[ut2004] Help? 64 bit linux client and poor fps.

Peter Hoff petehoff at pacbell.net
Wed Dec 1 13:05:11 EST 2004

Nathan Dale Van Eps wrote:

>>>>>BTW, I don't why you get more frames with "UseVSync=True", but this
>>>>>should be false.
>>>> From http://www.techspot.com/tweaks/jedioutcast/jedi-2.shtml
>>>>"*Video Sync*. With this option you can /enable///disable/ Vsync in the 
>>>>Game. Vsync effectively limits the frame rate to your current refresh 
>>>>rate at any given resolution. Frame rate can be perceivably /increased/ 
>>>>when set to *Off* (/disabled/), as frames are rendered as fast as they 
>>>>can be regardless of refresh rate, although you may experience image 
>>>>tearing &/or controller lag as a result. With this set to *On* 
>>>>(/enabled/) you won’t experience either image tearing nor controller 
>>>>lag, although frame rate will be limited to your refresh rate. As such 
>>>>I’d recommend /enabling/ Vsync for best image quality & /disabling/ it 
>>>>when you intend to benchmark system/game performance."
>>>>Turning off vsynch should only give an increase in framerates if refresh 
>>>>rate is the limiting factor. With a monitor less than a few years old 
>>>>that's very unlikely to be the case, so you're better off leaving it on.
>>>That doesn't explain why the original poster would get more FPS with
>>>VSync on.
>>>I've read benchmarking articles where disabling vsync could increase
>>>your FPS by ten or more. And they were on newer systems. But it wasn't
>>>on the Unreal engine. I believe it was on Quake2. And yes, you could
>>>easily get 60+ Hz with monitors in those days. I believe I had mine at
>>>Anyway, I can easily benchmark my UT and will be giving it a try with
>>>this option on and off.
>>One theory I've heard is that with vsynch off you end up with a bunch of 
>>junk in the pipeline, basically the undrawn remainder of the frame when 
>>it gets interrupted.
>>Yeah, I remember those days too. I remember always turning vsynch off, 
>>much like people perform religeous rituals. I don't remember it making 
>>any noticable difference though, even though my fps numbers were higher.
> Yes, that is what I remember too. VSync would stall rendering. I 
> believe I would get something like 5 more fps with it off on quake. 
> But I tested it with UT and it didn't make any difference (under 1fps 
> difference). That is weird option to leave in their if it doesn't make 
> any difference. I wonder if the windows binary is the same... 

IMO it's only useful for benchmarking. If you're testing video cards, 
you don't want to be limited by the monitor. I remember a while back 
where the next generation of cards had come out, but they were still 
using Quake3 as the main gaming benchmark, and they were getting like 
250fps in some cases. Being used as a benchmark is good advertising, so 
they have to keep those folks happy.

More information about the ut2004 mailing list