ver4a/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

SYCL: Refactor and enable FP16 in binary broadcast OPs (#12975)

Browse source 
* SYCL: refactor move to a separate file

* Fix binbcast

* Remove duplicates

* fix include formatting

* fix typo
This commit is contained in:
Akarshan Biswas 2025-04-18 19:27:56 +05:30 • committed by GitHub
parent b9154ecff9
commit 8d66005763
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
7 changed files with 393 additions and 372 deletions
Download patch file Download diff file Expand all files Collapse all files

47
ggml/src/ggml-sycl/element_wise.cpp
View file

@ -1261,27 +1261,6 @@ inline void ggml_sycl_op_acc(ggml_backend_sycl_context & ctx, ggml_tensor *dst)
}
inline void ggml_sycl_op_add(ggml_backend_sycl_context & ctx, ggml_tensor *dst) {
ggml_sycl_op_bin_bcast<bin_bcast_sycl<op_add>>(ctx, dst->src[0], dst->src[1], dst);
}
inline void ggml_sycl_op_sub(ggml_backend_sycl_context & ctx, ggml_tensor *dst) {
ggml_sycl_op_bin_bcast<bin_bcast_sycl<op_sub>>(ctx, dst->src[0], dst->src[1], dst);
}
inline void ggml_sycl_op_mul(ggml_backend_sycl_context & ctx, ggml_tensor *dst) {
ggml_sycl_op_bin_bcast<bin_bcast_sycl<op_mul>>(ctx, dst->src[0], dst->src[1], dst);
}
inline void ggml_sycl_op_div(ggml_backend_sycl_context & ctx, ggml_tensor *dst) {
ggml_sycl_op_bin_bcast<bin_bcast_sycl<op_div>>(ctx, dst->src[0], dst->src[1], dst);
}
void ggml_sycl_sqrt(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
GGML_SYCL_DEBUG("call %s: DST Tensor type: %s\n", __func__, ggml_type_name(dst->type));
ggml_sycl_op_sqrt(ctx, dst);
@ -1409,29 +1388,3 @@ void ggml_sycl_clamp(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
GGML_SYCL_DEBUG("call %s done\n", __func__);
}
void ggml_sycl_add(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
GGML_SYCL_DEBUG("call %s\n", __func__);
ggml_sycl_op_add(ctx, dst);
GGML_SYCL_DEBUG("call %s done\n", __func__);
}
void ggml_sycl_sub(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
GGML_SYCL_DEBUG("call %s\n", __func__);
ggml_sycl_op_sub(ctx, dst);
GGML_SYCL_DEBUG("call %s done\n", __func__);
}
void ggml_sycl_mul(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
GGML_SYCL_DEBUG("call %s\n", __func__);
ggml_sycl_op_mul(ctx, dst);
GGML_SYCL_DEBUG("call %s done\n", __func__);
}
void ggml_sycl_div(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
GGML_SYCL_DEBUG("call %s\n", __func__);
ggml_sycl_op_div(ctx, dst);
GGML_SYCL_DEBUG("call %s done\n", __func__);
}