Since I switched back to R196.34, the artifacts have disappeared, even under hefty load…
Conclusion: R196.75 are really crappy and dangerous.
Now the question: how can we trust the WHQL certification ????
Yesterday, I suddenly got artifacts on FluidMark (the graphics card was a GeForce GTS 250).
That happened just after a CPU intensive test with PhysX. And after those artifacts were also present on other OpenGL apps like GPU Caps Viewer like in this OpenGL 3 demo:
Maybe the multithreaded test I did with FluidMark has been a little bit too much intensive and maybe the NVIDIA driver has been corrupted. I don’t know. The GPU temperature was low (40°C) even when the 3D app was running so I think those artifacts are not related to a GPU overheating. Anyway, a reboot was enough to get rid of artifacts.
I hope these artifacts are not related to the buggy ForceWare 195.76 because I currently use this driver on my dev-station. But my GTS 250 has a Scythe VGA cooler that is not controlled by the card (then is not affected by the bug) so both fans of the VGA cooler work all the time…
7 thoughts on “[BUG] Artifacts Caused by PhysX? Nope, They are Caused by R196.75 (Update)”
Sometimes i see the same artefacts. But i do not use PhysiX. It happens, if you use the SceniX Raytracer based on OptiX. If your application crashes because of some internal bugs, not enough GPU Memory, etc, it will be like that.
It seems that the GPU memory has corrupted. I have seen this artifact but I haven’t used PhysX. Reboot the computer may help.
Yeah, drivers will kill your GPU, PhysX will kill you software, and then Nvidia wil come and kill yoouuuu !
JeGX, I prefer less provoking titles 🙂
I have seen those artifacts too and they had nothing to do with PhysX. Keep an eye on your GPU and get *rid* of the 195.76 drivers.
Regarding heating, it’s not just your GPU temperature that can result in a card going bad. A bad driver can cause any number of other components to fail.
Not to sound alarmist, but most Nvidia GPUs have thermal throttling, so the GPU (as in the GPU chip itself) will slow down in case it overheats and in worst case will BSOD to protect itself. The bug that you reported earlier in the driver must be due to some other component overheating. From the looks of it from the image on your site, it does. So the GPU temp reading might just show a cool 40 deg while something else gets fried.
Again I am not saying it is the reason, but it’s best to get rid of the drivers asap, maybe using a drivercleaner won’t be such a bad idea.
I gonna reinstall R196.34…
The R196.34 breaks some thing too, like Optix.
Basically, any modern NVIDIA driver has problems…
I’m afraid the 196.75 just fried your card 😀
No need for Furmark now!
Comments are closed.