DOOM3 Tek Discussion --NO SPOILERS PLEASE--
Moderator: enderzero
Well, if you set refresh to 60 and turn on V-Sync, that will limit the frame rate.
I think it might default to the desktop rate since it's opengl. I could be wrong, but it made a difference for me. Mine would go really fast, then slower for a second then fast. (Like it was the difference between 60 and 85.) I set it to 60 and it went away. Now it's a constant flow.
I think it might default to the desktop rate since it's opengl. I could be wrong, but it made a difference for me. Mine would go really fast, then slower for a second then fast. (Like it was the difference between 60 and 85.) I set it to 60 and it went away. Now it's a constant flow.
- McNevin
- Post Apocalyptic
- Posts: 2802
- Joined: Mon Mar 17, 2003 12:39 pm
- Location: Lat: 47.6062095, long: -122.3320708
- Contact:
I even put it on 640x480 and medium detail, and it still did it. From what this guy says, reformatting doesnt fix it either. I'd say it was an amd thing, but my roomate has it too on a p4-2.8/9700.Specs are in my sig. I run Doom3 at 1024x768 high settings. For the most part it runs smooth at around ~51fps but my problem is that when I move around any level sometimes I get this terrible lag where literally freezes for like 1-2 seconds. This happens anytime I'm moving around a level or if there are various monsters on the screen. I even formatted today and reinstalled the game but I still get the same lag and I don't know what it could be. I'm running the latest omega drivers and my video card is at stock settings. BTW, I also tried changing to 800x600 and medium settings and it still happens.
__________________
AMD Athlon XP 2500+ @ 2.2ghz
Shuttle A35N Ultra MB
512mb Mushkin pc3200 (2.5-3-3-8)
WD 80gb 7200rpm 8mb Hard drive
ATI Radeon 9800 Pro
SB Live 5.1
Also fast writes made no change.
Choki is running an i875 board, P4 2.4C, 1GB DDR400, and ATI 9700Pro. He runs at medium detail and it is silky smooth. His system is completely stock too. No OCs of any kind. He says it pauses once in a while, but I don't think for as long as you've been describing. Are you running those beta 4.9 drivers? (Have I asked you that before?) They are supposed to work pretty well. It shouldn't do that in medium detail. It probably should in high detail, because it decompresses enough stuff to fit in 256MB of local video memory. It never does it on mine at all. Seth's X800Pro ran the game extremely well too. I think it's more memory bandwidth and AGP texturing than anything when using HD on 128MB cards. Hmmmm....
- McNevin
- Post Apocalyptic
- Posts: 2802
- Joined: Mon Mar 17, 2003 12:39 pm
- Location: Lat: 47.6062095, long: -122.3320708
- Contact:
I'm running the 4.9 drivers. I uninstalled Doom, defragged, and re-installed, and defraged again. It made no change.
I changed the cache settings. Think its on 40mb for minK, and 256 for main. That seemed to improve play... still get occasional pauses, but nothing like before. Timedemo still looks like crap, how does the timedemo look on SpiderChoki's machine?
I changed the cache settings. Think its on 40mb for minK, and 256 for main. That seemed to improve play... still get occasional pauses, but nothing like before. Timedemo still looks like crap, how does the timedemo look on SpiderChoki's machine?
We've had no need to run timedemos, so I don't know... You can try this:
http://www.beyond3d.com/forum/viewtopic.php?t=14874
http://www.beyond3d.com/forum/viewtopic.php?t=14874
- McNevin
- Post Apocalyptic
- Posts: 2802
- Joined: Mon Mar 17, 2003 12:39 pm
- Location: Lat: 47.6062095, long: -122.3320708
- Contact:
All tests done twice, with a game exit between, and the results averaged.
Catalyst 4.7
Without Humus Tweak: 28.85
With Humus Tweak: 29.65
Catalyst 4.9 Beta
Without Tweak: 29.6
With Tweak: 29.85
Conclusion: Yes there are errors on both driver versions, so the miniscule increase is not worth it.
Catalyst 4.7
Without Humus Tweak: 28.85
With Humus Tweak: 29.65
Catalyst 4.9 Beta
Without Tweak: 29.6
With Tweak: 29.85
Conclusion: Yes there are errors on both driver versions, so the miniscule increase is not worth it.
nasty bright white areas with humus' code:
http://www.sunflower.com/~kyleb/with.jpg
but looks good without:
http://www.sunflower.com/~kyleb/without.jpg
Last edited by McNevin on Mon Aug 09, 2004 1:17 am, edited 1 time in total.
