HIP: enable WMMA-MMQ INT kernels for RDNA 3 (#17576)

* enabled wmma instructions for most quantizations other than q2k * fixed the last q2_k test case failure * address comments: fix out of bound write for RDNA4, add comments after #endif * clean up rebase: fix ne error in half2 * fix the EditorConfig CI
2025-12-05 09:17:37 +01:00
parent 03d9a77b85
commit 668ed76574
4 changed files with 94 additions and 29 deletions
@@ -1542,8 +1542,10 @@ static __device__ __forceinline__ void vec_dot_q2_K_q8_1_mma(
            tile_C Cm;
            if (k01 >= MMQ_TILE_NE_K * 3/4) {
                tile_A A1;
-                A1.x[0] = 0x01010101;
-                A1.x[1] = 0x01010101;
+#pragma unroll
+                for (int l = 0; l < tile_A::ne; ++l) {
+                    A1.x[l] = 0x01010101;
+                }
                mma(Cm, A1, B);
            }