Update ggml/src/ggml-cuda/cpy.cu

Co-authored-by: Johannes Gäßler <[email protected]>

Author: gcp

ggml/src/ggml-cuda/cpy.cu

@@ -85,7 +85,8 @@ static __device__ void cpy_blck_f32_q8_0(const char * cxi, char * cdsti) {
 static __device__ void cpy_blck_q8_0_f32(const char * cxi, char * cdsti) {
     float * cdstf = (float *)(cdsti);
 
-    for (int j = 0; j < QK8_0; j+=2) {
+#pragma unroll
+    for (int j = 0; j < QK8_0; j += 2) {
         float2 dq;
         dequantize_q8_0(cxi, 0, j, dq);
         *(cdstf + j) = dq.x;