- McNevin
- Post Apocalyptic
- Posts: 2802
- Joined: Mon Mar 17, 2003 12:39 pm
- Location: Lat: 47.6062095, long: -122.3320708
- Contact:
Sorry... typo (Corrected)
Ok i Read the whole thread, really interesting. It Had some proxy comunication with John C.
So they made some more tweaks to the same file and were able to get the same performance increase without the artifacts. This seemed to affect the x800 more than the 9800.
Also Realized I didn't Have Triple Buffer Enabled.
So with TB enabled and Vsync off, running the time demo twice (a 2nd run w/o restart caches some things and yeilds better FPS.) Everyone else online is posting there 2nd run benchmarks so i figure why not eh?
Here are the results:
Humus=NO | Vsync=NO | T.B.=YES
Run 1 = 35.7 , Run 2 (No Restart) = 39.0 , Run 3 (No Restart) 38.9
Humus=YES | Vsync=NO | T.B.=YES
Run 1 = 35.8 , Run 2 (No Restart) = 39.0 , Run 3 (No Restart) 38.9
Here is my CFG
Ok i Read the whole thread, really interesting. It Had some proxy comunication with John C.
So they made some more tweaks to the same file and were able to get the same performance increase without the artifacts. This seemed to affect the x800 more than the 9800.
Also Realized I didn't Have Triple Buffer Enabled.
So with TB enabled and Vsync off, running the time demo twice (a 2nd run w/o restart caches some things and yeilds better FPS.) Everyone else online is posting there 2nd run benchmarks so i figure why not eh?
Here are the results:
Humus=NO | Vsync=NO | T.B.=YES
Run 1 = 35.7 , Run 2 (No Restart) = 39.0 , Run 3 (No Restart) 38.9
Humus=YES | Vsync=NO | T.B.=YES
Run 1 = 35.8 , Run 2 (No Restart) = 39.0 , Run 3 (No Restart) 38.9
Here is my CFG
Code: Select all
set com_allowConsole "1"
set com_videoRam "128"
set r_swapInterval "1"
set r_displayRefresh "60"
set image_anisotropy "1"
set image_useCache "1"
set image_cacheMegs "256"
set image_cacheMinK "40960"
set image_downSizeLimit "512"
set image_downSize "1"
set image_downSizeBumpLimit "512"
Set image_downSizeBump "1"
set image_downSizeSpecularLimit "64"
set image_downSizeSpecular "1"
Those framerates aren't that bad. You could always use Bilinear The line with Linear_Mipmap_Linear (meaning Trilinear,) could be change to several different filtering methods, like linear_mipmap_nearest, or nearest_mipmap_nearest, or other combinations thereof. That may make it look bad though. I haven't played with those settings since some Quake game (either 2 or 3, can't remember.)
I finally got Quake 3 to stop tearing and chopping! It looked terrible on my card until the other night. I actually used settings out of my Doom 3 config file, and now Q3A is silky smooth. I had it set for V-Sync, but it was freaking out anyway, so I use the r_displayrefresh setting, and matched it to the fps I'm using. Worked perfectly after that. Now if I could just enable triple buffering. Nvidia took it out of their OpenGL drivers for some stupid reason. (That's the stupidest thing I've ever heard of. TB is absolutely necessary in my view.)
I finally got Quake 3 to stop tearing and chopping! It looked terrible on my card until the other night. I actually used settings out of my Doom 3 config file, and now Q3A is silky smooth. I had it set for V-Sync, but it was freaking out anyway, so I use the r_displayrefresh setting, and matched it to the fps I'm using. Worked perfectly after that. Now if I could just enable triple buffering. Nvidia took it out of their OpenGL drivers for some stupid reason. (That's the stupidest thing I've ever heard of. TB is absolutely necessary in my view.)
- McNevin
- Post Apocalyptic
- Posts: 2802
- Joined: Mon Mar 17, 2003 12:39 pm
- Location: Lat: 47.6062095, long: -122.3320708
- Contact:
I actually tried those mipmap_linear options and it didnt make much increase, so i just set it back.
