1.0.5 vs 1.1 Performance

[iluvquake4]

I have two copies of Quake 4 with configs configurations - but different versions. Actually the 1.0.5 I can set at ultra.

With 1.0.5 I get 60 fps with vsync on, high quality, 1024x768. With 1.1 it drops to the 30's in MP and SP.

Asus NF4, AMD 3800 x2, 7800 GT, 2 gigs RAM, Sound Blaster X-FI.

This is with vsync on.

Now I wonder .. if it's related to the mouse bug as well -- refresh rate issues?

[Oeloe]

Don't use vsync ever if you want proper mouse behaviour. It caused huge mouse lags in Q3 and it does in Q4 too. You obviously don't have a 30 FPS drop in 1.1. Disable vsync to see how much FPS you actually lost.

[D'Artagnan]

I get the same FPS on each PR, and never use v-sync...

[iluvquake4]

You obviously don't have a 30 FPS drop in 1.1. Disable vsync to see how much FPS you actually lost.

That would only show me how many frames the system never bothered to render -- obviously.

With vsync off, I get horrible tearing.

With 1.0.5 vsync on, the FPS stays at constant 60 FPS and the mouse is very very fast and smooth.

With 1.1 vsync on, the FPS drops to 30 at times, and the mouse is very sloppy.

My guess is that whatever is causing the sloppy / inaccurate mouse bug in 1.1. may also be causing the FPS drops with vsync on.

[Oeloe]

You obviously don't have a 30 FPS drop in 1.1. Disable vsync to see how much FPS you actually lost.

With vsync off, I get horrible tearing.

With 1.0.5 vsync on, the FPS stays at constant 60 FPS and the mouse is very very fast and smooth.

With 1.1 vsync on, the FPS drops to 30 at times, and the mouse is very sloppy.

My guess is that whatever is causing the sloppy / inaccurate mouse bug in 1.1. may also be causing the FPS drops with vsync on.

I know, tearing isn't nice at all, but if you try it in Q3 (r_swapinterval 0/1) you'll see that vsync really messes up mouse movement. In Q4 it's almost exactly the same for me in all pre 1.1 versions.

[KMG]

I can't stand no vsync.

[iluvquake4]

I know, tearing isn't nice at all, but if you try it in Q3 (r_swapinterval 0/1) you'll see that vsync really messes up mouse movement. In Q4 it's almost exactly the same for me in all pre 1.1 versions.

You are speaking for yourself and whatever system you are running. Continually stating your opinion on vsync won't fix the 1.1 problem, especially when it ran fine with 1.0.5. If vsync off works great for you, that's awesome, go have fun with it.

As to Quake 3 -- the only reason to turn off vsync (if you have monitor that doesn't reach 125hz refresh with vsync on), is due to the FPS dependant physics. Otherwise, if your system is fast enough, vsync on with Quake 3 is very accurate as well.

Bottom line is this:

Q4 1.0.5 = Vsync on = Consistent 60 FPS + Accurate Mouse

Q4 1.1 = Vsync on = Major FPS Drops + Sloppy Mouse

[SUPERCHUNK]

I haven't had any issues with vsync with q3 or q4. For those of us with LCD monitors yeah, it very well be might be an issue in terms of comparing them to CRT's. Still, for an LCD user as myself, I pretty much have to have vsync on to have an good gaming experience. I got this monitor when I was still playing q3, and really didn't notice much of a difference after a couple of hours.

The mouse movement differential between vsync- vs- non-vsync might really be apparent to some pros when aim comes down to a milimeter, but I wouldn't know. I play for fun, and while even hitting 73% rail (rarely) with a crappy computer that dropps to 11 fps in heavy firefights, my mouse does seem pretty smooth in vq4.

Since getting over the placebo effect of the beta patch, the new patch's fps seems to be an issue with me. I'm used to lots of fps drops with 1.0.4 (?) with 12+ players so it is pretty hard for me to judge the differences.

Still, the new patch seems to be more "stuttery" for me.

The vsync issue might have to do with the fact that I have a 4ms LCD? I dunno.

[Oeloe]

Vsync just causes an inconsistent drop in mouse sensitivity. At least it does on the different systems i tried it on. If you just sweep your mouse from left to right, you'll see that you'll cover a noticeably smaller angle with vsync on. It's really bigger than a millimeters difference.

[[xeno]Julios]

oeloe there may be a confound here:

in q3 if you use cl_mouseaccel, then the sensitivity of the mouse changes as the framerate does (which is why if you use cl_mouseaccel it's very important to have stable fps).

But it doesn't sound like this is the problem you're talking about, since I'm assuming you tested with v_sync at the same framerate as you did without v_sync?

[iluvquake4]

Vsync just causes an inconsistent drop in mouse sensitivity. At least it does on the different systems i tried it on. If you just sweep your mouse from left to right, you'll see that you'll cover a noticeably smaller angle with vsync on. It's really bigger than a millimeters difference.

Feel free to stop trolling this thread anytime now.

This thread is about the performance issues of 1.1, not about the merits of vsync on or off.

[SUPERCHUNK]

Vsync just causes an inconsistent drop in mouse sensitivity. At least it does on the different systems i tried it on. If you just sweep your mouse from left to right, you'll see that you'll cover a noticeably smaller angle with vsync on. It's really bigger than a millimeters difference.

Feel free to stop trolling this thread anytime now.

This thread is about the performance issues of 1.1, not about the merits of vsync on or off.

The vsync discussion was mostly my fault.

I still have the same problems with 1.1 as you. Maybe the only way to tell if it is a mouse issue vs fps is to run a timedemo in both versions? I dumped 1.1 so I can't test this, sorry.

Also, I have heard that having both versions on your comp might be still causing the stuttering issue, whether you are running 1.0 or 1.1 this might be inaccurate info.

[iluvquake4]

I was getting bad fps before doing the compare -- only running 1.1. Then I installed 1.0.5 to compare. 1.1 has issues. :icon31:

[^misantropia^]

Vsync just causes an inconsistent drop in mouse sensitivity. At least it does on the different systems i tried it on. If you just sweep your mouse from left to right, you'll see that you'll cover a noticeably smaller angle with vsync on. It's really bigger than a millimeters difference.

This is correct. With vsync on, the game renders a frame, then idles until the vertical retrace happens (which might take up to 8 ms with a 125 Hz refresh rate, even longer at lower refresh rates). No mouse movement is registered in that time. Depending on framerate and mouse sensitivity, this will have an appreciable impact.

[[xeno]Julios]

ew that sounds nasty.

[iluvquake4]

Apologies for the trolling comment. It was uncalled for.

I'm just frustrated and worried that 1.1 final will have the same FPS issues as 1.1 beta.

Vsync off isn't an option for me -- the tearing hurts my eyes.