, keep up the good job! Special thanks flies out to Mr_Slinky over at XtremeSystems forums for selling his nice clocking Opteron 165 to a very reasonable price and to my big brother for letting me use Microsoft Excel on his computer so I was able to make the nice graphs. EDIT : The graphs are included in the attached file below so you can browse the graphs faster & easier in case ImageShack is slow or the graphs doesn't show up at all (should be 4 for each game).