test freq

example/32_batched_gemm_scale_softmax_gemm/batched_gemm_scale_softmax_gemm_xdl_fp16.cpp

@@ -160,6 +160,8 @@ int main(int argc, char* argv[])
     ck::index_t BatchStrideC  = -1;
     float alpha               = 1;
 
+    int nrepeat = 1;
+
     if(argc == 1)
     {
         // use default case
@@ -183,7 +185,7 @@ int main(int argc, char* argv[])
 
         BatchCount = std::stoi(argv[8]);
     }
-    else if(argc == 18)
+    else if(argc == 19)
     {
         do_verification = std::stoi(argv[1]);
         init_method     = std::stoi(argv[2]);
@@ -207,6 +209,8 @@ int main(int argc, char* argv[])
         BatchStrideC  = std::stoi(argv[16]);
 
         alpha = std::stof(argv[17]);
+
+        nrepeat = std::stoi(argv[18]);
     }
     else
     {
@@ -348,6 +352,8 @@ int main(int argc, char* argv[])
         return 0;
     }
 
+    for(int i = 0; i < nrepeat; ++i)
+    {
         float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
 
         std::size_t flop      = (size_t(M) * N * K * 2 + size_t(M) * N * O * 2) * BatchCount;
@@ -359,8 +365,9 @@ int main(int argc, char* argv[])
 
         float gb_per_sec = num_btype / 1.E6 / ave_time;
 
-    std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s, "
-              << gemm.GetTypeString() << std::endl;
+        std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec
+                  << " GB/s, " << gemm.GetTypeString() << std::endl;
+    }
 
     c_g_m_o_device_buf.FromDevice(c_g_m_o_device_result.mData.data());