How accurate is the CPU usage? What's the temporal resolution of the CPU usage data? Since at 60fps frame times are 16ms, even a 1ms spike can significantly decrease the fps if the GPU has to wait for it to pass. The CPU usage statistics look quite coarse to me which leads me to believe that it is already some kind of moving average filtered signal. This does not really tell us much in terms of CPU limits to performance. 30% or 50% CPU usage should not make any difference to the framerate. It's the split second (single thread) spike that might cause the performance to degrade.
Announcement
Collapse
No announcement yet.
The Current CPU Driver Usage Difference Between RADV/RadeonSI & NVIDIA
Collapse
X
-
Originally posted by dwagner View PostAdmittedly off-topic, but does somebody know why https://cgit.freedesktop.org/~agd5f/linux/ says "No repositories found" right now? Repo moved, site down/hacked, or other reasons?
- Likes 2
Comment
-
Hopefully it's not hosted on the same server as lkml
Test signature
- Likes 2
Comment
-
..or they have some server configuration to do (mod_rewrite?) as it works with:
Comment
-
Originally posted by dwagner View PostAdmittedly off-topic, but does somebody know why https://cgit.freedesktop.org/~agd5f/linux/ says "No repositories found" right now? Repo moved, site down/hacked, or other reasons?
Downtime was _short_, at least here from Germany.
Comment
-
Something to think about (bridgman?):
Dave's latest radv (hash table patch).
cpupower govenor results
[Mesa-dev] [PATCH] radv/winsys: replace bo list searchs with a hash table.
Greetings,
Dieter
BTW "radv: don't emit unneeded vertex state." landed.
Comment
Comment