You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Upload of a reasonably minimal complete main.rs file that demonstrates the issue: vulkano/examples/src/bin/triangle.rs
Device: Laptop (Dell XPS 15 9570)
I launched the triangle.rs example and it locked up my machine.
I then googled around for this issue, and followed the advice in #923 by calling wait on the future after signal_fence_and_flush(), but the issue remained.
I then added some timing code to it and found that it was running as fast as 3378 fps.
Can this library be made to work with Nvidia drivers on Linux?
The text was updated successfully, but these errors were encountered:
I just tried to repro this on another machine, a desktop rather than a laptop using a GTX 1070 instead of a GTX 1050, but with the same versions of the same drivers and operating system.
Without the call to wait() after signal_fence_and_flush(), it monopolized one core of the CPU (utilizing 25%) without locking up the whole machine. After calling wait() after signal_fence_and_flush(), it now uses 1-2% of the CPU time.
This sounds definitely like a bug in the Nvidia driver. At least with the wait call you should never get a lockup and of course you shouldn't without the wait. Maybe point Nvidia towards this issue?
main.rs
file that demonstrates the issue:vulkano/examples/src/bin/triangle.rs
I launched the
triangle.rs
example and it locked up my machine.I then googled around for this issue, and followed the advice in #923 by calling wait on the future after
signal_fence_and_flush()
, but the issue remained.I then added some timing code to it and found that it was running as fast as 3378 fps.
Can this library be made to work with Nvidia drivers on Linux?
The text was updated successfully, but these errors were encountered: