..:: PCSX2 Forums ::..

^You can tell it's the video card only. Peak system consumption doesn't go as low as 14W.

I mean the link in truejdd's post.

(04-11-2010 12:41 AM)sakraycore Wrote: [ -> ]^You can tell it's the video card only. Peak system consumption doesn't go as low as 14W.

Technically, it's possible, just not with those parts.

Hi, I use the 5670, it's performance is good, most games at 60 FPS, Xenosaga and FFX at 45 and 47 FPS in some parts as a min, most of time 55+. I use the GSDX 0.1.15 plugin, and you can chose the directx to use, with the directx10, I got 8 FPS less (to 37 and 39 FPS min.), and with the directx9 it go down to 17 FPS in Xenogears! (all in internal resolution = native, I changed the resolution, but it doesn't seem to change the FPS) (*)
The 5750 and 5770 are a lot better. But, I don't think that using them with your power supply would be a good option, they need 450 W as min (recommended). Actually, the 8800 (GT I think, the other versions need a better one) and 5670 need a 400 W power supply at least.
Why you don't want to change the power supply? If you buy a good one, it can stay for later, when you buy a newer pc.

*: I changed the normal resolution, the native res. should change the FPS theoretically. And I always leave it = to the native.

PD: I forgot that I had some speedhacks on. lol The Xenosaga go at 36 FPS without speedhacks. With more speedhacks, it go to 50+ FPS always. I still have to try configurations to get the better performance.
I don't know how I got the 17 FPS in dx9, now I got 35 FPS, with the same config than before (with speedhacks). And the FPS with the DX10 also go up (now is 2 to 3 FPS less than the DX11). Well, the values are still the same. The HD5670 gets 36-38 FPS in Xenosaga without speedhacks, and 50+ with speedhacks (a lot of them). You can see how many gets the 8800 and compare them.

I changed all the resolutions, natives, all, and I always get the same speed, it doesn't make sense for me. x_x The quality increase, anyway. ^^
I didn't know that. I can play the game at maximum quality, and gets the same speed that if I play it in minimum quality. I feel so n00b. :$ I have to play with the configurations more often.

Thanks for all the answers, my normal power supply is 450w, but i have an UPS too, and it is 350w only, and a better is much more expensive than a videocard, but i'll think about it, i had a 7600gs before, with the same pc, and there wasn't any problem for years, but i think that only consumed about 30w at max, which isn't much... (btw I have Asus EN8800GT now)
Well, I think that if Xenosaga runs in native, and at maximum quality with the same fps, that means that the cpu is slow at emulating (maybe i'm not right), but even with the 7600GS in native most of the games were ~full fps, and playable for me, but in higher resolutions they were very slow, with 8800GT i can use higher resolutions, or native, it doesn't matter, if a game is slow, it is slow (probably because of the cpu, or they aren't perfectly emulated yet(?), like 3d in Sakura Wars 5, i don't know)
I'll try Xenosaga soon, with/without speedhacks and i'll post my results

I tried Xenosaga, this is interesting when the game starts (after the movies), fps was around 55 with my usual speedhacks, and GSDX at 1440x1080 D3Dres, so i tried native and it seems a little better, almost 60 fps, but it slows down too sometimes to ~55fps, but i can't say that it doesn't matter anything...
I had VU cycle stealing at 1 (which is ok at most times), now i turned it off, and the fps went down to ~40! and with all (recommended) speedhacks turned off it is around 35-38 fps, i switched back to native again, and it is around 39-41 fps, so it does matter 1-2 fps (i don't think it worths it, and in other games i haven't seen this difference).
Well, with the speedhacks back on again, and 1440x1080res, i tried Combat Mode, it is 58-60 fps too (maybe a little faster than the first screen with the white background), without speedhacks = 45fps |EE=90-100%, GS=30-40%|, with speedhacks GS seems higher, even 50-60%, and EE is the same. I don't see much changes in native mode either...
This was all with DX10, now i tried DX9, with speedhacks, it is around 50 fps in battle, instead of 58-60 |EE: 82-90, GS=45-55%|, without speedhacks it is 39-40 fps... so DX9 is slower (as expected), in both modes, but not with very much... all in all it doesn't seems perfect, but isn't unenjoyable either(with the speedhacks), i wonder if this is mainly because of the CPU or the GPU.

