It's more accurate.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3975>
*
* LDS is 128KB in WGP mode and 64KB in CU mode. Assume the WGP mode is used.
*/
- info->lds_size_per_cu = info->chip_class >= GFX10 ? 128 * 1024 : 64 * 1024;
+ info->lds_size_per_workgroup = info->chip_class >= GFX10 ? 128 * 1024 : 64 * 1024;
info->lds_granularity = info->chip_class >= GFX7 ? 128 * 4 : 64 * 4;
assert(util_is_power_of_two_or_zero(dma.available_rings + 1));
printf(" tcc_cache_line_size = %u\n", info->tcc_cache_line_size);
printf(" tcc_harvested = %u\n", info->tcc_harvested);
printf(" pc_lines = %u\n", info->pc_lines);
- printf(" lds_size_per_cu = %u\n", info->lds_size_per_cu);
+ printf(" lds_size_per_workgroup = %u\n", info->lds_size_per_workgroup);
printf(" lds_granularity = %i\n", info->lds_granularity);
printf(" max_memory_clock = %i\n", info->max_memory_clock);
printf(" ce_ram_size = %i\n", info->ce_ram_size);
uint32_t tcc_cache_line_size;
bool tcc_harvested;
unsigned pc_lines;
- uint32_t lds_size_per_cu;
+ uint32_t lds_size_per_workgroup;
uint32_t lds_granularity;
uint32_t max_memory_clock;
uint32_t ce_ram_size;
chunk->vram_size = rad_info->vram_size;
chunk->l2_cache_size = rad_info->l2_cache_size;
chunk->l1_cache_size = rad_info->l1_cache_size;
- chunk->lds_size = rad_info->lds_size_per_cu;
+ chunk->lds_size = rad_info->lds_size_per_workgroup;
strncpy(chunk->gpu_name, device->physical_device->name, SQTT_GPU_NAME_MAX_SIZE);
device->physical_device->rad_info.num_physical_wave64_vgprs_per_simd / vgprs);
}
- unsigned max_lds_per_simd = device->physical_device->rad_info.lds_size_per_cu / device->physical_device->rad_info.num_simd_per_compute_unit;
+ unsigned max_lds_per_simd = device->physical_device->rad_info.lds_size_per_workgroup / device->physical_device->rad_info.num_simd_per_compute_unit;
if (lds_per_wave)
max_simd_waves = MIN2(max_simd_waves, max_lds_per_simd / lds_per_wave);
max_simd_waves = MIN2(max_simd_waves, max_vgprs / conf->num_vgprs);
}
- unsigned max_lds_per_simd = sscreen->info.lds_size_per_cu / 4;
+ unsigned max_lds_per_simd = sscreen->info.lds_size_per_workgroup / 4;
if (lds_per_wave)
max_simd_waves = MIN2(max_simd_waves, max_lds_per_simd / lds_per_wave);