Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by Deleted member 255716, Jun 29, 2017.
completely breaks SLI+surround.
Do not trust too much the driver about resource binding tier... Tier 2 has several binding limitations (14 only CBV per shader stage, 64 UAV slots on all stages, and no "un-popoulated" root signature slots) which were all established for valid reasons (especially the first and the latter limitations, maybe the UAV slots limit could be bypassed by the driver). Even if with current games such limitations aren't a big deal, they are strictly related to NV hardware.
That's not the first time a driver tells a wrong cap-bit value.
Good for Fermi if they finally decided to public a working DX12 driver..
Is there any way to check yet the status for Shader Model 6 features yet? I'd really like news about SM 6.0 support on Kepler and 6.1 for Paxwell
SM 6.0 requires experimental mode to be enabled (which is not available for the common user), SM 6.1 is not available yet (and I am not sure NV supports barycentric semantic).
well Watch dogs2 crashes at start up with this driver
Oh look another garbage 38x.xx driver! Yay!
Every driver after 378.92 causes BSOD when watching videos via browser with my 1080Ti... looks like I will give this one a miss, since even IF it fixed my issue, it brought others with it. And I just bought Watch Dogs 2... :bang:
Has NVidia ever publically disclosed that Maxwell and Pascal are not fully bindless?
Are you using Chrome? Chrome's hardware acceleration seems to have a problem with NVidia's drivers, or vice versa.
Yeah, I disabled it yesterday and so far, so good. At first the BSODs were maybe 1-2/week, but with the last driver (the one before this) it went insane, pretty much at least 1 BSOD/day...
Someone on another forum I frequent had this problem. He fixed it by doing a clean install. It's possible some files might be corrupt on your windows install.
At any rate, Chrome's hardware acceleration is mediocre compared to Microsoft Edge. Edge has the best hardware acceleration of any browser by a long shot.
It's 99% driver related, if I roll back to 378.92 I'll have no BSODs even with Hardware Acceleration enabled.
Does this problem occur with Edge or other browsers? Although Edge is my primary browser, I use Chrome from time to time, and I've never had a problem with it regardless of driver set.
im still getting the issue with the WU recently sad times
I'm pretty sure it happens with any browser while having hardware acceleration on. I had reports of this same issue from a guy using Edge. Also, it's always with a 1080Ti.
NVIDIA never publicly disclose explicitly it's hardware architecture details and implementation, however some APIs capabilities and support can be read directly from CUDA virtual machine implementations.
All NVIDIA GPUs have a special cache buffer, where the driver puts the constant buffers (CBV, "constant buffer view" in DirectX jargon). NVIDIA constant buffer hardware is about 64-KB. Since under DirectX a CB is always limited to 4096-bytes (4-KB), the constant buffer limit under NV archtiecture is still 16 (precisely, 14 available to the programmer, 2 reserved for internal usage).
Just this made all current NV hardware a "tier 2" on resource binding tiers. Please note this is not a huge deal for current and upcoming games.
Just checked my idle clocks now with 382.53 and its 135MHz, its always been like that here..
Installing these now.
EDIT: still 135MHz with this release.
well guys i think the lower idle clocks is a placebo or some effect of the driver install.....
i dont know if it was due to rebooting system or installing the latest insider build over my previous but my idle clocks are back to normal again on my 1080ti`s 240//202 instead of 139/202 like before lol
Maxwell could always do that. Factory OCed Pascal cards were not reaching 135. Now they do.
No, i leave it at optimal power.
Still having driver crashes in both FF 55b5 and Phase Shift (rhythm game) with these.