diff options
| author | Samuel Pitoiset <[email protected]> | 2019-07-12 12:17:11 +0200 |
|---|---|---|
| committer | Samuel Pitoiset <[email protected]> | 2019-07-12 17:47:11 +0200 |
| commit | e510c5ee3b8c7f0d1d9afff28760469f43c24c02 (patch) | |
| tree | 9935211693651396e3f4e6f589db5690533c8d32 /src/amd/common | |
| parent | 5f4f8aec748427462bb4539dcfbe0e622842327a (diff) | |
ac: import ac_get_compute_resource_limits() from RadeonSI
Signed-off-by: Samuel Pitoiset <[email protected]>
Reviewed-by: Bas Nieuwenhuizen <[email protected]>
Diffstat (limited to 'src/amd/common')
| -rw-r--r-- | src/amd/common/ac_gpu_info.c | 32 | ||||
| -rw-r--r-- | src/amd/common/ac_gpu_info.h | 4 |
2 files changed, 36 insertions, 0 deletions
diff --git a/src/amd/common/ac_gpu_info.c b/src/amd/common/ac_gpu_info.c index 596a9ebe508..a501d840b25 100644 --- a/src/amd/common/ac_gpu_info.c +++ b/src/amd/common/ac_gpu_info.c @@ -895,3 +895,35 @@ ac_get_harvested_configs(struct radeon_info *info, } } } + +unsigned ac_get_compute_resource_limits(struct radeon_info *info, + unsigned waves_per_threadgroup, + unsigned max_waves_per_sh, + unsigned threadgroups_per_cu) +{ + unsigned compute_resource_limits = + S_00B854_SIMD_DEST_CNTL(waves_per_threadgroup % 4 == 0); + + if (info->chip_class >= GFX7) { + unsigned num_cu_per_se = info->num_good_compute_units / + info->max_se; + + /* Force even distribution on all SIMDs in CU if the workgroup + * size is 64. This has shown some good improvements if # of CUs + * per SE is not a multiple of 4. + */ + if (num_cu_per_se % 4 && waves_per_threadgroup == 1) + compute_resource_limits |= S_00B854_FORCE_SIMD_DIST(1); + + assert(threadgroups_per_cu >= 1 && threadgroups_per_cu <= 8); + compute_resource_limits |= S_00B854_WAVES_PER_SH(max_waves_per_sh) | + S_00B854_CU_GROUP_COUNT(threadgroups_per_cu - 1); + } else { + /* GFX6 */ + if (max_waves_per_sh) { + unsigned limit_div16 = DIV_ROUND_UP(max_waves_per_sh, 16); + compute_resource_limits |= S_00B854_WAVES_PER_SH_SI(limit_div16); + } + } + return compute_resource_limits; +} diff --git a/src/amd/common/ac_gpu_info.h b/src/amd/common/ac_gpu_info.h index c42548f8352..3ec3e44d665 100644 --- a/src/amd/common/ac_gpu_info.h +++ b/src/amd/common/ac_gpu_info.h @@ -167,6 +167,10 @@ void ac_get_harvested_configs(struct radeon_info *info, unsigned raster_config, unsigned *cik_raster_config_1_p, unsigned *raster_config_se); +unsigned ac_get_compute_resource_limits(struct radeon_info *info, + unsigned waves_per_threadgroup, + unsigned max_waves_per_sh, + unsigned threadgroups_per_cu); static inline unsigned ac_get_max_simd_waves(enum radeon_family family) { |