ATITool artifact detection: inconsistency?

N

Nihilanth99

2[H]4U
Joined
Aug 28, 2002
Messages
3,564
I've been using ATITool 0.27 beta 4, and have noticed that it detects massive amounts of artifacts (invisible) shortly after having it scan for them—but not always. Right after installing a new GPU cooler, it went hours without artifacts, but right after the next reboot, it was back to detecting them constantly. Temperatures seem fine (39c idle, 63c load), and this is at 100% stock speeds.

I'm using Catalyst 7.12's and the GPU is a Radeon X1950XTX with a Thermalright HR-03 Rev. A installed. ATITool's artifact sensitivity slider is all the way to the left, or maximum.
 
I've found that 7.9's worked best performance-wise for my XTX. 7.10's were close, but I ended up losing a little bit to 7.9's. I figure the 1950XTX is a dead breed in comparison to modern high end eqpt. I do have an XTX and am planning on installing an HR-03 Rev. As I can't afford a newer card.

I've also noticed some flaky-ness in ATI Tool's artifact scanner, I personally use .26 for my application as I do overclock my card and none of the .27 versions save the voltage settings correctly if at all. I generally use the 3D Fuzzy Cube for stress, but I've found the Doughnut shaped Fuzzy Cube Stability Test to put a heavier load on my GPU so I'll run that windowed at 1024x768, 6AA for a couple hours or more (to compensate for when I'm gaming a lot and ensure stability), plus you'll know if you have artifact issues pretty quickly with this method. I'm at 708/1080 with just 1.525 VGPU, but if I increase the rest I can hit 715/1136, but the temps are insane even at 100% fan speed...that's where HR-03 comes in. ATI Tool is a great program, W1z has done an excellent job with it over the years as I've used it since my 9600, but for some applications I find some of the options in it to be unusable or insufficient for the needs of that application so I'll find other ways to get what I need verified.

Until a cooler install I've been using a Bios mod for clocks that I did with Rabit and ATI Tool for more clocks since Rabit can only adjust VGPU, and ATI Tool for fan speed since no matter what I edit in RABIT (I even set all at the same speed!) it does not change the card's speed algorithm...I suppose if I could hex-edit I might have a chance..oh well.
 
Thanks for your detailed reply.

I just installed the 8.1's using the old Driver Cleaner method, and ATITool still has the detection errors. Here's a sample from the log; it's pretty convincing that something's not right:

Code: 
2008-01-16 19:55:31	I Scan for Artifacts started...	Core: 648.00 MHz	Memory: 999.00 MHz
2008-01-16 19:55:31	I Scan for Artifacts running for: 0:00:00
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.23.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.25.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.26.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.28.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.29.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.29.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.29.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.29.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.29.	1 pixels don't match.
2008-01-16 19:56:06	I Artifacts detected after 00:35.29.	1 pixels don't match.
It's always 1 pixel, and the window indicates a delta of 22 each time. Again, temperatures are well within limits and I see no visible artifacts in either the test window or in any applications.
 
I gave up using ATI tool for stability checking quite a while ago as it always claims I can clock WAAAAY higher than I really can, even when set to hardest testing.
I thought this was due to it being an NVidia card but it does the same on my old X1800 card.
The temps never get as high as in game testing.

Even if you can get it to stop displaying those errors, I wouldnt trust it.
Use games/benchmarks to verify it works.
I found Oblivion on max settings gets my card the hottest when rapidly looking left to right with the big city in full view from a distance so it fits across the screen and Battle For Proxycon test in 3DM03 is the best for testing memory stability.
If you have any weather mods for Oblivion, only test in bright sunlight otherwise the number of objects drawn can be hugely reduced.
 
I know this is an old thread, but I'm wondering if anyone else has run across this problem with ATITool as well. I think I may have the same problem.

I overclocked my 8800GT to 725 (with linked shaders at about 1800+), and 1000 for memory using RivaTuner, since ATITool still doesn't work with the 8800GT. I ran the Crysis GPU benchmark and 3DMark06 without any problems, but ATITool tells me there's artifacts, even though I don't see any. Temperatures are fine, never exceeding 60C when running the ATITool test.

My ATITool log is very much like Nihilanth99's, except mine says 2 pixels don't match instead of 1 pixels don't match.

Has there been any sort of conclusion on this? Is ATITool reliable for artifact testing, especially with newer generation cards?
 
I just ran ATITool along with Rivatuner Hardware monitoring and found some odd things.
First the GPU useage never got over 95% w/ the 3d window open.
Second ATITool showed the fan speed at 60% at idel and Rivatuner had it at 20%, nder load it got to 46% acording to Rivatuner and ATITool had it listed at 100.
I got no errors when running the Scan for artifacts.

Running some of the ATI tech demos I got my video card to 100%, the temp differance between ATITools 95% and ATI tech Demos 100% was about 6deg C. Not sure how well a tool designed for overclocking can do very well if it cant get the GPU to 100% load.

I had been trying to get a differant fan profile setup useing Rivatuner and cant seem to get it figured out, and I am afraid of the fan speed reading error I am getting with ATITool. I have used ATITool in the past to remap the fan but it always gove the the correct reading with older cards.

Maybe ATITool is gettting a little old to be useful on newer video cards.
 
Check ATItools config, you can tell it to use 100% GPU in Settings/Artifact Scanning.
Move the slider all the way to left.

ATITool and Rivatuner have their own fan profiles which override the NVidia ones.
They are likely to have different settings.

When I used ATItool a long time ago to set the fan speeds on my old X1800 card, it would occasionally invert the fan profile.
ie when it was supposed to run at 20%, it would run at 80%, as the card got hotter the fan slowed down more, quite deadly!
Also when running the GPU 100% and getting no artifacts, I can get much higher GPU clocks than I can use in games, so its not very good as a overclocking tester.
I dont trust ATItool.
fyi
 
Hm, guess it does sound like ATITool is getting a bit outdated.

What would be the best way to test stability aside from ATITool then? Would just playing something graphics-intensive like Crysis be a good way to test?

Thanks!
 
Find the game that is most sensitive to overclocking.
Once you have that game stable, everything else should be too.
 
Well, I played Crysis for 40 minutes or so, ran the benchmark, ran 3DMark06, and played some TF2 without any problems. Does that suggest that it's stable enough?
 
If it does everything you need without hiccups then yes.
Dont forget to watch temps as well.
 
Yup, no hiccups at all, temps on the card maxing out below 60C, which should be fine for it I think.
 
Looks good :)
By watching the temps every so often, you can tell when the HSF needs a clean.
You may notice a crash before you notice the higher temps, so if it crashes, check the temps.
fyi
 
Hm, so I'm bringing this thread back up again because I've just noticed something with ATITool. I tried scanning for artifacts on it again, but this time using the old method (that checkbox in settings), and I'm now running almost 50 minutes without it detecting any artifacts.

Maybe there's an issue with the new method it uses since I'm perfectly stable in games, and now I haven't been able to find any artifacts with the old method?
 
You must log in or register to reply here.
Back
Top