Merge branch 'develop' into amd-develop

profiler/src/profile_gemm_multiply_multiply.cpp

@@ -76,7 +76,7 @@ int profile_gemm_multiply_multiply(int argc, char* argv[])
     int n_warmup      = 1;
     int n_iter        = 10;
     uint64_t rotating = 0;
-    if(argc == 18)
+    if(argc == 19)
     {
         n_warmup = std::stoi(argv[16]);
         n_iter   = std::stoi(argv[17]);

test/smfmac_op/smfmac_op_xdl.cpp

@@ -13,6 +13,7 @@
 #include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 #include "test/smfmac_op/smfmac_op_util.hpp"
+#include "ck/host_utility/device_prop.hpp"
 
 using BF16        = ck::bhalf_t;
 using F16         = ck::half_t;
@@ -39,6 +40,8 @@ class TestSmfmac : public ::testing::Test
     void Run()
     {
         bool pass = true;
+        if(ck::get_device_name() == "gfx942")
+        {
             constexpr auto matmul_default = ck::smfmac_op_util::matmul<Src1Type,
                                                                        Src1VecSize,
                                                                        Src2Type,
@@ -52,7 +55,8 @@ class TestSmfmac : public ::testing::Test
 
             constexpr auto smfmac_kernel_container = std::make_tuple(matmul_default);
 
-        ck::static_for<0, std::tuple_size_v<decltype(smfmac_kernel_container)>, 1>{}([&](auto i) {
+            ck::static_for<0, std::tuple_size_v<decltype(smfmac_kernel_container)>, 1>{}(
+                [&](auto i) {
                     pass &= ck::smfmac_op_util::TestSmfmac<
                         std::tuple_element_t<i.value, decltype(smfmac_kernel_container)>,
                         Src1Type,
@@ -71,7 +75,7 @@ class TestSmfmac : public ::testing::Test
                         N,
                         K>{}(std::get<ck::Number<i>{}>(smfmac_kernel_container));
                 });
-
+        }
         EXPECT_TRUE(pass);
     }
 };