Baker - PQL Help

[necromunga]

I've been using your Pro Quake Launcher for a few weeks now and it runs great. The only problem I am noticing is that I can't get the frames per second that I should be getting when using ProQuake GL.

When I use 'Enhanced GL Single Player' I can get up to 780fps when I run a timedemo and when I run the ProQuake 3.50 (launched off of PQL) I can only pull a 180 framerate, which is a huge difference. My system is deff capable of consistant 500 fps in Q1. I have the vsync off and the pq_maxfps to 999 when performing this also.

It's irritating not getting all the power I can out of my expensive video card If you have any ideas please let me know, thanks man!

Necro

[Baker]

I do know the answer, it's ugly though ... and if I could have put that sole missing item in the ProQuake Launcher, I would have done so.

Here is what you have to do ...

1. Click Start
2. Click Control Panel
3. Click Display Adapter
4. Click the Settings Tab

You will now be on this screen:



5. Click Advanced
6. There should be a tab on the window that opened up labelled "GeForce 440 MX" or "Radeon 9700" or whatever your video card is. Click that tab. Here is mine with my [circa 2002] GeForce2:



7. Look around in the settings for Refresh Rate, it may be dependent on your video resolution. This will look different probably for every video card. Here is mine.



HOWEVER

Myth busting time!

1. According to Rook -- the Qrack engine author -- your FPS max is limited to the refresh rate by your monitor. You are not actually getting 780 FPS with Enhanced GLQuake in the way you are thinking ... my guess is that you are getting 180.

The Enhanced GLQuake engine may update the screen 780 times per second because vsync is OFF, but your monitor is probably at 180 Hz so it updates 180 times per second.

Thread excerpt from:

http://www.quakeone.com/forums/showthread.php?t=1551

Monster,

Let me first say that you are not really getting 1000 fps in darkplaces.
You are getting the data throughput but your monitor is limited to your
refreshrate. So even if your showfps say 99999 you are only getting the max the monitor / card can pump out.

2. ProQuake doesn't have the ability to set v_sync off.

/aguirRe or Rook or anyone else super familiar with Quake engines feel free to correct me above if I am wrong in my interpretation of posts I've read.

3. According to Yugo2Heck, who is incredibly smart, the human eye has absolutely no chance of noticing a difference between a 180 FPS or 500 FPS (if you can actually get the latter).

[necromunga]

It seems I get rather decent FPS during online gameplay anyways. Just when I run a timedemo through PQL it runs poorly. Hmmmm... Maybe I will be alright.

The thing is, I'm running on a laptop with a 15.4" monitor, so there is not much I can do about refresh rates I don't think? I have an ATI RADEON 9800 in case you were wondering what video card I was using.

I don't want to abandon the PQL because of its user friendly interface. I just realized I am getting above 200 FPS during gameplay on multiplayer mode. My next concern is my ping that gets affected when I play with the pq_maxfps. I'm also gaming on a wireless connection which I found out today is horrible for playing games, hehe. So I gotta go get some CAT5.

[Baker]

The ProQuake Launcher doesn't have anything to do with FPS whatsoever, it just sets the command line in a very convenient manner.

Perhaps you selected a lower performance combination of settings from the ProQuake Launcher than your normal command line? (screen resolution, memory allocation, color depth, transparent water <-- watch out!).

Btw: pq_drawfps = the ONLY stat that matters. Timedemo is notoriously inaccurate. That's why ProQuake/DarkPlaces/Qrack/JoeQuake/FuhQuake/EZQuake/etc. have ways to display the FPS, hehe. Those timedemo numbers are horrible ... many of us discovered that in 2002 when engines started including REAL fps displays.

[necromunga]

Thanks a lot for your time! I just wanted to clear some of those things up, sorry dude. I'm actually glad to hear that the timedemo is notoriously inaccurate because otherwise I'd be pissed.

PQL works great, I've been using it since you released it and was only back to playing Quake a few weeks before that. You've made me actually switch to GL Quake because of your program

[aguirRe]

AFAIK, the fps numbers from a timedemo should be correct as long as vsynch is off (set in display properties of desktop). You're only limited by the refresh rate when vsynch is on, therefore it should always be off when testing.

Also, when vsynch is off properly, no application should be able to turn it on. As for what the human eye can perceive, it's not just about screen flicker, in games it's a lot about what moves that are possible. The reported numbers seems very high to me though, so I don't know if there's any practical difference.

I don't know what demo that was run in this test, if it was the built-in demo1/2/3, you could try doing e.g. bigass1 or some other classic test case. It might differentiate the various engines more.

[Baker]

Thanks a lot for your time! I just wanted to clear some of those things up, sorry dude.

No problem at all.

PQL works great, I've been using it since you released it and was only back to playing Quake a few weeks before that. You've made me actually switch to GL Quake because of your program

I'm glad you like it. I tried to set it up so someone who wanted to play Quake could just play Quake, instead of learn about commands/settings/commandlines and all of the other hassles.

AFAIK, the fps numbers from a timedemo should be correct as long as vsynch is off (set in display properties of desktop). You're only limited by the refresh rate when vsynch is on, therefore it should always be off when testing.

I will have to check this out. Monster seems seemed to have the FPS limitation with only ProQuake as well. Maybe ProQuake is "special", heheh

I know those video card control panels scare me and I try to stay out of them. Plus they are different for every video card so I can't assemble screenshots to cover them all

At least FPS seems to work really well in Enhanced GLQuake

[aguirRe]

I should correct myself in that the fps numbers should be "comparable" (with other engines on the same system), if they're "correct" is maybe a bit stretching it. Also, the numbers shown are of course very much depending on the selected demo and settings (resolution and loads of other stuff that affect framerates).

From the benchmarks I've run with various engines, simple engines like the original GLQuake 0.97 is usually also the fastest, even in demos like bigass1. And some engines can't even complete the test without crashing ...

[Baker]

Back in 2002 when the first Quake engine that had a high degree of control over FPS came out (XQuake), players went insane with all kinds of "tweaking" to see what kind of FPS they could get ... or at least what the number would show if the cl_showfps 1 was on or pq_drawfps 1 was on.

Back then, FPS was very important because few players had good machines and/or graphics cards. I used a 400 Mhz Compaq with a Voodoo card and barely managed 35 FPS with GLQuake.

I think you are right. Maybe I was thinking of the "timerefresh" command ... the only one that worked in with DOS Quake (quake.exe).

I can't remember the technical details from back then, hehe. I do think if pq_drawfps 1 shows a constant 180 FPS Necro's in good shape. It would interesting to see what the limit on updates with the typical CRT is -- my monitor shows 59 Hz.

/an interesting subject, heh

[Baker]

Add: From Wikipedia

"An average 17" monitor can reach 85 Hz, meaning that any performance reached by the game over 85 fps is discarded. For that reason it is not uncommon to limit the frame rate to the refresh rate of the monitor in a process called vertical synchronization. However, many players feel that NOT synchronizing every frame produces better in-game performance, at the cost of some "tearing" of the images."

http://en.wikipedia.org/wiki/Frames_per_second

[aguirRe]

Yes timerefresh seems strange, I prefer timedemo for benchmarks. As for leaving vsynch off while playing, I can't stand the screen tearing that is caused by the engine updating different parts of the screen with different scenes. When turning quickly it's usually very obvious that the screen isn't updated in one sweep, causing anomalies.