AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Simplify hipDispatchLatency sample (#1793)

Browse Source 
* Use hipExtLaunchKernelGGL in dispatchlatency sample

* Let it run on NVCC path too

* Refactoring

* Add test_kernel source

* Remove ResultDB

* Remove error checks
This commit is contained in:
Rahul Garg
2020-01-22 14:37:47 -08:00
committed by GitHub
parent 6a881a647c
commit c7280bec4d
5 changed files with 140 additions and 690 deletions
Download Patch File Download Diff File Expand all files Collapse all files
samples/1_Utils/hipDispatchLatency/hipDispatchLatency.cpp
+110 -121
View File
@@ -1,16 +1,13 @@
/*
Copyright (c) 2015-present Advanced Micro Devices, Inc. All rights reserved.
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
@@ -21,142 +18,134 @@ THE SOFTWARE.
*/
#include "hip/hip_runtime.h"
#ifdef __HIP_PLATFORM_HCC__
#include "hip/hip_ext.h"
#endif
#include <iostream>
#include <time.h>
#include "ResultDatabase.h"
#define PRINT_PROGRESS 0
#define check(cmd) \
{ \
hipError_t status = cmd; \
if (status != hipSuccess) { \
printf("error: '%s'(%d) from %s at %s:%d\n", hipGetErrorString(status), status, #cmd, \
__FILE__, __LINE__); \
abort(); \
} \
}
#define LEN 1024 * 1024
#include <chrono>
#include <algorithm>
#define NUM_GROUPS 1
#define GROUP_SIZE 64
#define TEST_ITERS 20
#define DISPATCHES_PER_TEST 100
#define GROUP_SIZE 1
#define WARMUP_RUN_COUNT 10
#define TIMING_RUN_COUNT 100
#define TOTAL_RUN_COUNT WARMUP_RUN_COUNT + TIMING_RUN_COUNT
#define BATCH_SIZE 1000
const unsigned p_tests = 0xfffffff;
#define FILE_NAME "test_kernel.code"
#define KERNEL_NAME "test"
__global__ void EmptyKernel() { }
// HCC optimizes away fully NULL kernel calls, so run one that is nearly null:
__global__ void NearlyNull(float* Ad) {
if (Ad) {
Ad[0] = 42;
}
void print_timing(std::string test, const std::array<float, TOTAL_RUN_COUNT> &results, int batch = 1) {
float total_us = 0.0f, mean_us = 0.0f, stddev_us = 0.0f;
// skip warm-up runs
auto start_iter = std::next(results.begin(), WARMUP_RUN_COUNT);
auto end_iter = results.end();
// mean
std::for_each(start_iter, end_iter, [&](const float &run_ms) {
total_us += (run_ms * 1000) / batch;
});
mean_us = total_us / TIMING_RUN_COUNT;
// stddev
total_us = 0;
std::for_each(start_iter, end_iter, [&](const float &run_ms) {
float dev_us = ((run_ms * 1000) / batch) - mean_us;
total_us += dev_us * dev_us;
});
stddev_us = sqrt(total_us / TIMING_RUN_COUNT);
// display
printf("\n %s: %.1f us, std: %.1f us\n", test.c_str(), mean_us, stddev_us);
}
ResultDatabase resultDB;
void stopTest(hipEvent_t start, hipEvent_t stop, const char* msg, int iters) {
float mS = 0;
check(hipEventRecord(stop));
check(hipDeviceSynchronize());
check(hipEventElapsedTime(&mS, start, stop));
resultDB.AddResult(std::string(msg), "", "uS", mS * 1000 / iters);
if (PRINT_PROGRESS & 0x1) {
std::cout << msg << "\t\t" << mS * 1000 / iters << " uS" << std::endl;
}
if (PRINT_PROGRESS & 0x2) {
resultDB.DumpSummary(std::cout);
}
}
int main() {
hipError_t err;
float* Ad;
check(hipMalloc(&Ad, 4));
hipStream_t stream;
check(hipStreamCreate(&stream));
hipEvent_t start, sync, stop;
check(hipEventCreate(&start));
check(hipEventCreateWithFlags(&sync, hipEventBlockingSync));
check(hipEventCreate(&stop));
int main() {
hipStream_t stream0 = 0;
hipDevice_t device;
hipDeviceGet(&device, 0);
hipCtx_t context;
hipCtxCreate(&context, 0, device);
hipModule_t module;
hipFunction_t function;
hipModuleLoad(&module, FILE_NAME);
hipModuleGetFunction(&function, module, KERNEL_NAME);
void* params = nullptr;
std::array<float, TOTAL_RUN_COUNT> results;
hipEvent_t start, stop;
hipEventCreate(&start);
hipEventCreate(&stop);
/************************************************************************************/
/* HIP kernel launch enqueue rate: */
/* Measure time taken to enqueue a kernel on the GPU */
/************************************************************************************/
if (p_tests & 0x1) {
hipEventRecord(start);
hipLaunchKernelGGL(NearlyNull, dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream0, Ad);
stopTest(start, stop, "FirstKernelLaunch", 1);
// Timing hipModuleLaunchKernel
for (auto i = 0; i < TOTAL_RUN_COUNT; ++i) {
