CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-08 10:07:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

ggml-zdnn: fix missing load tensor

Browse Source 
Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>
This commit is contained in:
Aaron Teo
2025-07-28 23:42:24 +08:00
parent fc9260deab
commit 2cfa118fa9
1 changed files with 9 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
ggml/src/ggml-zdnn/ggml-zdnn-rewrite.cpp
View File

@@ -47,6 +47,11 @@ inline void ggml_zdnn_create_tensor(zdnn_tensor_desc & pre_tfm_desc,
ZDNN_CHECK(zdnn_init_ztensor_with_malloc(&pre_tfm_desc, &tfm_desc, &ztensor));
}
inline void ggml_zdnn_load_tensor(zdnn_ztensor & ztensor,
void * buffer) {
ZDNN_CHECK(zdnn_transform_ztensor(&ztensor, buffer));
}
static void ggml_zdnn_mul_mat_op(ggml_backend_zdnn_context * ctx, const ggml_tensor * src0, const ggml_tensor * src1, ggml_tensor * dst) {
GGML_TENSOR_BINARY_OP_LOCALS;
@@ -98,7 +103,10 @@ static void ggml_zdnn_mul_mat_op(ggml_backend_zdnn_context * ctx, const ggml_ten
ggml_zdnn_create_tensor(ptd_output, td_output, zt_output, output, output_dim, ZDNN_2D);
void * bias_data = (void *)calloc(ne0, ggml_element_size(output));
ZDNN_CHECK(zdnn_transform_ztensor(&zt_bias, bias_data));
ggml_zdnn_load_tensor(zt_weights, weights->data);
ggml_zdnn_load_tensor(zt_inputs, inputs->data);
ggml_zdnn_load_tensor(zt_bias, bias_data);
ggml_zdnn_load_tensor(zt_output, output->data);
ZDNN_CHECK(zdnn_matmul_transpose_op(&zt_inputs, &zt_weights, &zt_bias,
false, true, MATMUL_OP_ADDITION, &zt_output));