diff --git a/modules/sdnq/triton_mm.py b/modules/sdnq/triton_mm.py
index beda09f03..955761e88 100644
--- a/modules/sdnq/triton_mm.py
+++ b/modules/sdnq/triton_mm.py
@@ -12,16 +12,31 @@ import torch
 import triton
 import triton.language as tl
 
+try:
+    from .common import is_rdna2_and_older
+except Exception:
+    is_rdna2_and_older = False
 
-matmul_configs = [
-    triton.Config({'BLOCK_SIZE_M': BM, 'BLOCK_SIZE_N': BN, "BLOCK_SIZE_K": BK, "GROUP_SIZE_M": GM}, num_warps=w, num_stages=s)
-    for BM in [32, 64, 128, 256]
-    for BN in [32, 64, 128, 256]
-    for BK in [32, 64, 128]
-    for GM in [4, 8]
-    for w in [4, 8]
-    for s in [2]
-]
+if is_rdna2_and_older:
+    matmul_configs = [
+        triton.Config({'BLOCK_SIZE_M': BM, 'BLOCK_SIZE_N': BN, "BLOCK_SIZE_K": BK, "GROUP_SIZE_M": GM}, num_warps=w, num_stages=s)
+        for BM in [64, 128]
+        for BN in [64, 128]
+        for BK in [64]
+        for GM in [2, 4]
+        for w in [2, 4]
+        for s in [2]
+    ]
+else:
+    matmul_configs = [
+        triton.Config({'BLOCK_SIZE_M': BM, 'BLOCK_SIZE_N': BN, "BLOCK_SIZE_K": BK, "GROUP_SIZE_M": GM}, num_warps=w, num_stages=s)
+        for BM in [32, 64, 128, 256]
+        for BN in [32, 64, 128, 256]
+        for BK in [32, 64, 128]
+        for GM in [4, 8]
+        for w in [4, 8]
+        for s in [2]
+    ]
 
 
 @triton.autotune(configs=matmul_configs, key=["M", "N", "K", "stride_bk", "ACCUMULATOR_DTYPE"], cache_results=True)