CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-17 11:37:10 +00:00
Code Issues Packages Projects Releases Wiki Activity

Revert "sycl: add usage of enqueue_functions extension (#14244)" (#15910)

Browse Source 
* Revert "sycl: add usage of enqueue_functions extension (#14244)"

This reverts commit 8308f98c7f.

* fix missed revert code, format the code
This commit is contained in:
Neo Zhang Jianyu
2025-09-12 09:15:12 +08:00
committed by GitHub
parent 360d6533db
commit 704d90c987
20 changed files with 845 additions and 674 deletions
Download Patch File Download Diff File Expand all files Collapse all files

41
ggml/src/ggml-sycl/concat.cpp
View File

@@ -89,24 +89,33 @@ static void concat_f32_sycl(const float *x, const float *y, float *dst,
sycl::range<3> gridDim(ne2, ne1, num_blocks);
switch (dim) {
case 0:
sycl_parallel_for(stream,
sycl::nd_range<3>(gridDim * sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE),
sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE)),
[=](sycl::nd_item<3> item_ct1) { concat_f32_dim0(x, y, dst, ne0, ne00, item_ct1); });
break;
stream->parallel_for(
sycl::nd_range<3>(gridDim *
sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE),
sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE)),
[=](sycl::nd_item<3> item_ct1) {
concat_f32_dim0(x, y, dst, ne0, ne00, item_ct1);
});
break;
case 1:
sycl_parallel_for(stream,
sycl::nd_range<3>(gridDim * sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE),
sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE)),
[=](sycl::nd_item<3> item_ct1) { concat_f32_dim1(x, y, dst, ne0, ne01, item_ct1); });
break;
stream->parallel_for(
sycl::nd_range<3>(gridDim *
sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE),
sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE)),
[=](sycl::nd_item<3> item_ct1) {
concat_f32_dim1(x, y, dst, ne0, ne01, item_ct1);
});
break;
// dim >=2 will be dispatched to the default path
default:
sycl_parallel_for(stream,
sycl::nd_range<3>(gridDim * sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE),
sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE)),
[=](sycl::nd_item<3> item_ct1) { concat_f32_dim2(x, y, dst, ne0, ne02, item_ct1); });
break;
stream->parallel_for(
sycl::nd_range<3>(gridDim *
sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE),
sycl::range<3>(1, 1, SYCL_CONCAT_BLOCK_SIZE)),
[=](sycl::nd_item<3> item_ct1) {
concat_f32_dim2(x, y, dst, ne0, ne02, item_ct1);
});
break;
}
}
@@ -120,7 +129,7 @@ static void concat_f32_sycl_non_cont(
int64_t ne2, int64_t ne3, uint64_t nb0, uint64_t nb1, uint64_t nb2,
uint64_t nb3, int32_t dim) {
sycl::range<3> gridDim(ne3, ne2, ne1);
sycl_parallel_for(stream, sycl::nd_range<3>(gridDim, sycl::range<3>(1, 1, 1)), [=](sycl::nd_item<3> item_ct1) {
stream->parallel_for(sycl::nd_range<3>(gridDim, sycl::range<3>(1, 1, 1)), [=](sycl::nd_item<3> item_ct1) {
int64_t i3 = item_ct1.get_group(0);
int64_t i2 = item_ct1.get_group(1);
int64_t i1 = item_ct1.get_group(2);