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

ggml : upgrade init_tensor API to return a ggml_status (#11854)

Browse source 
* Upgrade init_tensor API to return a ggml_status

To prepare for an 'abort-free' ggml
(ggml not to abort on OOMs but return a OOM status),
as agreeed with Diego in the ggml repo,
upgrade the init_tensor() and view_init() APIs
to return a ggml_status.

* misc fixes

---------

Co-authored-by: slaren <slarengh@gmail.com>
This commit is contained in:
William Tambellini 2025-02-28 05:41:47 -08:00 • committed by GitHub
parent c43a3e7996
commit 70680c48e5
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
16 changed files with 136 additions and 97 deletions
Download patch file Download diff file Expand all files Collapse all files

5
ggml/src/ggml-cann/ggml-cann.cpp
View file

@ -796,11 +796,11 @@ static bool need_transform(ggml_type type) {
* @param buffer The CANN buffer from which to initialize the tensor.
* @param tensor Pointer to the tensor to be initialized.
*/
static void ggml_backend_cann_buffer_init_tensor(
static enum ggml_status ggml_backend_cann_buffer_init_tensor(
ggml_backend_buffer_t buffer, ggml_tensor* tensor) {
if (tensor->view_src != NULL && tensor->view_offs == 0) {
GGML_ASSERT(tensor->view_src->buffer->buft == buffer->buft);
return;
return GGML_STATUS_SUCCESS;
}
// TODO: can backend doesn't support quantized yet. Just leave the code
@ -817,6 +817,7 @@ static void ggml_backend_cann_buffer_init_tensor(
memset_size, 0, memset_size));
}
}
return GGML_STATUS_SUCCESS;
}
// TODO: need handle tensor which has paddings.