diff options
| author | Nirmoy Das <nirmoy.aiemd@gmail.com> | 2020-06-25 14:07:23 +0200 |
|---|---|---|
| committer | Christian König <christian.koenig@amd.com> | 2020-06-26 14:16:29 +0200 |
| commit | d41a39dda1407ff50c7c4bfbd307c188f1ae364b (patch) | |
| tree | f668464bee4abac9f7f2f5ff445b80ad499acda0 /drivers/gpu/drm/scheduler/sched_main.c | |
| parent | 0dc9b286b8d2c219a55f6ad05d144aa4da495e97 (diff) | |
drm/scheduler: improve job distribution with multiple queues
This patch uses score to select a new drm scheduler for better
loadbalance between multiple drm schedulers instead of num_jobs.
Below are test results after running amdgpu_test for ~10 times.
Before this patch:
sched_name num of many times it got schedule
========= ==================================
sdma0 1463
sdma1 198
comp_1.0.1 280
After this patch:
sched_name num of many times it got schedule
========= ==================================
sdma0 925
sdma1 928
comp_1.0.1 177
comp_1.1.1 44
comp_1.2.1 43
comp_1.3.1 44
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/373000/
Signed-off-by: Christian König <christian.koenig@amd.com>
Diffstat (limited to 'drivers/gpu/drm/scheduler/sched_main.c')
| -rw-r--r-- | drivers/gpu/drm/scheduler/sched_main.c | 14 |
1 files changed, 8 insertions, 6 deletions
diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 8e731ed0d9d9..25a9e6911602 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -92,6 +92,7 @@ void drm_sched_rq_add_entity(struct drm_sched_rq *rq, if (!list_empty(&entity->list)) return; spin_lock(&rq->lock); + atomic_inc(&rq->sched->score); list_add_tail(&entity->list, &rq->entities); spin_unlock(&rq->lock); } @@ -110,6 +111,7 @@ void drm_sched_rq_remove_entity(struct drm_sched_rq *rq, if (list_empty(&entity->list)) return; spin_lock(&rq->lock); + atomic_dec(&rq->sched->score); list_del_init(&entity->list); if (rq->current_entity == entity) rq->current_entity = NULL; @@ -647,7 +649,7 @@ static void drm_sched_process_job(struct dma_fence *f, struct dma_fence_cb *cb) struct drm_gpu_scheduler *sched = s_fence->sched; atomic_dec(&sched->hw_rq_count); - atomic_dec(&sched->num_jobs); + atomic_dec(&sched->score); trace_drm_sched_process_job(s_fence); @@ -712,7 +714,7 @@ drm_sched_pick_best(struct drm_gpu_scheduler **sched_list, { struct drm_gpu_scheduler *sched, *picked_sched = NULL; int i; - unsigned int min_jobs = UINT_MAX, num_jobs; + unsigned int min_score = UINT_MAX, num_score; for (i = 0; i < num_sched_list; ++i) { sched = sched_list[i]; @@ -723,9 +725,9 @@ drm_sched_pick_best(struct drm_gpu_scheduler **sched_list, continue; } - num_jobs = atomic_read(&sched->num_jobs); - if (num_jobs < min_jobs) { - min_jobs = num_jobs; + num_score = atomic_read(&sched->score); + if (num_score < min_score) { + min_score = num_score; picked_sched = sched; } } @@ -860,7 +862,7 @@ int drm_sched_init(struct drm_gpu_scheduler *sched, spin_lock_init(&sched->job_list_lock); atomic_set(&sched->hw_rq_count, 0); INIT_DELAYED_WORK(&sched->work_tdr, drm_sched_job_timedout); - atomic_set(&sched->num_jobs, 0); + atomic_set(&sched->score, 0); atomic64_set(&sched->job_id_count, 0); /* Each scheduler will run on a seperate kernel thread */ |