PDA

View Full Version : Card downclocks itself


Fackamato
09-21-05, 02:38 PM
Hi

I have an Inno3D GeForce 6800GT 256MB AGP, which defaults to 350/1000mhz. I've changed the BIOS so it defaults to 370/1000mhz. With Rivatuner and 3DMark 2005 I've found that the highest stable overclock is 410/1200mhz. That's with no artifacts at all. Higher than that on gpu or mem = artifacts. (check out http://www.tehjunkyard.net/clock/3dmark2005.JPG ). I thought that I would stresstest the card even more by running the realtime HDR demo (http://www.daionet.gr.jp/~masa/rthdribl/ ) which I ran at 1024x768, 141 taps depth of field, with detail bump maps and 16x multisample. It had run for 14 minutes when it downclocked it self to default (370/1000mhz, just as in 2D mode), but the HDR demo was still running (now with reduced FPS). The temperature was at stable 80c degrees, with small tops of 81c sometimes here and there. Why did it do this? Why didn't it stay on the overclocked mhz? Was this the throttle in use? It didn't even reach the throttle limit (afaik). I'm using driver version 78.03.

Any ideas?

rohit
09-21-05, 04:28 PM
Hello,
Does it happen only with rthdribl, or with other games/apps too?

Fackamato
09-21-05, 04:49 PM
Hello,
Does it happen only with rthdribl, or with other games/apps too?

I don't know, do you know of any 3D intensive application that I ran run in a window (to monitor the frequency)? I guess it has to be DX9 to fully utilize the GPU... I know that 3DMark can be run in a window, but that requires registration :( (and you can't loop-loop it, there's a little pause between every test where the GPU can rest :( ).

Thanks!

Edit: Hm, I'll try with Farcry, installing it atm!

rohit
09-21-05, 05:16 PM
What i meant was, try running a game, like Doom3, HL2, etc or 3dmark 05.
Kepp RivaTuner running in thebackground, and start Hardware monitiorin (recording).
play for 2-3 mins. quit and check the clocks in rivatuner HW Monitor.

Fackamato
09-21-05, 07:34 PM
This is the strangest thing. I benchmarked Farcry, with patch 1.32. Check out the results at http://www.tehjunkyard.net/clock/farcrybenches/ . Isn't it really strange?! Just look at the FPS, the same in every resolution. I don't get it. Vsync was of course off. Is it because I am CPU limited? (P4 2.6C @ 3ghz, 5:4)

(BTW this time the GPU mhz stayed the same the whole time, but FarCry didn't stress it enough I guess because the temperature was never higher than 70c)

Edit: A friend of mine with a 3200+ A64 (S939), a 6800GT @ 415/1150 gets ~45fps average in the regulator demo, which I get 23.06 average fps in! Both on ultra quality, same settings and all (16x aniso 4x fsaa, 1600x1200). He's using 77.77, but that can't be it, can it? Is it that I'm that much CPU limited?

TierMann
09-21-05, 09:55 PM
Edit: A friend of mine with a 3200+ A64 (S939), a 6800GT @ 415/1150 gets ~45fps average in the regulator demo, which I get 23.06 average fps in! Both on ultra quality, same settings and all (16x aniso 4x fsaa, 1600x1200). He's using 77.77, but that can't be it, can it? Is it that I'm that much CPU limited?

From your benches, it says you have 768mb's of RAM. How much does your friend have? Ultra Quality eats up ram like candy, so that could be the difference.

Fackamato
09-22-05, 07:25 AM
From your benches, it says you have 768mb's of RAM. How much does your friend have? Ultra Quality eats up ram like candy, so that could be the difference.


