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

SWDEV-508776 - support VGPRs validation (#274)

Browse Source 
Clarify some VGPRs terms description.
Fix some wrong query logics of availableVGPRs_ and
availableRegistersPerCU_ in device info.
Add hipDeviceAttributeMaxAvailableVgprsPerThread
attribute query.
Remove hardcoding of following
 info_.vgprAllocGranularity_
 info_.vgprsPerSimd_
This commit is contained in:
Sang, Tao
2025-07-09 10:46:52 -04:00
committed by GitHub
parent 71c6535b07
commit 397f303d97
7 changed files with 32 additions and 93 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rocclr/device/devkernel.cpp
+1 -1
View File
@@ -619,7 +619,7 @@ Kernel::Kernel(const amd::Device& dev, const std::string& name, const Program& p
workGroupInfo_.usedGPRs_ = 0;
workGroupInfo_.availableSGPRs_ = 0;
workGroupInfo_.usedSGPRs_ = 0;
workGroupInfo_.availableVGPRs_ = 0;
workGroupInfo_.availableVGPRs_ = dev.info().availableVGPRs_;
workGroupInfo_.usedVGPRs_ = 0;
workGroupInfo_.availableLDSSize_ = 0;
workGroupInfo_.usedLDSSize_ = 0;