Im sitting at almost 40 FPS, that is i think the best i can get. If the web r 'n d guys come up with somthing major, ill try it. Otherwise, im happy.
Seems to not be pausing quite as much now too, only did it once last night. There was a lot of "new" stuff happening on the screen.
Im sitting at almost 40 FPS, that is i think the best i can get. If the web r 'n d guys come up with somthing major, ill try it. Otherwise, im happy.
Seems to not be pausing quite as much now too, only did it once last night. There was a lot of "new" stuff happening on the screen.
Choki runs in medium quality, and I can't see a difference between his graphics and mine. They were side by side on Saturday too. I'm sure if we snapped some screenshots and magnified, that we'd see a tiny bit of compression-related artifacting or something, but during play, it looks the same. There is no difference in effects, lighting, or anything like that, so you may just want to decrease a quality level, and bump your res up one, or just enjoy the faster/smoother play.
alright, i finally read through this thread and realized that I have some BIZZZZZZZZARRRRRRRRRRO optimizing to do!
I don't why my shit is slow compared to everyone else (I'm sure it's Rylands fault one way or another), but I just ran a test and got 28.1 fps runnning at 1024 X 768 in medium mode. With my system specs (P4 3.7Ghz, 1GB Dual DDR PC3200 RAM, 9800 Pro 128MB), I'm hoping I can squeeze out a few more frames.
Here's what I've changed so far to bring me to a solid 28 FPS:
J3RK, begin compiling optimization routines for charlyne 2.0!
I don't why my shit is slow compared to everyone else (I'm sure it's Rylands fault one way or another), but I just ran a test and got 28.1 fps runnning at 1024 X 768 in medium mode. With my system specs (P4 3.7Ghz, 1GB Dual DDR PC3200 RAM, 9800 Pro 128MB), I'm hoping I can squeeze out a few more frames.
Here's what I've changed so far to bring me to a solid 28 FPS:
Thanks Kevlar!set com_allowConsole "1"
set com_videoRam "128"
set r_swapInterval "1"
set r_displayRefresh "60"
set image_anisotropy "1"
set image_useCache "1"
set image_cacheMegs "256"
set image_cacheMinK "40960"
set image_downSizeLimit "512"
set image_downSize "1"
set image_downSizeBumpLimit "512"
Set image_downSizeBump "1"
set image_downSizeSpecularLimit "64"
set image_downSizeSpecular "1"
J3RK, begin compiling optimization routines for charlyne 2.0!
I think it has a lot to do with running in HQ mode on 128MB hardware. They decompress a lot more stuff in HQ. The level of compression is the only REAL difference in the modes. I'd probably be seeing similar performance hits to what you guys are, if I ran in UQ mode. Simply because the amount of texture data is twice what my card has locally.
Mine looks like it is running at sync all the time. I haven't really seen a slowdown yet, even with monsters flying all over the place. My CPU is running slower than yours, (currently around 3.2GHz.) I'm running 1GB of PC3200 cheapo-RAM too. The only real difference is that I'm using the 6800 GT and it's clocked at Ultra speeds. You should at least be able to get higher than 28FPS in 1024 on that thing. On the closest system to yours on the HardOCP Doom 3 Hardware Guide, (Medium Quality 1024x768,) these were the scores: Low=22FPS, Max=60FPS, Avg=50.4FPS.
That was a 3.0GHz P4 on an i875 board with 1GB DDR400, and a 9800 Pro.
I'd run mine in medium detail if I thought it would get me any more speed. I'm already hitting sync though, so it wouldn't do anything for me.
Mine looks like it is running at sync all the time. I haven't really seen a slowdown yet, even with monsters flying all over the place. My CPU is running slower than yours, (currently around 3.2GHz.) I'm running 1GB of PC3200 cheapo-RAM too. The only real difference is that I'm using the 6800 GT and it's clocked at Ultra speeds. You should at least be able to get higher than 28FPS in 1024 on that thing. On the closest system to yours on the HardOCP Doom 3 Hardware Guide, (Medium Quality 1024x768,) these were the scores: Low=22FPS, Max=60FPS, Avg=50.4FPS.
That was a 3.0GHz P4 on an i875 board with 1GB DDR400, and a 9800 Pro.
I'd run mine in medium detail if I thought it would get me any more speed. I'm already hitting sync though, so it wouldn't do anything for me.
