Extreme FPS issue

[theq]

i forced VSync off in my Omega drivers (sadly i cant create a rule so i always have to manually turn VSync off because else it has no effect on audiosurf) and it works A LOT better now. It still goes down to ~ 25 fps sometimes but i think i wouldn't even have noticed it if my frames counter wouldn't become red if the framerate is <= 25 fps.

Conclusion:
PLEASE turn off Vsync in a Patch or at least add a setting to the options menu to turn it off.

[NightRyder]

I also turned vsync off, and noticed it isn't as choppy, still lower frames then I want (below 40), but much more playable on enhanced.

[UnknownAssassinX]

I also turned vsync off, and noticed it isn't as choppy, still lower frames then I want (below 40), but much more playable on enhanced.

Same. The frames stayed the same, but laginess was definitely reduced. This helps, but I still wish I wasn't getting 15FPS in premium.

Machine specs for future reference:
2GB RAM (DDR2 I think)
AMD Athlon 64 x2 Dual Core Processor 4200+
ATI Radeon x1650 Pro PCI-E 512MB

Two 21" 1680x1050 monitors (I run the game in this resolution, at 60 hertz; though I used ATI Tray Tools to lock 1680x1050 at 75hertz, so it may be overriding that)

[EdZ]

Tried disabling v-sync, and if anything, the juttering got slightly worse with Normal detail. No change in Enhanced mode. I can't think what could have changed (graphically) since the beta that caused such a massive drop in performance.

[k3Ck]

I also noticed beta ran more smoothly

strange

[Phetal]

okay so, i can confirm that disabling V Sync helped performance. It would be great to include this option so that I don't have to force V Sync off...

And, performance improved only slightly... it's still not as good as the beta.

[Lim-Dul]

Hmmm - usually VSync has NOTHING to do with performance (contrary to popular belief) as it only limits the fps to your monitor refresh rate (because your monitor can't display more fps anyway and huge differences in fps and the refresh rate might cause screen tearing). However, it can affect the performance if something is screwed in the drivers or the game itself - the typical symptom of this isn't low performance, but constant fps jumps up and down (like every half a second or every second).

[DremoraLord]

Disabling Vsync will not affect performance as such, it will make the drops in fps seem less severe. I would definately like a no-vsync option as it is an easy way to perceive better performance

[Lim-Dul]

I disagree. The fps drops should be even more obvious since your max fps would go ABOVE your monitor refresh rate and the minimum fps would be unaffected, hence the amplitude would be even greater. As to the difference between 60 and 75, and surely above 75 fps - it's imperceivable to the human eye.

[riktw]

my cpu is always on 100% when i play audiosurf.
even at startup and character select
i have a amd athlon *64 3000+ on the moment, not so fast
 

[DremoraLord]

Lim-Dul I have SEEN this change, and turning Vsync off does improve the feel of the game for me. I'm not going to scientifically prove it, just offer a useful performance tweak that has worked for me.

[Lim-Dul]

That might be a placebo effect - just like all the guys screaming to turn off vsync in e.g. Counter-Strike, because they think it improves performance (and in fact it only lowers the visual quality of the image by introducing screen tearing).

I'm curious - would you be so kind and run a benchmark with Fraps on a given song with vsync turned on and off? It'd be nice if it were a fast section, since on the slower ones the performance would be compromised by the non-limited max-fps which would disturb the calculation of the average.

The main thing that you should look at is the min fps value - if it is the same for vsync on and off than that setting doesn't do anything.

[DremoraLord]

This is what I was trying to make clear in my last post, sorry if I didn't. 

There is NO change in actual benchmarks, the min fps will be the same, but the frame rate drops will be less noticable than if vsync was on due to the way Vsync works.

Basically, at any one time if the graphics card is unable to produce a fps of 60, then it will automatically have to revert to 30 fps to 'synch' with the screen properly. I have seen this at work in Guild Wars, where it does this every time I entered a crowded scene with Vsync on. Triple Buffering should eliminate this need to go down to 30 fps, but even so there is some responsiveness gained (hence the feeling of less lag, even though this is not true) by disabling Vsync.

Those Counter-Strikers are right, it makes for a less quality gaming experience, but a faster and more responsive one as the game doesn't have to waste milliseconds synching with the screen, thus causing the undesirable side-effect, tearing. I have seen this at work in HL2 and many other games, and audiosurf definately benefits with it off, its worked for quite a few people so there you are 

[eeep]

I would say that part of the mouse processing is linked to the redraw functions, which explains the responsiveness phenomena with sudden jumps in framerate. 

Say rendering is at 60fps, so there's 16.6... ms between frames.  If it drops suddenly to 50 fps, this goes to 20ms.

Let's say mouse coordinates are read every frame, and now let's say that you move your mouse 60 pixels in 1 second.  So that's a speed of 60pixels/s.  If we take a look at the pixel/frame speed though, that would be 60 pix/s divided by 60 frames/s giving 1 pixel/frame vs 60 pix/s divided by 50 frames/s which is 1.2 pixel/frame:

1 pix/frame vs 1.2 pix/frame.

So going from 60 to 50 fps has given a sensation of speed change of 20%, which definitely is noticeable.  Of course, this depends on the assumptions being correct.

With vsync turned on, the video card has to redraw the screen before doing anything else.  This forces the computer to wait until drawing is done.  This can explain the perception of lagginess as perhaps the computer must do some more things as well when drawing the frame, thus delaying mouse input reading some more.

Turning off vsync removes this possible lag, as well as letting the video card fps to go beyond the monitor refresh frequency.

With an Nvidia Geforce 7600 GT with: vsync off, normal quality w/ no AA, 1600x1050 resolution and a Core2Duo E6400 at 2.13 GHz, fps goes from 90-70, usually sitting around 79ish.  This specifically helps with the lag per frame issue.

Also possible, notice that the pixel/frame speed difference occurs due to the ratio of the different fps's.  A 10 fps hit is more noticeable between 60/50(=1.2) than at 80/70(=1.14..).

If you want to experience this for yourself, find a fast song, where mouse responsiveness is automatically dialed up to begin with.  Then use fraps (demo should work for just this purpose) to record once at low fps, then another time at high fps, with sync on for both runs (I think this effectively forces vsync).

You should be able to feel the difference in movement speeds, lag not withstanding.

[petethepanda]

Forcing VSync off definitely helped a little, but when I hit more than one brick at once it still slows down a lot.