Fixed dmmv dequant for k<= GGML_SYCL_DMMV_X

OuadiElfarouki · Sep 2, 2024 · 3ba9d04 · 3ba9d04
1 parent b60074f
commit 3ba9d04
Showing 1 changed file with 2 additions and 0 deletions.
diff --git a/ggml/src/ggml-sycl/dmmv.cpp b/ggml/src/ggml-sycl/dmmv.cpp
@@ -47,13 +47,15 @@ static void dequantize_mul_mat_vec(const void * __restrict__ vx, const dfloat *
 
     for (int i = 0; i < ncols; i += iter_stride) {
         const int col = i + vals_per_iter*tid;
+        if (col >= ncols) break; 
         const int ib = (row*ncols + col)/qk; // x block index
         const int iqs = (col%qk)/qr; // x quant index
         const int iybs = col - col%qk; // y block start index
 
 // processing >2 values per i iter is faster for fast GPUs
 #pragma unroll
         for (int j = 0; j < vals_per_iter; j += 2) {
+            if (col + j >= ncols) break;
             // process 2 vals per j iter
 
             // dequantize