portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2026-08-29 22:09:35 +01:00
Code Releases Activity

Lattice_reduction_gpu: demote timing logs to Debug, disable by default

Browse Source 
skills/mpi-heterogeneous: add Bug Class 4 for Frontier GTL/libamdhip64 ABI mismatch

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Peter Boyle
2026-05-21 12:34:30 -04:00
co-authored by Claude Sonnet 4.6
parent 32654db366
commit 7803580aa6
2 changed files with 50 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Grid/lattice/Lattice_reduction_gpu.h
+4 -4
View File
@@ -198,7 +198,7 @@ __global__ void reduceKernel(const vobj *lat, sobj *buffer, Iterator n) {
// Possibly promote to double and sum
/////////////////////////////////////////////////////////////////////////////////////////////////////////
#define GRID_REDUCTION_TIMING
#undef GRID_REDUCTION_TIMING
template <class vobj>
inline typename vobj::scalar_objectD sumD_gpu_small(const vobj *lat, Integer osites)
@@ -230,7 +230,7 @@ inline typename vobj::scalar_objectD sumD_gpu_small(const vobj *lat, Integer osi
acceleratorCopyFromDevice(buffer_v,&result,sizeof(result));
#ifdef GRID_REDUCTION_TIMING
t_d2h += usecond();
std::cout << GridLogMessage << " sumD_gpu_small"
std::cout << GridLogDebug << " sumD_gpu_small"
<< " sizeof(sobj)=" << sizeof(sobj)
<< " blocks=" << numBlocks << " threads=" << numThreads
<< " kernel+barrier=" << t_kernel << " us"
@@ -362,7 +362,7 @@ inline void sumD_gpu_reduce_words(const vobj *lat, Integer osites,
acceleratorCopyFromDevice(buffer_v, &result, sizeof(result));
#ifdef GRID_REDUCTION_TIMING
t_d2h += usecond();
std::cout << GridLogMessage << " sumD_gpu_reduce_words R=" << R
std::cout << GridLogDebug << " sumD_gpu_reduce_words R=" << R
<< " base=" << base
<< " kernel=" << t_kernel << " D2H=" << t_d2h << " us" << std::endl;
#endif
@@ -391,7 +391,7 @@ inline typename vobj::scalar_objectD sumD_gpu_large(const vobj *lat, Integer osi
while (w < words) { sumD_gpu_reduce_words< 1>(lat, osites, ret_p, w); w += 1; }
#ifdef GRID_REDUCTION_TIMING
t_large += usecond();
std::cout << GridLogMessage << "sumD_gpu_large"
std::cout << GridLogDebug << "sumD_gpu_large"
<< " sizeof(sobjD)=" << sizeof(sobjD)
<< " words=" << words << " total=" << t_large << " us" << std::endl;
#endif