diff --git a/example/65_gemm_multiply_multiply/CMakeLists.txt b/example/65_gemm_multiply_multiply/CMakeLists.txt
index 5d2a097576..8d51d43c65 100644
--- a/example/65_gemm_multiply_multiply/CMakeLists.txt
+++ b/example/65_gemm_multiply_multiply/CMakeLists.txt
@@ -7,7 +7,7 @@ add_example_executable(example_gemm_multiply_multiply_xdl_int8 gemm_multiply_mul
 add_example_executable(example_moe_gemm1_xdl_fp8 moe_gemm1_xdl_fp8.cpp)
 add_example_executable(example_moe_gemm2_xdl_fp8 moe_gemm2_xdl_fp8.cpp)
 
-list(APPEND gpu_list gfx942)
+list(APPEND gpu_list gfx942 gfx950)
 set(target 0)
 foreach(gpu IN LISTS GPU_TARGETS)
     if(gpu IN_LIST gpu_list AND target EQUAL 0)
diff --git a/example/65_gemm_multiply_multiply/moe_gemm2_xdl_pk_i4.cpp b/example/65_gemm_multiply_multiply/moe_gemm2_xdl_pk_i4.cpp
index b9621cc9b3..3745e3d0af 100644
--- a/example/65_gemm_multiply_multiply/moe_gemm2_xdl_pk_i4.cpp
+++ b/example/65_gemm_multiply_multiply/moe_gemm2_xdl_pk_i4.cpp
@@ -281,7 +281,7 @@ int main(int argc, char* argv[])
         break;
     case 4:
         a0_t_k_k.GenerateTensorValue(GeneratorTensor_1<A0DataType>{});
-        b0_e_n_k.GenerateTensorValue(GeneratorTensor_2<A0DataType>{-2, 2});
+        b0_e_n_k.GenerateTensorValue(GeneratorTensor_2<B0DataType>{-2, 2});
         d0_t_n.GenerateTensorValue(GeneratorTensor_1<D0DataType>{});
         d1_e_n.GenerateTensorValue(GeneratorTensor_1<D1DataType>{});
         d2_e_n.GenerateTensorValue(GeneratorTensor_1<D2DataType>{});