He has 1GB of ram. But would that do all the FPS difference? (I didn't see any RAM related issues in the benchmark on my computer, it was smooth but low FPS).

rohit
09-22-05, 04:23 PM
Try another game, doom3 or anything..what else do you have?
BTW, when i switched from p4 2.8 to amd 3200+, the FPS in doom3 went up by 10.

so i can imgine what would happen in FC (20 fps difference is expected)

Fackamato
09-22-05, 07:34 PM
Try another game, doom3 or anything..what else do you have?
BTW, when i switched from p4 2.8 to amd 3200+, the FPS in doom3 went up by 10.

so i can imgine what would happen in FC (20 fps difference is expected)

I'll try Doom 3 too. This is my complete computer:

Antec 480W Truepower PSU
MSI 875P mainboard
Pentium 4 2.6C (800/HT) @ 230mhz fsb (5:4 mem divider)
1*512mb noname 400mhz + 1*256mb noname 400mhz
SB PCI 128 sound card
Inno3D 6800GT 256MB AGP
160GB Seagate S-ATA & 120GB WD P-ATA

That's about it... Now doom 3!

Fackamato
09-22-05, 09:02 PM
I'll try Doom 3 too. This is my complete computer:

Antec 480W Truepower PSU
MSI 875P mainboard
Pentium 4 2.6C (800/HT) @ 230mhz fsb (5:4 mem divider)
1*512mb noname 400mhz + 1*256mb noname 400mhz
SB PCI 128 sound card
Inno3D 6800GT 256MB AGP
160GB Seagate S-ATA & 120GB WD P-ATA

That's about it... Now doom 3!

Allright, Doom 3 benched (demo1). My config can be found here: http://www.tehjunkyard.net/clock/doom3/ . Edit: Basically I've set ingame settings to high quality, and followed a tweakguide on tweaktown to improve image quality (such as lod, compressed stuff, image cache etc, check the config file if you really wanna know). I benched with 16x aniso and 4x fsaa in all resolutions. The results are the best of three runs.

1024x768 : 57.5fps
1280x1024: 51.4fps
1600x1200: 42.2fps

Edit: I'll see if I can benchmark HL2 too...

rohit
09-23-05, 03:07 PM
2 questions,
did the throttling take place in Doom3?
what were the temps like?
IMO, doom3 performance is fine.

Also.
i remembered.
When i had p4 2.8e HT, i used to get 50+ fps in doom3, but wehn i ran the custom demo (for benching) i got 42-44 fps.
When i changed to AMD A64 3200+, i still got 50+ fps in doom3, but when benching in custom demo, the FPS jumped to 58-60..
so this solves you low performance in FC compared to your friends.

BTW, wtf are you doin with i875P mobo..with p4 2.6..i mean... normally ppl who bought p4 2.6 bought a budget mobo. and who bought p4 3.0+ bought i875..

Fackamato
09-23-05, 03:22 PM
BTW, wtf are you doin with i875P mobo..with p4 2.6..i mean... normally ppl who bought p4 2.6 bought a budget mobo. and who bought p4 3.0+ bought i875..

The mobo was cheap. I bought the 2.6c because I wanted to overclock and get a high FSB, but I should've bought a used 2.4C instead (they were out of stock everywhere). Still, I'm running the 2.6 @ 3ghz, remember.

The card did not throttle in the Doom 3 benchmarks (not a custom demo, but the demo1).

So you're saying that my Doom 3 scores are allright? In other words, Farcry is far more CPU bound than Doom 3?... That's the conclusion I'm about to make. :confused:

rohit
09-23-05, 03:37 PM
So you're saying that my Doom 3 scores are allright?
Yes. if they are the FPS you are getting while playing, then they are fine.
Farcry is far more CPU bound than Doom 3
never said that.
All i did, was stated my experience of doom3 with p4 and amd.

I bought the 2.6c because I wanted to overclock and get a high FSB, but I should've bought a used 2.4C instead.
yeah. when i had got my p4, amny ppl told me, 2.6's dont O/C that well, and 2.4 is the way to go, or the 2.8. but 2.8c was replaced by 2.8e, and i got screwed..coz that bych heated up like i kept in an oven.

EDIT: i think your card is just fine, keep trying more games.. i hope you have no issues.