[CI, CK examples] Disable time_kernel for CI tests and examples (#3464)

* Disable kernel timing in tests

* default time_kernel = false in old CK examples

example/02_gemm_bilinear/gemm_bilinear_wmma_fp16.cpp

@@ -119,7 +119,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 3840;

example/02_gemm_bilinear/gemm_bilinear_wmma_int8.cpp

@@ -119,7 +119,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 3840;

example/12_reduce/reduce_blockwise.cpp

@@ -31,7 +31,7 @@ class SimpleAppArgs
     bool do_verification = true;
     int data_type        = 1;
     int init_method      = 2;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     public:
     void show_usage(const char* cmd)

example/12_reduce/reduce_multiblock_atomic_add.cpp

@@ -31,7 +31,7 @@ class SimpleAppArgs
     bool do_verification = true;
     int data_type        = 1;
     int init_method      = 2;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     public:
     void show_usage(const char* cmd)

example/12_reduce/reduce_threadwise_multi_d.cpp

@@ -31,7 +31,7 @@ class SimpleAppArgs
     bool do_verification = true;
     int data_type        = 1;
     int init_method      = 2;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     public:
     void show_usage(const char* cmd)

example/13_pool2d_fwd/pool2d_fwd_fp16.cpp

@@ -53,7 +53,7 @@ int main(int argc, char* argv[])
     {
         do_verification = true;
         init_method     = 1;
-        time_kernel     = true;
+        time_kernel     = false;
     }
     else if(argc == 4)
     {

example/15_grouped_gemm/grouped_gemm_multiple_d_splitk_xdl_fp16.cpp

@@ -90,7 +90,7 @@ struct ExecutionConfig final
     bool do_verification = true;
     int init_method      = 1;
     int k_batch          = 128;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 };
 
 bool run_grouped_gemm(const ProblemSize& problem_size, const ExecutionConfig& config)

example/15_grouped_gemm/grouped_gemm_multiple_d_xdl_fp16.cpp

@@ -89,7 +89,7 @@ struct ExecutionConfig final
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 };
 
 bool run_grouped_gemm(const ProblemSize& problem_size, const ExecutionConfig& config)

example/16_gemm_multi_d_multi_reduces/gemm_add_add_mean_meansquare_xdl_fp16.cpp

@@ -268,7 +268,7 @@ int main()
         pass &= ck::utils::check_err(r1_m, r1_m_host, "Error: Incorrect results d1", 1e-2, 1e-2);
     }
 
-    bool time_kernel = true;
+    bool time_kernel = false;
     if(time_kernel)
     {
         float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});

example/16_gemm_multi_d_multi_reduces/gemm_add_addsquare_xdl_int8.cpp

@@ -302,7 +302,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 1024;

example/16_gemm_multi_d_multi_reduces/gemm_max_xdl_bf16.cpp

@@ -106,7 +106,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 1024;

example/16_gemm_multi_d_multi_reduces/gemm_max_xdl_fp16.cpp

@@ -106,7 +106,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 1024;

example/16_gemm_multi_d_multi_reduces/gemm_max_xdl_fp32.cpp

@@ -106,7 +106,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 1024;

example/16_gemm_multi_d_multi_reduces/gemm_max_xdl_int4.cpp

@@ -108,7 +108,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 1024;

example/16_gemm_multi_d_multi_reduces/gemm_max_xdl_int8.cpp

@@ -105,7 +105,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 1024;

example/16_gemm_multi_d_multi_reduces/gemm_mean_meansquare_xdl_bf16.cpp

@@ -112,7 +112,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 1024;

example/16_gemm_multi_d_multi_reduces/gemm_mean_meansquare_xdl_fp16.cpp

@@ -112,7 +112,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 1024;

example/16_gemm_multi_d_multi_reduces/gemm_mean_meansquare_xdl_fp32.cpp

@@ -112,7 +112,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M = 1024;

example/22_cgemm/cgemm_xdl_int4.cpp

@@ -81,7 +81,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // CGEMM shape
     ck::index_t M = 1024;

example/23_softmax/softmax_blockwise.cpp

@@ -65,7 +65,7 @@ class SimpleAppArgs
 
     bool do_verification = true;
     int init_method      = 2;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     public:
     void show_usage(const char* cmd)

example/24_batched_gemm/run_batched_gemm_example_fp16int4_b_scale.inc

@@ -27,7 +27,7 @@ struct ExecutionConfig final
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 };
 
 template <typename DataType>

example/29_batched_gemm_bias_e_permute/batched_gemm_bias_e_permute_wmma_fp16.cpp

@@ -248,7 +248,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     ck::index_t G0 = 1;
     ck::index_t G1 = 2;

example/30_grouped_conv_fwd_multiple_d/common.hpp

@@ -92,7 +92,7 @@ struct ExecutionConfig final
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 };
 
 #define DefaultConvParam                                                       \

example/30_grouped_conv_fwd_multiple_d/common_wmma.hpp

@@ -92,7 +92,7 @@ struct ExecutionConfig final
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 };
 
 #define DefaultConvParam                                                       \

example/33_multiple_reduce/dual_reduce_common.hpp

@@ -40,7 +40,7 @@ class SimpleAppArgs
 
     bool do_verification = true;
     int init_method      = 2;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     public:
     SimpleAppArgs()

example/35_splitK_gemm/common.hpp

@@ -44,7 +44,7 @@ struct ExecutionConfig final
 {
     bool do_verification = true;
     int init_method      = 2;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 };
 
 template <ck::index_t... Is>

example/36_sparse_embedding/sparse_embedding3_forward_layernorm.cpp

@@ -56,7 +56,7 @@ template<> struct emb_kernel<ck::half_t, 8192> { using kernel_type = DeviceInsta
 
 int main(int argc, char* argv[])
 {
-    bool time_kernel = true;
+    bool time_kernel = false;
 
     ck::index_t num_rows          = 65536;
     constexpr auto dims           = ck::Sequence<256, 512, 768, 1024, 1536, 2048, 4096, 8192>{};

example/37_batched_gemm_add_add_relu_gemm_add/batched_gemm_add_add_relu_gemm_add_xdl_fp16.cpp

@@ -195,7 +195,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t M              = 1024;

example/40_conv2d_fwd_quantization/conv2d_fwd_dl_bias_relu_perchannel_quantization_int8.cpp

@@ -86,7 +86,7 @@ using DeviceGroupedConvNDFwdInstance =
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/40_conv2d_fwd_quantization/conv2d_fwd_dl_bias_relu_perlayer_quantization_int8.cpp

@@ -84,7 +84,7 @@ using DeviceGroupedConvNDFwdInstance =
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/40_conv2d_fwd_quantization/conv2d_fwd_dl_bias_tanh_perchannel_quantization_int8.cpp

@@ -87,7 +87,7 @@ using DeviceGroupedConvNDFwdInstance =
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/40_conv2d_fwd_quantization/conv2d_fwd_dl_bias_tanh_perlayer_quantization_int8.cpp

@@ -84,7 +84,7 @@ using DeviceGroupedConvNDFwdInstance =
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/40_conv2d_fwd_quantization/conv2d_fwd_dl_perchannel_quantization_int8.cpp

@@ -84,7 +84,7 @@ using DeviceGroupedConvNDFwdInstance =
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/40_conv2d_fwd_quantization/conv2d_fwd_xdl_bias_relu_perchannel_quantization_int8.cpp

@@ -90,7 +90,7 @@ using DeviceGroupedConvNDFwdInstance =
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/40_conv2d_fwd_quantization/conv2d_fwd_xdl_bias_relu_perlayer_quantization_int8.cpp

@@ -88,7 +88,7 @@ using DeviceGroupedConvNDFwdInstance =
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/40_conv2d_fwd_quantization/conv2d_fwd_xdl_perchannel_quantization_int8.cpp

@@ -88,7 +88,7 @@ using DeviceGroupedConvNDFwdInstance =
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/42_groupnorm_fwd/run_groupnorm_fwd_example.inc

@@ -12,7 +12,7 @@ int run_groupnorm_fwd_example(int argc, char* argv[])
     ck::index_t C = 128;
 
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
     bool log_kernel      = true;
 
     if(argc == 1)

example/44_elementwise_permute/elementwise_binary_4D_fp16.cpp

@@ -53,7 +53,7 @@ using DeviceElementwisePermuteInstance = ck::tensor_operation::device::DeviceEle
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     std::vector<std::size_t> nchw = {16, 128, 32, 64};
 

example/44_elementwise_permute/elementwise_permute_4D_fp16.cpp

@@ -46,7 +46,7 @@ using DeviceElementwisePermuteInstance = ck::tensor_operation::device::DeviceEle
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/44_elementwise_permute/elementwise_permute_4D_fp16_col.cpp

@@ -50,7 +50,7 @@ using DeviceElementwisePermuteInstance = ck::tensor_operation::device::DeviceEle
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/44_elementwise_permute/elementwise_permute_4D_fp16_row.cpp

@@ -50,7 +50,7 @@ using DeviceElementwisePermuteInstance = ck::tensor_operation::device::DeviceEle
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/44_elementwise_permute/elementwise_permute_4D_fp32_col.cpp

@@ -49,7 +49,7 @@ using DeviceElementwisePermuteInstance = ck::tensor_operation::device::DeviceEle
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/44_elementwise_permute/elementwise_permute_4D_fp32_row.cpp

@@ -50,7 +50,7 @@ using DeviceElementwisePermuteInstance = ck::tensor_operation::device::DeviceEle
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/44_elementwise_permute/elementwise_scale_permute_amax_2D_fp16_fp8.cpp

@@ -121,7 +121,7 @@ void reference_scale_permute_amax(Tensor<InputDataType>& input,
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     const float scale = 2.f;
 

example/44_elementwise_permute/elementwise_trinary_4D_fp16.cpp

@@ -58,7 +58,7 @@ using DeviceElementwisePermuteInstance = ck::tensor_operation::device::DeviceEle
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     if(argc == 1)
     {

example/45_elementwise_normalization/elementwise_layernorm_blockwise.cpp

@@ -84,7 +84,7 @@ void host_elementwise2D(HostTensorC& C,
 int main(int argc, char* argv[])
 {
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     ck::index_t M = 48 * 256;
     ck::index_t N = 1024;

example/65_gemm_multiply_multiply/moe_gemm1_xdl_fp8.cpp

@@ -205,7 +205,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // GEMM shape
     ck::index_t N               = 4096;

example/65_gemm_multiply_multiply/moe_gemm1_xdl_fp8_blockscale.cpp

@@ -193,7 +193,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 #if 1
     // GEMM shape
     ck::index_t N       = 4096;

example/65_gemm_multiply_multiply/moe_gemm1_xdl_pk_i4.cpp

@@ -194,7 +194,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // per expert:
     // GEMM shape

example/65_gemm_multiply_multiply/moe_gemm2_xdl_fp8.cpp

@@ -185,7 +185,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // per expert:
     // GEMM shape

example/65_gemm_multiply_multiply/moe_gemm2_xdl_fp8_blockscale.cpp

@@ -188,7 +188,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // tokens = 1
     // topk = 1

example/65_gemm_multiply_multiply/moe_gemm2_xdl_pk_i4.cpp

@@ -164,7 +164,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // per expert:
     // GEMM shape

example/67_gemm_microscaling/moe_gemm1_xdl_mx_fp4.cpp

@@ -178,7 +178,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // per expert:
     // GEMM shape

example/67_gemm_microscaling/moe_gemm1_xdl_mx_fp4_bns.cpp

@@ -178,7 +178,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // per expert:
     // GEMM shape

example/67_gemm_microscaling/moe_gemm1_xdl_mx_fp4_bpreshuffle.cpp

@@ -208,7 +208,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // per expert:
     // GEMM shape

example/67_gemm_microscaling/moe_gemm2_xdl_mx_fp4.cpp

@@ -171,7 +171,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // per expert:
     // GEMM shape

example/67_gemm_microscaling/moe_gemm2_xdl_mx_fp4_bns.cpp

@@ -171,7 +171,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // per expert:
     // GEMM shape

example/67_gemm_microscaling/moe_gemm2_xdl_mx_fp4_bpreshuffle.cpp

@@ -204,7 +204,7 @@ int main(int argc, char* argv[])
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
 
     // per expert:
     // GEMM shape

test/batched_gemm_multi_d/test_batched_gemm_multi_d_dl.cpp

@@ -61,7 +61,7 @@ class TestBatchedGemmMultiD : public ::testing::Test
                 true,  // do_verification
                 1,     // init_method
                 false, // do_log
-                1,     // time_kernel,
+                false, // time_kernel,
                 M,
                 N,
                 K,

test/gemm/gemm_standalone_xdl_fp16.cpp

@@ -104,7 +104,7 @@ int main(int argc, char* argv[])
     };
 
     bool do_verification = true;
-    bool time_kernel     = true;
+    bool time_kernel     = false;
     int problem_index    = -1;
 
     if(argc == 1)

test/wrapper/test_wrapper_gemm_xdl.cpp

@@ -306,7 +306,7 @@ void PerformGemm(const ck::index_t M,
 
     const auto kernel =
         DeviceGemm<DataType, GemmTraits, scalar_per_vector, BlockShape, ThreadLayout, DoPadding>;
-    const float avg_time = launch_and_time_kernel(StreamConfig{nullptr, true},
+    const float avg_time = launch_and_time_kernel(StreamConfig{nullptr, false},
                                                   kernel,
                                                   dim3(grid_size_x, grid_size_y, 1),
                                                   dim3(ck::wrapper::size(thread_layout)),