CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-16 11:27:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

sycl: add usage of enqueue_functions extension (#14244)

Browse Source 
* Add header and namespace to use enqueue_functions extension

* Convert submit and parallel_for to use new extension in convert.cpp

* Convert submit and parallel_for to use extension in ggml-sycl.cpp

* Convert submit and parallel_for to use extension in gla.cpp

* Convert submit and parallel_for in mmq.cpp

* Convert submit and parallel_for in mmvq.cpp

* Convert submit and parallel_for in remaining files

* Convert all simple parallel_for to nd_launch from enqueue_functions
extension

* Wrapping extension in general function

Create a general function that enable the enqueue_functions extension if
it is enable in the compiler, otherwise call the general SYCL function
to launch kernels.

---------

Signed-off-by: nscipione <nicolo.scipione@codeplay.com>
This commit is contained in:
Nicolò Scipione
2025-06-20 15:07:21 +02:00
committed by GitHub
parent 6369be0735
commit 8308f98c7f
19 changed files with 750 additions and 986 deletions
Download Patch File Download Diff File Expand all files Collapse all files

44
ggml/src/ggml-sycl/rope.cpp
View File

@@ -235,20 +235,22 @@ static void rope_norm_sycl(const T * x, T * dst, const int ne0, const int ne1, c
the limit. To get the device limit, query
info::device::max_work_group_size. Adjust the work-group size if needed.
*/
stream->parallel_for(sycl::nd_range<3>(block_nums * block_dims, block_dims), [=](sycl::nd_item<3> item_ct1) {
rope_norm<T, false>(x, dst, ne0, ne1, s1, s2, n_dims, pos, freq_scale, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors, item_ct1);
});
sycl_parallel_for(stream, sycl::nd_range<3>(block_nums * block_dims, block_dims),
[=](sycl::nd_item<3> item_ct1) {
rope_norm<T, false>(x, dst, ne0, ne1, s1, s2, n_dims, pos, freq_scale, ext_factor,
attn_factor, corr_dims, theta_scale, freq_factors, item_ct1);
});
} else {
/*
DPCT1049:41: The work-group size passed to the SYCL kernel may exceed
the limit. To get the device limit, query
info::device::max_work_group_size. Adjust the work-group size if needed.
*/
stream->parallel_for(sycl::nd_range<3>(block_nums * block_dims, block_dims), [=](sycl::nd_item<3> item_ct1) {
rope_norm<T, true>(x, dst, ne0, ne1, s1, s2, n_dims, pos, freq_scale, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors, item_ct1);
});
sycl_parallel_for(stream, sycl::nd_range<3>(block_nums * block_dims, block_dims),
[=](sycl::nd_item<3> item_ct1) {
rope_norm<T, true>(x, dst, ne0, ne1, s1, s2, n_dims, pos, freq_scale, ext_factor,
attn_factor, corr_dims, theta_scale, freq_factors, item_ct1);
});
}
}
@@ -267,15 +269,17 @@ static void rope_neox_sycl(const T * x, T * dst, const int ne0, const int ne1, c
dpct::has_capability_or_fail(stream->get_device(), { sycl::aspect::fp16 });
if (freq_factors == nullptr) {
stream->parallel_for(sycl::nd_range<3>(block_nums * block_dims, block_dims), [=](sycl::nd_item<3> item_ct1) {
rope_neox<T, false>(x, dst, ne0, ne1, s1, s2, n_dims, pos, freq_scale, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors, item_ct1);
});
sycl_parallel_for(stream, sycl::nd_range<3>(block_nums * block_dims, block_dims),
[=](sycl::nd_item<3> item_ct1) {
rope_neox<T, false>(x, dst, ne0, ne1, s1, s2, n_dims, pos, freq_scale, ext_factor,
attn_factor, corr_dims, theta_scale, freq_factors, item_ct1);
});
} else {
stream->parallel_for(sycl::nd_range<3>(block_nums * block_dims, block_dims), [=](sycl::nd_item<3> item_ct1) {
rope_neox<T, true>(x, dst, ne0, ne1, s1, s2, n_dims, pos, freq_scale, ext_factor, attn_factor, corr_dims,
theta_scale, freq_factors, item_ct1);
});
sycl_parallel_for(stream, sycl::nd_range<3>(block_nums * block_dims, block_dims),
[=](sycl::nd_item<3> item_ct1) {
rope_neox<T, true>(x, dst, ne0, ne1, s1, s2, n_dims, pos, freq_scale, ext_factor,
attn_factor, corr_dims, theta_scale, freq_factors, item_ct1);
});
}
}
@@ -298,12 +302,12 @@ static void rope_multi_sycl(const T * x, T * dst, const int ne0, const int ne1,
}
// launch kernel
if (freq_factors == nullptr) {
stream->parallel_for(nd_range, [=](sycl::nd_item<3> item_ct1) {
sycl_parallel_for(stream, nd_range, [=](sycl::nd_item<3> item_ct1) {
rope_multi<T, false>(x, dst, ne0, ne1, ne2, s1, s2, n_dims, pos, freq_scale, ext_factor, attn_factor,
corr_dims, theta_scale, freq_factors, sections, item_ct1);
});
} else {
stream->parallel_for(nd_range, [=](sycl::nd_item<3> item_ct1) {
sycl_parallel_for(stream, nd_range, [=](sycl::nd_item<3> item_ct1) {
rope_multi<T, true>(x, dst, ne0, ne1, ne2, s1, s2, n_dims, pos, freq_scale, ext_factor, attn_factor,
corr_dims, theta_scale, freq_factors, sections, item_ct1);
});
@@ -333,12 +337,12 @@ static void rope_vision_sycl(const T * x, T * dst, const int ne0, const int ne1,
}
// launch kernel
if (freq_factors == nullptr) {
stream->parallel_for(nd_range, [=](sycl::nd_item<3> item_ct1) {
sycl_parallel_for(stream, nd_range, [=](sycl::nd_item<3> item_ct1) {
rope_vision<T, false>(x, dst, ne0, ne1, ne2, s1, s2, n_dims, pos, freq_scale, ext_factor, attn_factor,
corr_dims, theta_scale, freq_factors, sections, item_ct1);
});
} else {
stream->parallel_for(nd_range, [=](sycl::nd_item<3> item_ct1) {
sycl_parallel_for(stream, nd_range, [=](sycl::nd_item<3> item_ct1) {
rope_vision<T, true>(x, dst, ne0, ne1, ne2, s1, s2, n_dims, pos, freq_scale, ext_factor, attn_factor,
corr_dims, theta_scale, freq_factors, sections, item_ct1);
});