FUCK OFF!!!! Yeah, I got yer timedemos right here:
2nd run: 57.5
3rd run: 57.7
That's in 1280x1024 High Detail, 4x anisotropic, cache tweaks. I didn't turn off V-Sync, but that doesn't matter since it's a couple frames under.
(Unless timedemo does that for you.)
I think if I was able to triple buffer it, I'd be sitting at sync. (or if not, it would at least look as smooth as if it was.)
1280x1024 is worth those coupla frames though for me.
Incidentally, 640x480 looks incredible on the television.
Nvidia hasn't yet enabled UltraShadow 2 in their drivers yet, but it will supposedly yield some extra speed in the stencil shadow department when it is.
Oh, almost forgot, that's the performance with g_showplayershadow "1" on.
2nd run: 57.5
3rd run: 57.7
That's in 1280x1024 High Detail, 4x anisotropic, cache tweaks. I didn't turn off V-Sync, but that doesn't matter since it's a couple frames under.
(Unless timedemo does that for you.)
I think if I was able to triple buffer it, I'd be sitting at sync. (or if not, it would at least look as smooth as if it was.)
1280x1024 is worth those coupla frames though for me.
Incidentally, 640x480 looks incredible on the television.
Nvidia hasn't yet enabled UltraShadow 2 in their drivers yet, but it will supposedly yield some extra speed in the stencil shadow department when it is.
Oh, almost forgot, that's the performance with g_showplayershadow "1" on.
- enderzero
- Site Admin
- Posts: 3442
- Joined: Mon Mar 17, 2003 2:40 am
- Location: Highland Park, Los Angeles, CA
- Contact:
My road to 40+ fps
Got it all installed and dialed. Please let me know if you see anything I should change or hear of anything new. Is the general consensus that the Humus tweak is not worth it?
Main system specs:
P4 2.8GHz (533MHz), 1GB slow RAM, 4x AGP, 9800 Pro 128MB, DirectX 9.0C,
All tests are run in Doom 3 Medium Quality, 1024x768 with Fast Writes On. Scores are from the second pass.
First test - Default cfg settings, Catalyst 4.7, Triple Buffer not forced: 33.1
Second test - cfg memory cache tweaked to 128MB & 8192KB: 33.0
Third test - cfg memory cache uped to 256MB & 40960KB: 33.0
Fourth test - Triple Buffer set in ATI Cpanel: 39.9
Fifth Test - CFG tweaked further: 40.2
Sixth test - ATI Catalyst 4.9 beta drivers: 41.9
Whenever I add the lines
set com_allowConsole "1"
set com_videoRam "128"
they are erased when I exit doom. I tried them with seta (why do you write it without the a?) and the same results. The tilde will pull down the console until I leave the game once. What"s up?
The 4.9beta drivers are v6.14.10.6476 dated 7/27 right?
so I am pretty happy with 41.9. Now I should go play!!
Main system specs:
P4 2.8GHz (533MHz), 1GB slow RAM, 4x AGP, 9800 Pro 128MB, DirectX 9.0C,
All tests are run in Doom 3 Medium Quality, 1024x768 with Fast Writes On. Scores are from the second pass.
First test - Default cfg settings, Catalyst 4.7, Triple Buffer not forced: 33.1
Second test - cfg memory cache tweaked to 128MB & 8192KB: 33.0
Third test - cfg memory cache uped to 256MB & 40960KB: 33.0
Fourth test - Triple Buffer set in ATI Cpanel: 39.9
Fifth Test - CFG tweaked further: 40.2
Code: Select all
seta image_downSizeLimit "512"
seta image_downSizeBumpLimit "512"
seta image_downSizeSpecularLimit "64"
seta image_downSizeBump "1"
seta image_downSizeSpecular "1"
seta image_useCache "1"
seta image_cacheMegs "256"
seta image_cacheMinK "40960"
seta image_forceDownSize "0"
seta image_downSize "1"
seta image_anisotropy "1"
seta image_filter "GL_LINEAR_MIPMAP_LINEAR"
Whenever I add the lines
set com_allowConsole "1"
set com_videoRam "128"
they are erased when I exit doom. I tried them with seta (why do you write it without the a?) and the same results. The tilde will pull down the console until I leave the game once. What"s up?
The 4.9beta drivers are v6.14.10.6476 dated 7/27 right?
so I am pretty happy with 41.9. Now I should go play!!