ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-19 22:39:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

[CK][EXAMPLES] (#2826)

Browse Source 
-Added parameter to enable/disable verification and timing of kernel in various examples that missed it.
-Added parameter to change number of groups to execute in grouped_gemm_examples.

Signed-off-by: Michal Kulikowski <Michal.Kulikowski@amd.com>
This commit is contained in:
Michał Kulikowski
2025-09-11 21:33:00 +02:00
committed by GitHub
parent f3239395dc
commit ffe9775e70
33 changed files with 578 additions and 116 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
example/39_permute/permute_1xHxW_fp16.cpp
View File

@@ -17,4 +17,23 @@ using DevicePermuteInstance = ck::tensor_operation::device::DevicePermuteImpl
#include "run_permute_element_example.inc"
int main() { return !run_permute_element_example({1, 32000, 80}, {0, 2, 1}); }
int main(int argc, char* argv[])
{
bool time_kernel = false;
if(argc == 1)
{
// use default
}
else if(argc == 2)
{
time_kernel = std::stoi(argv[1]);
}
else
{
printf("arg1: time kernel (0=no, 1=yes, default=0)\n");
exit(0);
}
return !run_permute_element_example({1, 32000, 80}, {0, 2, 1}, time_kernel);
}

21
example/39_permute/permute_HxWx4_fp16.cpp
View File

@@ -19,4 +19,23 @@ using DevicePermuteInstance = ck::tensor_operation::device::DevicePermuteImpl
#include "run_permute_bundle_example.inc"
int main() { return !run_permute_bundle_example({1, 80, 32000}, {0, 2, 1}); }
int main(int argc, char* argv[])
{
bool time_kernel = false;
if(argc == 1)
{
// use default
}
else if(argc == 2)
{
time_kernel = std::stoi(argv[1]);
}
else
{
printf("arg1: time kernel (0=no, 1=yes, default=0)\n");
exit(0);
}
return !run_permute_bundle_example({1, 80, 32000}, {0, 2, 1}, time_kernel);
}

21
example/39_permute/permute_NxHxW_fp16.cpp
View File

@@ -17,4 +17,23 @@ using DevicePermuteInstance = ck::tensor_operation::device::DevicePermuteImpl
#include "run_permute_element_example.inc"
int main() { return !run_permute_element_example({121, 768, 80}, {0, 2, 1}); }
int main(int argc, char* argv[])
{
bool time_kernel = false;
if(argc == 1)
{
// use default
}
else if(argc == 2)
{
time_kernel = std::stoi(argv[1]);
}
else
{
printf("arg1: time kernel (0=no, 1=yes, default=0)\n");
exit(0);
}
return !run_permute_element_example({121, 768, 80}, {0, 2, 1}, time_kernel);
}

10
example/39_permute/run_permute_bundle_example.inc
View File

@@ -3,7 +3,7 @@
#pragma once
bool run_permute_bundle(const Problem& problem)
bool run_permute_bundle(const Problem& problem, bool time_kernel)
{
const auto& input_bundle_shape = problem.shape;
const auto& input_bundle_axes = problem.axes;
@@ -41,7 +41,7 @@ bool run_permute_bundle(const Problem& problem)
};
auto invoker = permute.MakeInvoker();
float ave_time = invoker.Run(argument, StreamConfig{nullptr, true});
float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
std::cout << "Perf: " << ave_time << " ms" << std::endl;
@@ -72,7 +72,9 @@ bool run_permute_bundle(const Problem& problem)
1e-6);
}
bool run_permute_bundle_example(const Problem::Shape& shape, const Problem::Axes& axes)
bool run_permute_bundle_example(const Problem::Shape& shape,
const Problem::Axes& axes,
bool time_kernel)
{
return run_permute_bundle(Problem{shape, axes});
return run_permute_bundle(Problem{shape, axes}, time_kernel);
}

10
example/39_permute/run_permute_element_example.inc
View File

@@ -3,7 +3,7 @@
#pragma once
bool run_permute_element(const Problem& problem)
bool run_permute_element(const Problem& problem, bool time_kernel)
{
const auto& input_shape = problem.shape;
const auto& input_axes = problem.axes;
@@ -40,7 +40,7 @@ bool run_permute_element(const Problem& problem)
};
auto invoker = permute.MakeInvoker();
float ave_time = invoker.Run(argument, StreamConfig{nullptr, true});
float ave_time = invoker.Run(argument, StreamConfig{nullptr, time_kernel});
std::cout << "Perf: " << ave_time << " ms" << std::endl;
@@ -59,7 +59,9 @@ bool run_permute_element(const Problem& problem)
1e-6);
}
bool run_permute_element_example(const Problem::Shape& shape, const Problem::Axes& axes)
bool run_permute_element_example(const Problem::Shape& shape,
const Problem::Axes& axes,
bool time_kernel)
{
return run_permute_element(Problem{shape, axes});
return run_permute_element(Problem{shape, axes}, time_kernel);
}