*After the battle the cutscenes are around 40-50 fps (sometimes 60), without speedhacks 30-40 (up to 50), but they are like very slow, and the audio is out of sync, interesting, is it the same for you too?
after the cutscenes/talking ends, the gameplay is back to normal again, around 45-50fps (probably background slows here down too, not like the white bg when the game starts), in native it is 60 fps... without speedhacks it is around 55 fps in native too, but gameplay seems a little slow like this too compared to other games, it seems speedhacks doesn't changes that much here, but native gives +10 fps, if not more. The battles are ok, it is around the same when the game begins, only the real city is slow (probably the flame effects)

I also think it can be the CPU, but it's in 50-80% of use when I get those results, most of time in 50-60%.
With a lot of speedhacks that scene goes to 40-55 FPSs (40 FPSs in some seconds). I tried it with the VU cycle, and in moderate (2), it goes at 60 FPSs in that scene, all the time, but you see it like normal (graphics have the same speed that at 40-55 FPS), and with the VU cycle the audio goes a lot more out of sync. Without the VU cycle, it goes some seconds out of sync, but less than using the VU cycle. If you use VU cycle in 1 (low), it go 50-60 FPSs, but still the graphics have the same speed, and the audio is out of sync (more out of sync that if you don't use VU cycle, but less that if you use it in 2).
In the real city, with speedhacks, I got 60 FPSs. Without them, I got 55 FPSs, and 49 at the parts with flames and enemies. It's the same if I play it on native or not (except that the graphics are horrible in native).
I think the audio out of sync is a audio plugin problem, I use the SPU2 1.2. I got an error loading the 1.4.
I don't like the VU cycle, it increase the FPS, but the speed seems equal to me, even if it saids its running at more FPSs.
If you put "Wait VSync" in the graphic plugin options, it goes a little slower, but its synchronized. With the VU cycle and wait vsync option, it doesn't have any effect, and it seems to use more CPU %, the graphics have the same speed, but it goes at 60 FPSs, and the sound is at 60 FPSs, the video don't.

Conclusion: Don't use VU cycle, it increase the FPSs, but they don't increase in reality.
The 8800GT and HD5670 have similar speeds in PCSX2, but the 8800GT is a little faster.
Increasing a lot the resolution doesn't have any important speed effect (in my case, not even 1 FPS of difference). I don't know what is limiting my speed.

Hmm, Interesting... maybe the CPU matters too, according to a review E6750 is a little better than the x2 240, but the speeds really seems similiar, I wonder what will happen when there will be directx11 support, in pc game reviews dx11 is slower, when it should be faster according to Microsoft. Also both of our cards are 512 mb, there are 1 gb versions too, in theory does that matter anything for pcsx2?
But thanks anyways, it is good to see the comparison results!

The memory of GPU doesn't matter a lot, maybe some extra FPS, but not much.
I haven't tried DX11 to much, but so far I found it a little better than the DX10, but they are like the same. The GSDX 0.1.15 I own has compatibility with DX11, while the 0.1.16 don't. That's strange. But I think the 0.1.15 doesn't use DX11 new options. We have to wait until 0.1.17 I hope, for a good use of the DX11. (I also need to try the other plugin)
The CPU matters, but I think something else is limiting us, if it was the CPU, the CPU usage should go to 99% all time (and mine doesn't reach 90%).
I think I'm getting limited by the clock speeds of my RAM and GPU. And also I only have 2 GB of RAM, maybe I need more.

CPU load will never reach 99%, since that is almost impossible in correctly coded threaded applications. If you get around 80-90% (on the task manager) with a dual core it means you are CPU bottlenecked. The % GSdx reports in its' title bar is the CPU % of the GS thread meaning it measures GPU load. That's what you want to watch out for when seeking GPU bottlenecks. If that % is above 90% it means you are probably GPU limited

Oh, I see, thanks, I didn't knew that.
I got 70-80% in task manager on the slow situations (and the inactive process gets a 15-17%, so I can use a little more IMO). And my GPU is 50-65%. And I get 40-50 FPS, what's limiting me?

CPU clearly.