ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-17 00:58:44 +00:00
Code Issues Packages Projects Releases Wiki Activity

1s cold and 1s warm kernel run for mi300 perf measurement

Browse Source
This commit is contained in:
Harisankar Sadasivan
2024-04-15 13:29:13 -04:00
parent dd34ab6e64
commit 04afd30e5b
2 changed files with 33 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
include/ck/host_utility/kernel_launch.hpp Normal file → Executable file
View File

@@ -20,6 +20,34 @@ float launch_and_time_kernel(const StreamConfig& stream_config,
#if CK_TIME_KERNEL
if(stream_config.time_kernel_)
{
if(ck::get_device_name() == "gfx940" || ck::get_device_name() == "gfx941" ||
ck::get_device_name() == "gfx942")
{
hipEvent_t start, stop;
hip_check_error(hipEventCreate(&start));
hip_check_error(hipEventCreate(&stop));
hip_check_error(hipDeviceSynchronize());
hip_check_error(hipEventRecord(start, stream_config.stream_id_));
for(int i = 0; i < stream_config.nrepeat_; ++i)
{
kernel<<<grid_dim, block_dim, lds_byte, stream_config.stream_id_>>>(args...);
hip_check_error(hipGetLastError());
}
hip_check_error(hipEventRecord(stop, stream_config.stream_id_));
hip_check_error(hipEventSynchronize(stop));
float total_time = 0;
hip_check_error(hipEventElapsedTime(&total_time, start, stop));
total_time /= 10;
stream_config.cold_niters_ =
(1000.0 / total_time); // we need longer runtime to ramp up the clk on MI300s
stream_config.nrepeat_ = stream_config.cold_niters_;
}
#if DEBUG_LOG
printf("%s: grid_dim {%d, %d, %d}, block_dim {%d, %d, %d} \n",
__func__,

10
include/ck/stream_config.hpp Normal file → Executable file
View File

@@ -8,9 +8,9 @@
struct StreamConfig
{
hipStream_t stream_id_ = nullptr;
bool time_kernel_ = false;
int log_level_ = 0;
int cold_niters_ = 5;
int nrepeat_ = 50;
hipStream_t stream_id_ = nullptr;
bool time_kernel_ = false;
int log_level_ = 0;
mutable int cold_niters_ = 5;
mutable int nrepeat_ = 50;
};