From 0fc2f528e071b1bfc9369ce822aef25ec8a1a6f3 Mon Sep 17 00:00:00 2001
From: Rostyslav Geyyer <rosty.geyyer@amd.com>
Date: Wed, 30 Apr 2025 20:03:27 +0000
Subject: [PATCH] Fix conflicts

---
 include/ck/utility/amd_xdlops.hpp | 1 +
 test/mx_mfma_op/mx_mfma_op.hpp    | 4 ++--
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/include/ck/utility/amd_xdlops.hpp b/include/ck/utility/amd_xdlops.hpp
index fc2528c4d0..2e3f8f0091 100644
--- a/include/ck/utility/amd_xdlops.hpp
+++ b/include/ck/utility/amd_xdlops.hpp
@@ -653,6 +653,7 @@ struct intrin_mfma_scale_f32_16x16x128f8f6f4<16, 16>
 #endif
     }
 
+    template <class FloatC>
     __device__ static void Run(const bf8x32_t& reg_a,
                                const int32_t& scale_a,
                                const bf8x32_t& reg_b,
diff --git a/test/mx_mfma_op/mx_mfma_op.hpp b/test/mx_mfma_op/mx_mfma_op.hpp
index 45a3deabe2..a0f130f3b4 100644
--- a/test/mx_mfma_op/mx_mfma_op.hpp
+++ b/test/mx_mfma_op/mx_mfma_op.hpp
@@ -981,10 +981,10 @@ matmul(const AType* a, const ScaleType* xa, const BType* b, const ScaleType* xb,
     auto fragXb  = BScaleFragT{};
 
     // Load the inputs.
-    fragA = load_mx_A_row_major<AType, AFragT, ScaleType, ScaleFragT, BLOCK_M, BLOCK_K, BLOCK_X>(
+    fragA = load_mx_A_row_major<AType, AFragT, ScaleType, AScaleFragT, BLOCK_M, BLOCK_K, BLOCK_X>(
         a, xa, fragXa);
 
-    fragB = load_mx_B_col_major<BType, BFragT, ScaleType, ScaleFragT, BLOCK_K, BLOCK_N, BLOCK_X>(
+    fragB = load_mx_B_col_major<BType, BFragT, ScaleType, BScaleFragT, BLOCK_K, BLOCK_N, BLOCK_X>(
         b, xb, fragXb);
 
     // Scaled Matrix multiply-accumulate using MFMA units