auto start = std::chrono::high_resolution_clock::now();
hipModuleLaunchKernel(function, 1, 1, 1, 1, 1, 1, 0, 0, &params, nullptr);
auto stop = std::chrono::high_resolution_clock::now();
results[i] = std::chrono::duration<float, std::milli>(stop - start).count();
}
print_timing("hipModuleLaunchKernel enqueue rate", results);
if (p_tests & 0x2) {
hipEventRecord(start);
hipLaunchKernelGGL(NearlyNull, dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream0, Ad);
stopTest(start, stop, "SecondKernelLaunch", 1);
// Timing hipLaunchKernelGGL
for (auto i = 0; i < TOTAL_RUN_COUNT; ++i) {
auto start = std::chrono::high_resolution_clock::now();
hipLaunchKernelGGL((EmptyKernel), dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream0);
auto stop = std::chrono::high_resolution_clock::now();
results[i] = std::chrono::duration<float, std::milli>(stop - start).count();
}
print_timing("hipLaunchKernelGGL enqueue rate", results);
/***********************************************************************************/
/* Single dispatch execution latency using HIP events: */
/* Measures latency to start & finish executing a kernel with GPU-scope visibility */
/***********************************************************************************/
if (p_tests & 0x4) {
for (int t = 0; t < TEST_ITERS; t++) {
hipEventRecord(start);
for (int i = 0; i < DISPATCHES_PER_TEST; i++) {
hipLaunchKernelGGL(NearlyNull, dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream0, Ad);
hipEventRecord(sync);
hipEventSynchronize(sync);
}
stopTest(start, stop, "NullStreamASyncDispatchWait", DISPATCHES_PER_TEST);
}
}
if (p_tests & 0x10) {
for (int t = 0; t < TEST_ITERS; t++) {
hipEventRecord(start);
for (int i = 0; i < DISPATCHES_PER_TEST; i++) {
hipLaunchKernelGGL(NearlyNull, dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream, Ad);
hipEventRecord(sync);
hipEventSynchronize(sync);
}
stopTest(start, stop, "StreamASyncDispatchWait", DISPATCHES_PER_TEST);
}
}
#if 1
if (p_tests & 0x40) {
for (int t = 0; t < TEST_ITERS; t++) {
hipEventRecord(start);
for (int i = 0; i < DISPATCHES_PER_TEST; i++) {
hipLaunchKernelGGL(NearlyNull, dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream0, Ad);
}
stopTest(start, stop, "NullStreamASyncDispatchNoWait", DISPATCHES_PER_TEST);
}
}
if (p_tests & 0x80) {
for (int t = 0; t < TEST_ITERS; t++) {
hipEventRecord(start);
for (int i = 0; i < DISPATCHES_PER_TEST; i++) {
hipLaunchKernelGGL(NearlyNull, dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream, Ad);
}
stopTest(start, stop, "StreamASyncDispatchNoWait", DISPATCHES_PER_TEST);
}
//Timing directly the dispatch
#ifdef __HIP_PLATFORM_HCC__
for (auto i = 0; i < TOTAL_RUN_COUNT; ++i) {
hipExtLaunchKernelGGL((EmptyKernel), dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream0, start, stop, 0);
hipEventSynchronize(stop);
hipEventElapsedTime(&results[i], start, stop);
}
print_timing("Timing directly single dispatch latency", results);
#endif
resultDB.DumpSummary(std::cout);
//Timing around the dispatch
for (auto i = 0; i < TOTAL_RUN_COUNT; ++i) {
hipEventRecord(start, 0);
hipLaunchKernelGGL((EmptyKernel), dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream0);
hipEventRecord(stop, 0);
hipEventSynchronize(stop);
hipEventElapsedTime(&results[i], start, stop);
}
print_timing("Timing around single dispatch latency", results);
check(hipEventDestroy(start));
check(hipEventDestroy(sync));
check(hipEventDestroy(stop));
/*********************************************************************************/
/* Batch dispatch execution latency using HIP events: */
/* Measures latency to start & finish executing each dispatch in a batch */
/*********************************************************************************/
for (auto i = 0; i < TOTAL_RUN_COUNT; ++i) {
hipEventRecord(start, 0);
for (int j = 0; j < BATCH_SIZE; j++) {
hipLaunchKernelGGL((EmptyKernel), dim3(NUM_GROUPS), dim3(GROUP_SIZE), 0, stream0);
}
hipEventRecord(stop, 0);
hipEventSynchronize(stop);
hipEventElapsedTime(&results[i], start, stop);
}
print_timing("Batch dispatch latency", results, BATCH_SIZE);
hipEventDestroy(start);
hipEventDestroy(stop);
hipCtxDestroy(context);
}