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
I got vramfs working nicely, but it continuously uses 100% CPU (one core) after I write something to the filesystem.
Here is a stack trace of the running thread:
#0 0x00007ffff6bb3a87 in sched_yield () from /usr/lib/libc.so.6
#1 0x00007ffff589a15f in ?? () from /usr/lib/libnvidia-opencl.so.1
#2 0x00007ffff59a4ca0 in ?? () from /usr/lib/libnvidia-opencl.so.1
#3 0x00007ffff6e9408c in start_thread () from /usr/lib/libpthread.so.0
#4 0x00007ffff6bcbe7f in clone () from /usr/lib/libc.so.6
The text was updated successfully, but these errors were encountered:
Hi,
the exact same thing happens to me (and OpenCL works fine with other applications).
I could see that:
The 100% of CPU usage depends on the file size and goes away after a while: if few megabytes, it is imperceptible, if 50MB, lasts a few seconds, if 512MB lasts 1 minute and a half until CPU usage drops from 100% to 0%.
This happens not only in writing, but in reading as well.
GPU usage seems low while CPU usage remains at 100% (monitored with watch -n0,5 "nvidia-settings -q GPUUtilization -q useddedicatedgpumemory")
dd returns almost immediately (less than 0.5s for 512M), although CPU usage remains high for quite some time.
I got vramfs working nicely, but it continuously uses 100% CPU (one core) after I write something to the filesystem.
Here is a stack trace of the running thread:
The text was updated successfully, but these errors were encountered: