add reviewers comments

example/ck_tile/03_gemm/instances/gemm_universal_mem_f16_f16_f16_mk_nk_mn.cpp

@@ -6,22 +6,6 @@
 using Row = ck_tile::tensor_layout::gemm::RowMajor;
 using Col = ck_tile::tensor_layout::gemm::ColumnMajor;
 
-template float gemm_<trait_<ck_tile::half_t,
-                            ck_tile::half_t,
-                            float,
-                            ck_tile::half_t,
-                            Row,
-                            Col,
-                            Row,
-                            128,
-                            32,
-                            64,
-                            4,
-                            1,
-                            1,
-                            32,
-                            32,
-                            8,
-                            false,
-                            false,
-                            false>>(const A&, const S&);
+// clang-format off
+template float gemm_<gemm_traits_<ck_tile::half_t, ck_tile::half_t, float, ck_tile::half_t, Row, Col, Row, 128, 32, 64, 4, 1, 1, 32, 32, 8, false, false, false>>(const A&, const S&);
+// clang-format on

example/ck_tile/03_gemm/instances/gemm_universal_mem_instance_common.hpp

@@ -2,50 +2,11 @@
 // Copyright (c) 2025, Advanced Micro Devices, Inc. All rights reserved.
 #include <ck_tile/core.hpp>
 #include <iostream>
-#include "gemm_basic.hpp"
+#include "gemm.hpp"
 
 using A = ck_tile::GemmHostArgs;
 using S = ck_tile::stream_config;
 
-template <typename ADataType_,
-          typename BDataType_,
-          typename AccDataType_,
-          typename CDataType_,
-          typename ALayout_,
-          typename BLayout_,
-          typename CLayout_,
-          ck_tile::index_t M_Tile_,
-          ck_tile::index_t N_Tile_,
-          ck_tile::index_t K_Tile_,
-          ck_tile::index_t M_Warp_,
-          ck_tile::index_t N_Warp_,
-          ck_tile::index_t K_Warp_,
-          ck_tile::index_t M_Warp_Tile_,
-          ck_tile::index_t N_Warp_Tile_,
-          ck_tile::index_t K_Warp_Tile_,
-          bool kPadM_,
-          bool kPadN_,
-          bool kPadK_>
-using trait_ = gemm_traits_<ADataType_,
-                            BDataType_,
-                            AccDataType_,
-                            CDataType_,
-                            ALayout_,
-                            BLayout_,
-                            CLayout_,
-                            M_Tile_,
-                            N_Tile_,
-                            K_Tile_,
-                            M_Warp_,
-                            N_Warp_,
-                            K_Warp_,
-                            M_Warp_Tile_,
-                            N_Warp_Tile_,
-                            K_Warp_Tile_,
-                            kPadM_,
-                            kPadN_,
-                            kPadK_>;
-
 template <typename Traits_>
 float gemm_(const ck_tile::GemmHostArgs& args, const ck_tile::stream_config& s)
 {

example/ck_tile/03_gemm/run_gemm_example.inc

@@ -2,6 +2,29 @@
 // Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
 #pragma once
 
+auto create_args(int argc, char* argv[])
+{
+    ck_tile::ArgParser arg_parser;
+    arg_parser.insert("m", "3840", "m dimension")
+        .insert("n", "4096", "n dimension")
+        .insert("k", "2048", "k dimension")
+        .insert("a_layout", "R", "A tensor data layout - Row by default")
+        .insert("b_layout", "R", "B tensor data layout - Row by default")
+        .insert("c_layout", "R", "C tensor data layout - Row by default")
+        .insert("stride_a", "0", "Tensor A stride")
+        .insert("stride_b", "0", "Tensor B stride")
+        .insert("stride_c", "0", "Tensor C stride")
+        .insert("v", "2", "0. No validation, 1. Validation on CPU, 2. Validation on GPU")
+        .insert("prec", "fp16", "data type. fp16/bf16/fp8/bf8")
+        .insert("warmup", "50", "number of iterations before benchmark the kernel")
+        .insert("repeat", "100", "number of iterations to benchmark the kernel")
+        .insert("timer", "gpu", "gpu:gpu timer, cpu:cpu timer")
+        .insert("split_k", "1", "splitK value");
+
+    bool result = arg_parser.parse(argc, argv);
+    return std::make_tuple(result, arg_parser);
+}
+
 template <typename ALayout, typename BLayout, typename CLayout>
 float invoke_gemm(ck_tile::DeviceMem& a_m_k_dev_buf,
                   ck_tile::DeviceMem& b_k_n_dev_buf,
@@ -28,6 +51,7 @@ float invoke_gemm(ck_tile::DeviceMem& a_m_k_dev_buf,
     args.stride_B = stride_B;
     args.stride_C = stride_C;
 
+    // TODO: Change datatypes in future to allow mixed precision gemms!
     gemm_traits traits{DataTypeTraits<ADataType>{}.name,
                        std::is_same_v<ALayout, Row>,
                        std::is_same_v<BLayout, Row>,

example/ck_tile/03_gemm/universal_gemm.cpp

@@ -10,30 +10,7 @@
 #include <tuple>
 
 #include "ck_tile/host.hpp"
-#include "gemm_basic.hpp"
-
-auto create_args(int argc, char* argv[])
-{
-    ck_tile::ArgParser arg_parser;
-    arg_parser.insert("m", "3840", "m dimension")
-        .insert("n", "4096", "n dimension")
-        .insert("k", "2048", "k dimension")
-        .insert("a_layout", "R", "A tensor data layout - Row by default")
-        .insert("b_layout", "R", "B tensor data layout - Row by default")
-        .insert("c_layout", "R", "C tensor data layout - Row by default")
-        .insert("stride_a", "0", "Tensor A stride")
-        .insert("stride_b", "0", "Tensor B stride")
-        .insert("stride_c", "0", "Tensor C stride")
-        .insert("v", "2", "0. No validation, 1. Validation on CPU, 2. Validation on GPU")
-        .insert("prec", "fp16", "data type. fp16/bf16/fp8/bf8")
-        .insert("warmup", "50", "number of iterations before benchmark the kernel")
-        .insert("repeat", "100", "number of iterations to benchmark the kernel")
-        .insert("timer", "gpu", "gpu:gpu timer, cpu:cpu timer")
-        .insert("split_k", "1", "splitK value");
-
-    bool result = arg_parser.parse(argc, argv);
-    return std::make_tuple(result, arg_parser);
-}
+#include "gemm.hpp"
 
 #include "run_gemm_example.inc"