TheRock/rocm-systems/projects/clr/opencl/tests/ocltst/module/perf/OCLPerfSVMMemcpy.cpp

/*
 * Copyright (c) Advanced Micro Devices, Inc., or its affiliates.
 *
 * SPDX-License-Identifier: MIT
 */

#include "OCLPerfSVMMemcpy.h"

#include <Timer.h>
#include <assert.h>
#include <stdio.h>

#include <sstream>
#include <string>

#include "CL/cl.h"
#include "CL/cl_ext.h"

// Quiet pesky warnings
#ifdef WIN_OS
#define SNPRINTF sprintf_s
#else
#define SNPRINTF snprintf
#endif

#define NUM_SIZES 5
#define NUM_SRC_FLAGS 2
#define NUM_DST_FLAGS 2
#define NUM_FG_FLAGS 3

static size_t sizeList[NUM_SIZES] = {
    0x040000, 0x080000, 0x100000, 0x200000, 0x400000,
};

#if defined(CL_VERSION_2_0)
static const cl_svm_mem_flags srcFlagList[NUM_SRC_FLAGS] = {CL_MEM_READ_WRITE, CL_MEM_READ_ONLY};
static const cl_svm_mem_flags dstFlagList[NUM_DST_FLAGS] = {CL_MEM_READ_WRITE, CL_MEM_WRITE_ONLY};
static const cl_svm_mem_flags FGFlags[NUM_FG_FLAGS] = {
    0,
    CL_MEM_SVM_FINE_GRAIN_BUFFER,
    CL_MEM_SVM_FINE_GRAIN_BUFFER | CL_MEM_SVM_ATOMICS,
};
#endif

OCLPerfSVMMemcpy::OCLPerfSVMMemcpy() {
  _numSubTests = (NUM_SRC_FLAGS * NUM_DST_FLAGS * NUM_FG_FLAGS + 1) * NUM_SIZES;
  failed_ = false;
  skip_ = false;
}

OCLPerfSVMMemcpy::~OCLPerfSVMMemcpy() {}

void OCLPerfSVMMemcpy::open(unsigned int test, char* units, double& conversion,
                            unsigned int deviceId) {
  OCLTestImp::open(test, units, conversion, deviceId);
  CHECK_RESULT((error_ != CL_SUCCESS), "Error opening test");

#if defined(CL_VERSION_2_0)
  FGSystem_ = (test >= (NUM_SIZES * NUM_SRC_FLAGS * NUM_DST_FLAGS * NUM_FG_FLAGS));
  testFGFlag_ = (test / (NUM_SIZES * NUM_DST_FLAGS * NUM_SRC_FLAGS)) % (NUM_FG_FLAGS);
  testSrcFlag_ = (test / (NUM_SIZES * NUM_DST_FLAGS)) % (NUM_SRC_FLAGS);
  testDstFlag_ = (test / NUM_SIZES) % (NUM_DST_FLAGS);
  testSize_ = test % NUM_SIZES;

  cl_device_svm_capabilities caps;
  error_ = clGetDeviceInfo(devices_[deviceId], CL_DEVICE_SVM_CAPABILITIES,
                           sizeof(cl_device_svm_capabilities), &caps, NULL);
  CHECK_RESULT(error_ != CL_SUCCESS, "clGetDeviceInfo failed");
  if ((caps & CL_DEVICE_SVM_COARSE_GRAIN_BUFFER) == 0) {
    skip_ = true;  // Should never happen as OCL 2.0 devices are required to
                   // support coarse grain SVM
    testDescString = "Coarse Grain Buffer  NOT supported. Test Skipped.";
    return;
  } else if ((testFGFlag_ > 0) && (caps & CL_DEVICE_SVM_FINE_GRAIN_BUFFER) == 0) {
    skip_ = true;  // No support for fine grain buffer SVM
    testDescString = "Fine Grain Buffer NOT supported. Test Skipped.";
    return;
  } else if (FGSystem_ && (caps & CL_DEVICE_SVM_FINE_GRAIN_SYSTEM) == 0) {
    skip_ = true;  // No support for fine grain system SVM
    testDescString = "Fine Grain System NOT supported. Test Skipped.";
    return;
  } else if ((testFGFlag_ == 2) && ((caps & CL_DEVICE_SVM_ATOMICS) == 0)) {
    skip_ = true;  // No support for SVM Atomic
    testDescString = "SVM Atomic        NOT supported. Test Skipped.";
    return;
  }

  cl_device_type deviceType;
  error_ = _wrapper->clGetDeviceInfo(devices_[deviceId], CL_DEVICE_TYPE, sizeof(deviceType),
                                     &deviceType, NULL);
  CHECK_RESULT((error_ != CL_SUCCESS), "CL_DEVICE_TYPE failed");

  if (!(deviceType & CL_DEVICE_TYPE_GPU)) {
    printf("GPU device is required for this test!\n");
    failed_ = true;
    return;
  }
  return;
#else
  skip_ = true;
  testDescString = "SVM NOT supported for < 2.0 builds. Test Skipped.";
  return;
#endif
}

static void CL_CALLBACK notify_callback(const char* errinfo, const void* private_info, size_t cb,
                                        void* user_data) {}

void OCLPerfSVMMemcpy::run(void) {
  if (skip_) {
    return;
  }

  if (failed_) {
    return;
  }
#if defined(CL_VERSION_2_0)
  cl_uint *src = NULL, *dst = NULL;
  CPerfCounter timer;

  size_t bufSize = sizeList[testSize_] * sizeof(cl_int4);
  size_t iter = 100;

  cl_mem_flags srcFlags = srcFlagList[testSrcFlag_] | FGFlags[testFGFlag_];
  cl_mem_flags dstFlags = dstFlagList[testDstFlag_] | FGFlags[testFGFlag_];

  size_t gws[1] = {bufSize / sizeof(cl_int4)};
  size_t lws[1] = {64};

  if (!FGSystem_) {
    src = (cl_uint*)clSVMAlloc(context_, srcFlags, bufSize, 0);
    CHECK_RESULT(src == 0, "Allocation failed");
    dst = (cl_uint*)clSVMAlloc(context_, dstFlags, bufSize, 0);
    CHECK_RESULT(dst == 0, "Allocation failed");
  } else {  // FGSystem_ == true
    src = (cl_uint*)malloc(bufSize);
    dst = (cl_uint*)malloc(bufSize);
  }

  timer.Reset();
  timer.Start();
  for (size_t i = 0; i < iter; ++i) {
    clEnqueueSVMMemcpy(cmdQueues_[_deviceId], false, dst, src, bufSize, 0, NULL, NULL);
  }
  _wrapper->clFinish(cmdQueues_[_deviceId]);
  timer.Stop();

  if (!FGSystem_) {
    clSVMFree(context_, (void*)src);
    clSVMFree(context_, (void*)dst);
  } else {  // FGSystem_ = true
    free(src);
    free(dst);
  }

  char pSrcFlags[5];
  pSrcFlags[0] = (testSrcFlag_ == 0 || testSrcFlag_ == 1) ? 'R' : '_';  // CL_MEM_READ_ONLY
  pSrcFlags[1] = (testSrcFlag_ == 0) ? 'W' : '_';                       // CL_MEM_WRITE_ONLY
  pSrcFlags[2] =
      (testFGFlag_ == 1 || testFGFlag_ == 2) ? 'F' : '_';  // CL_MEM_SVM_FINE_GRAIN_BUFFER
  pSrcFlags[3] = (testFGFlag_ == 2) ? 'A' : '_';           // CL_MEM_SVM_ATOMICS
  pSrcFlags[4] = '\0';

  char pDstFlags[5];
  pDstFlags[0] = (testDstFlag_ == 0) ? 'R' : '_';
  pDstFlags[1] = (testDstFlag_ == 0 || testDstFlag_ == 1) ? 'W' : '_';
  pDstFlags[2] = (testFGFlag_ == 1 || testFGFlag_ == 2) ? 'F' : '_';
  pDstFlags[3] = (testFGFlag_ == 2) ? 'A' : '_';
  pSrcFlags[4] = '\0';

  char buf[256];

  if (FGSystem_) {
    SNPRINTF(buf, sizeof(buf), "Fine Grain System   SVMMemcpy (GB/s) for %6d KB, from:%4s to:%4s",
             (int)bufSize / 1024, pSrcFlags, pDstFlags);
  } else if (testFGFlag_ == 0) {
    SNPRINTF(buf, sizeof(buf), "Coarse Grain Buffer SVMMemcpy (GB/s) for %6d KB, from:%4s to:%4s",
             (int)bufSize / 1024, pSrcFlags, pDstFlags);
  } else {
    SNPRINTF(buf, sizeof(buf), "Fine Grain Buffer   SVMMemcpy (GB/s) for %6d KB, from:%4s to:%4s",
             (int)bufSize / 1024, pSrcFlags, pDstFlags);
  }

  testDescString = buf;
  double sec = timer.GetElapsedTime();
  _perfInfo = static_cast<float>((bufSize * iter * (double)(1e-09)) / sec);
#endif
}

unsigned int OCLPerfSVMMemcpy::close(void) { return OCLTestImp::close(); }