diff --git a/ggml/src/ggml-zdnn/ggml-zdnn-impl.h b/ggml/src/ggml-zdnn/ggml-zdnn-impl.h
index bf9e19a8b0..626361d5c6 100644
--- a/ggml/src/ggml-zdnn/ggml-zdnn-impl.h
+++ b/ggml/src/ggml-zdnn/ggml-zdnn-impl.h
@@ -56,4 +56,15 @@ typedef unsigned long long ulong64x2_t __attribute__((vector_size(16)));
         GGML_ASSERT(status == ZDNN_OK); \
     } while (0);
 
+struct ggml_backend_zdnn_buffer {
+    void * data;
+    size_t size;
+
+    zdnn_tensor_desc pre_tfm_desc;
+    zdnn_tensor_desc tfm_desc;
+    zdnn_ztensor     ztensor;
+
+    char name[GGML_MAX_NAME];
+};
+
 #endif  // GGML_ZDNN_IMPL
diff --git a/ggml/src/ggml-zdnn/ggml-zdnn-rewrite.cpp b/ggml/src/ggml-zdnn/ggml-zdnn-rewrite.cpp
index ba01b58260..3f4c0cb2c7 100644
--- a/ggml/src/ggml-zdnn/ggml-zdnn-rewrite.cpp
+++ b/ggml/src/ggml-zdnn/ggml-zdnn-rewrite.cpp
@@ -63,6 +63,7 @@ static bool ggml_zdnn_op_mul_mat(struct ggml_backend_zdnn_context * ctx, const g
     void * bias_data = (void *)calloc(ne0, ggml_element_size(output));
     ZDNN_CHECK(zdnn_transform_ztensor(&ztensor_bias, bias_data));
 
+    std::raise(SIGINT);
     ZDNN_CHECK(zdnn_matmul_transpose_op(&inputs_extra->ztensor, &weights_extra->ztensor, &ztensor_bias,
                                         false, true, MATMUL_OP_ADDITION, &output_extra->ztensor));
     ZDNN_CHECK(zdnn_transform_ztensor(&output_extra->ztensor, output->data));
@@ -169,17 +170,6 @@ static void ggml_zdnn_free(struct ggml_backend_zdnn_context * ctx) {
     free(ctx);
 }
 
-struct ggml_backend_zdnn_buffer {
-    void * data;
-    size_t size;
-
-    zdnn_tensor_desc pre_tfm_desc;
-    zdnn_tensor_desc tfm_desc;
-    zdnn_ztensor     ztensor;
-
-    char name[GGML_MAX_NAME];
-};
-
 struct ggml_backend_zdnn_buffer_context {
     void * all_data;
     size_t all_size;