public-mirrors/linux
1
0
Fork 0
mirror of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git synced 2025-11-18 04:35:11 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

drm/amdkfd: Fix signal handling performance again

Browse source 
It turns out that idr_for_each_entry is really slow compared to just
iterating over the slots. Based on measurements the difference is
estimated to be about a factor 64. That means using idr_for_each_entry
is only worth it with very few allocated events.

Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
This commit is contained in:
Felix Kuehling 2018-05-01 17:56:08 -04:00 • committed by Oded Gabbay
parent f8ea72d097
commit eeb27b7eb3
1 changed files with 1 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
drivers/gpu/drm/amd/amdkfd/kfd_events.c
View file

@ -496,7 +496,7 @@ void kfd_signal_event_interrupt(unsigned int pasid, uint32_t partial_id,
pr_debug_ratelimited("Partial ID invalid: %u (%u valid bits)\n", pr_debug_ratelimited("Partial ID invalid: %u (%u valid bits)\n",
partial_id, valid_id_bits); partial_id, valid_id_bits);
if (p->signal_event_count < KFD_SIGNAL_EVENT_LIMIT/2) { if (p->signal_event_count < KFD_SIGNAL_EVENT_LIMIT / 64) {
/* With relatively few events, it's faster to /* With relatively few events, it's faster to
* iterate over the event IDR * iterate over the event IDR
*/ */