From 5c07cd0e15ec009f7c6616cf2f58db0915a73b59 Mon Sep 17 00:00:00 2001
From: Jiming Ruan <Jiming.Ruan@amd.com>
Date: Thu, 19 Jun 2025 04:17:20 +0000
Subject: [PATCH] A workaround for the case that get_slice_tile() doesn't work.

---
 .../gemm/block/block_gemm_areg_bsmem_creg_v2.hpp   | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/include/ck_tile/ops/gemm/block/block_gemm_areg_bsmem_creg_v2.hpp b/include/ck_tile/ops/gemm/block/block_gemm_areg_bsmem_creg_v2.hpp
index 0181c0eec8..66a08ccd18 100644
--- a/include/ck_tile/ops/gemm/block/block_gemm_areg_bsmem_creg_v2.hpp
+++ b/include/ck_tile/ops/gemm/block/block_gemm_areg_bsmem_creg_v2.hpp
@@ -1,5 +1,5 @@
 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2024, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2025, Advanced Micro Devices, Inc. All rights reserved.
 
 #pragma once
 
@@ -61,11 +61,11 @@ struct BlockGemmARegBSmemCRegV2
 
         constexpr auto c_block_outer_dstr_encoding = tile_distribution_encoding<
             sequence<>,
-            tuple<sequence<MIterPerWarp, MWarp>, sequence<NIterPerWarp, NWarp>>,
+            tuple<sequence<MIterPerWarp, MWarp>, sequence<NWarp, NIterPerWarp>>,
             tuple<sequence<1, 2>>,
-            tuple<sequence<1, 1>>,
+            tuple<sequence<1, 0>>,
             sequence<1, 2>,
-            sequence<0, 0>>{};
+            sequence<0, 1>>{};
 
         constexpr auto c_block_dstr_encode = detail::make_embed_tile_distribution_encoding(
             c_block_outer_dstr_encoding, typename WG::CWarpDstrEncoding{});
@@ -214,11 +214,11 @@ struct BlockGemmARegBSmemCRegV2
 
         constexpr auto c_block_outer_dstr_encoding = tile_distribution_encoding<
             sequence<>,
-            tuple<sequence<MIterPerWarp, MWarp>, sequence<NIterPerWarp, NWarp>>,
+            tuple<sequence<MIterPerWarp, MWarp>, sequence<NWarp, NIterPerWarp>>,
             tuple<sequence<1, 2>>,
-            tuple<sequence<1, 1>>,
+            tuple<sequence<1, 0>>,
             sequence<1, 2>,
-            sequence<0, 0>>{};
+            sequence<0, 1>>{};
 
         constexpr auto c_block_dstr_encode = detail::make_embed_tile_distribution_encoding(
             c_block_outer_dstr_encoding, typename WG::CWarpDstrEncoding{});