portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-15 02:05:37 +00:00
Code Releases Activity

New software moves cl::sycl

Browse Source
This commit is contained in:
Peter Boyle 2024-10-10 22:03:04 +00:00
parent ec1395a304
commit 68f112d576
2 changed files with 12 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
Grid/lattice/Lattice_reduction_sycl.h
View File

@ -16,11 +16,11 @@ inline typename vobj::scalar_objectD sumD_gpu_tensor(const vobj *lat, Integer os
Integer nsimd= vobj::Nsimd();
{
sycl::buffer<sobj, 1> abuff(&ret, {1});
theGridAccelerator->submit([&](cl::sycl::handler &cgh) {
auto Reduction = cl::sycl::reduction(abuff,cgh,identity,std::plus<>());
cgh.parallel_for(cl::sycl::range<1>{osites},
theGridAccelerator->submit([&](sycl::handler &cgh) {
auto Reduction = sycl::reduction(abuff,cgh,identity,std::plus<>());
cgh.parallel_for(sycl::range<1>{osites},
Reduction,
[=] (cl::sycl::id<1> item, auto &sum) {
[=] (sycl::id<1> item, auto &sum) {
auto osite = item[0];
sum +=Reduce(lat[osite]);
});
@ -75,11 +75,11 @@ template<class Word> Word svm_xor(Word *vec,uint64_t L)
Word ret = 0;
{
sycl::buffer<Word, 1> abuff(&ret, {1});
theGridAccelerator->submit([&](cl::sycl::handler &cgh) {
auto Reduction = cl::sycl::reduction(abuff,cgh,identity,std::bit_xor<>());
cgh.parallel_for(cl::sycl::range<1>{L},
theGridAccelerator->submit([&](sycl::handler &cgh) {
auto Reduction = sycl::reduction(abuff,cgh,identity,std::bit_xor<>());
cgh.parallel_for(sycl::range<1>{L},
Reduction,
[=] (cl::sycl::id<1> index, auto &sum) {
[=] (sycl::id<1> index, auto &sum) {
sum ^=vec[index];
});
});

8
Grid/lattice/Lattice_slicesum_core.h
View File

@ -141,11 +141,11 @@ inline void sliceSumReduction_sycl_small(const vobj *Data,
});
for (int r = 0; r < rd; r++) {
theGridAccelerator->submit([&](cl::sycl::handler &cgh) {
auto Reduction = cl::sycl::reduction(&mysum[r],std::plus<>());
cgh.parallel_for(cl::sycl::range<1>{subvol_size},
theGridAccelerator->submit([&](sycl::handler &cgh) {
auto Reduction = sycl::reduction(&mysum[r],std::plus<>());
cgh.parallel_for(sycl::range<1>{subvol_size},
Reduction,
[=](cl::sycl::id<1> item, auto &sum) {
[=](sycl::id<1> item, auto &sum) {
auto s = item[0];
sum += rb_p[r*subvol_size+s];
});