portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2026-09-15 05:53:17 +01:00
Code Releases Activity

Fix alignment on StencilEntry struct for HIP

Browse Source 
CUDA neatly aligns the StencilEntry struct to 16-bytes, which is not used on HIP.
Mapping the StencilEntry struct to 16-bytes on HIP yields around 5% better performance
on MI355X-class GPUs.
This commit is contained in:
Julio Maia
2026-09-11 10:56:02 -05:00
parent 75bb6bdb11
commit 487ec8d410
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Grid/stencil/Stencil.h
+1 -1
View File
@@ -82,7 +82,7 @@ void DslashLogPartial(void);
void DslashLogDirichlet(void); void DslashLogDirichlet(void);
struct StencilEntry { struct StencilEntry {
#ifdef GRID_CUDA #if defined(GRID_CUDA) || defined(GRID_HIP)
uint64_t _byte_offset; // 8 bytes uint64_t _byte_offset; // 8 bytes
uint32_t _offset; // 4 bytes uint32_t _offset; // 4 bytes
#else #else