On Mon, Nov 02, 2015 at 10:28:53AM +0100, Steinar H. Gunderson wrote: >> vaSyncSurface() just calls drm_intel_bo_wait_rendering() to make sure >> all GPU operations with the surface are finished. >> drm_intel_bo_wait_rendering() is implemented with SET_DOMAIN ioctl(), I >> don't think it waits with busylooping. > Then why does it show up in perf?
I also noticed that gen6_ring_get_seqno uses some CPU, again from vaSyncSurface() -> (something) -> drm_intel_gem_bo_start_gtt_access -> drmIoctl. /* Steinar */ -- Homepage: https://www.sesse.net/ _______________________________________________ Libva mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/libva
