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: add bias ztensor and data free

Browse Source 
Signed-off-by: Aaron Teo <aaron.teo1@ibm.com>
This commit is contained in:
Aaron Teo
2025-07-28 15:59:52 +08:00
parent bee7dd3020
commit f800c80281
1 changed files with 8 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
ggml/src/ggml-zdnn/ggml-zdnn-rewrite.cpp
View File

@@ -329,7 +329,14 @@ static void ggml_backend_zdnn_buffer_free_buffer(ggml_backend_buffer_t buffer) {
struct ggml_backend_zdnn_buffer_context * ctx = (struct ggml_backend_zdnn_buffer_context *)buffer->context; struct ggml_backend_zdnn_buffer_context * ctx = (struct ggml_backend_zdnn_buffer_context *)buffer->context;
for (int i = 0; i < ctx->n_buffers; i++) { for (int i = 0; i < ctx->n_buffers; i++) {
zdnn_free_ztensor_buffer(&ctx->buffers[i].ztensor); struct ggml_backend_zdnn_buffer * buf = &ctx->buffers[i];
// free any extra buffers (e.g., bias)
if (buf->extra != nullptr) {
zdnn_free_ztensor_buffer(&buf->extra->ztensor);
free(buf->extra->data);
}
zdnn_free_ztensor_buffer(&buf->ztensor);
} }
if (ctx->owned) { if (ctx->owned) {