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

[doc]shfl*sync update

Browse Source 
1. Updated FAQ with shft*sync not supported hip_faq.md
2. Corrected some of input parameter description in hcc_details/hip_runtime_api.h
3. Redirect shfl*() to shfl_*_sync() for nvcc path where CUDA > 9.0

Change-Id: I3d8184db5fcc622852c9bad96b706348e8dfc16c


[ROCm/hip commit: 83b11f9a61]
This commit is contained in:
Sarbojit Sarkar
2020-05-20 06:19:31 -04:00
parent 72a611f9e3
commit 587471a46c
3 changed files with 31 additions and 19 deletions
Download Patch File Download Diff File Expand all files Collapse all files
projects/hip/docs/markdown/hip_faq.md
+4 -1
View File
@@ -239,4 +239,7 @@ $HIP/include/hip/hcc_detail/host_defines.h
Or pass "-DGENERIC_GRID_LAUNCH=0" to hipcc at application compilation time.
### What is maximum limit of Generic Grid Launch parameters (grid and block)?
Product of (grid.x and block.x), (grid.y and block.y) or (grid.z and block.z) should always be less than UINT_MAX.
Product of (grid.x and block.x), (grid.y and block.y) or (grid.z and block.z) should always be less than UINT_MAX.
### Are __shfl_*_sync functions supported on HIP platform?
__shfl_*_sync is not supported on HIP but for nvcc path CUDA 9.0 and above all shuffle calls get redirected to it's sync version.