Skip to content

SYCL: set extras only on GGML_TYPE_Q4_0 #12366

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Mar 17, 2025
Merged

SYCL: set extras only on GGML_TYPE_Q4_0 #12366

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
6de75cd
SYCL: set extras only on GGML_TYPE_Q4_0
qnixsynapse Mar 13, 2025
0cee320
release tensor_extras in reset buffer interface
qnixsynapse Mar 14, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
29 changes: 22 additions & 7 deletions ggml/src/ggml-sycl/ggml-sycl.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -333,10 +333,11 @@ ggml_backend_sycl_buffer_init_tensor(ggml_backend_buffer_t buffer,
assert(tensor->view_src->buffer->buft == buffer->buft);
return GGML_STATUS_SUCCESS;
}

ggml_tensor_extra_gpu * extra = new ggml_tensor_extra_gpu{};
tensor->extra = extra;
ctx->tensor_extras.push_back(extra); //used to release it when destroy ctx.
if (tensor->type == GGML_TYPE_Q4_0) {
ggml_tensor_extra_gpu * extra = new ggml_tensor_extra_gpu{};
tensor->extra = extra;
ctx->tensor_extras.push_back(extra); //used to release it when destroy ctx.
}

if (ggml_is_quantized(tensor->type)) {
// initialize padding to 0 to avoid possible NaN values
Expand Down Expand Up @@ -486,6 +487,22 @@ catch (sycl::exception const &exc) {
std::exit(1);
}

static void ggml_backend_sycl_buffer_reset(ggml_backend_buffer_t buffer) {
GGML_SYCL_DEBUG("[SYCL] call %s\n", __func__);
if (buffer == nullptr) {
return;
}

ggml_backend_sycl_buffer_context * ctx = (ggml_backend_sycl_buffer_context *) buffer->context;

if (ctx != nullptr) {
for (ggml_tensor_extra_gpu * extra : ctx->tensor_extras) {
release_extra_gpu(extra);
}
ctx->tensor_extras.clear(); // reset the tensor_extras vector
}
}

static const ggml_backend_buffer_i ggml_backend_sycl_buffer_interface = {
/* .free_buffer = */ ggml_backend_sycl_buffer_free_buffer,
/* .get_base = */ ggml_backend_sycl_buffer_get_base,
Expand All @@ -495,7 +512,7 @@ static const ggml_backend_buffer_i ggml_backend_sycl_buffer_interface = {
/* .get_tensor = */ ggml_backend_sycl_buffer_get_tensor,
/* .cpy_tensor = */ ggml_backend_sycl_buffer_cpy_tensor,
/* .clear = */ ggml_backend_sycl_buffer_clear,
/* .reset = */ NULL,
/* .reset = */ ggml_backend_sycl_buffer_reset,
};

// sycl buffer type
Expand Down Expand Up @@ -576,7 +593,6 @@ ggml_backend_buffer_type_t ggml_backend_sycl_buffer_type(int device) {
static std::mutex mutex;
std::lock_guard<std::mutex> lock(mutex);

GGML_SYCL_DEBUG("[SYCL] call ggml_backend_sycl_buffer_type\n");

auto dev_count = ggml_backend_sycl_get_device_count();

Expand Down Expand Up @@ -3761,7 +3777,6 @@ bool ggml_backend_is_sycl(ggml_backend_t backend) {
}

int ggml_backend_sycl_get_device_count() {
GGML_SYCL_DEBUG("[SYCL] call ggml_backend_sycl_get_device_count\n");
return ggml_sycl_info().device_count;
}

Expand Down
Loading