AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

rocr: Add all sysfs entries for L2 Cache

Browse Source 
For L2 Cache and above, we report the total amount of cache for the
whole partition, so we add up the L2 Cache entry for each partition.


[ROCm/ROCR-Runtime commit: fc561ff37a]
This commit is contained in:
David Yat Sin
2025-05-20 18:13:01 +00:00
committed by Gu, Tim
parent 8f7c7458aa
commit d2982b797a
1 changed files with 8 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/rocr-runtime/runtime/hsa-runtime/core/runtime/amd_gpu_agent.cpp
+8 -1
View File
@@ -1430,7 +1430,14 @@ hsa_status_t GpuAgent::GetInfo(hsa_agent_info_t attribute, void* value) const {
const size_t num_cache = cache_props_.size();
for (size_t i = 0; i < num_cache; ++i) {
const uint32_t line_level = cache_props_[i].CacheLevel;
if (reinterpret_cast<uint32_t*>(value)[line_level - 1] == 0)
/*
* L1 Cache is per CU.
* For L2 Cache and above, we report total for the partition so we sum
* all the node entries.
*/
if (line_level >= 2)
reinterpret_cast<uint32_t*>(value)[line_level - 1] += cache_props_[i].CacheSize * 1024;
else if (reinterpret_cast<uint32_t*>(value)[line_level - 1] == 0)
reinterpret_cast<uint32_t*>(value)[line_level - 1] = cache_props_[i].CacheSize * 1024;
}
} break;