TestReduce1.cc

// -*- tab-width: 2; indent-tabs-mode: nil; coding: utf-8-with-signature -*-
//-----------------------------------------------------------------------------
// Copyright 2000-2026 CEA (www.cea.fr) IFPEN (www.ifpenergiesnouvelles.com)
// See the top-level COPYRIGHT file for details.
// SPDX-License-Identifier: Apache-2.0
//-----------------------------------------------------------------------------
 
#include <gtest/gtest.h>
 
#include "arccore/base/PlatformUtils.h"
 
#include "arccore/common/accelerator/Runner.h"
#include "arccore/common/accelerator/RunQueue.h"
#include "arccore/common/NumArray.h"
 
#include "arccore/accelerator/NumArrayViews.h"
#include "arccore/accelerator/RunCommandLoop.h"
#include "arccore/accelerator/Reduce.h"
#include "arccore/accelerator/internal/Initializer.h"
 
#include "./TestCommon.h"
 
/*---------------------------------------------------------------------------*/
/*---------------------------------------------------------------------------*/
 
using namespace Arcane;
using namespace Arcane::Accelerator;
 
/*---------------------------------------------------------------------------*/
/*---------------------------------------------------------------------------*/
 
extern "C++" Int64
_testReduceDirect(RunQueue queue, SmallSpan<const Int64> c, Int32 nb_thread,
                  Int32 nb_value, Int32 nb_part, Int32 nb_loop, bool is_async);
extern "C++" Int64
_testReduceGridStride(RunQueue queue, SmallSpan<const Int64> c, Int32 nb_thread,
                      Int32 nb_value, Int32 nb_part, Int32 nb_loop, bool is_async);
 
extern "C++" Int64
_testReduceEmpty(RunQueue queue, SmallSpan<const Int64> c);
 
void _doTestReduceDirect(bool use_accelerator, Int32 max_allowed_thread)
{
  Accelerator::Initializer x(use_accelerator, max_allowed_thread);
  Runner runner(x.executionPolicy());
  RunQueue queue(makeQueue(runner));
  if (queue.isAcceleratorPolicy())
    queue.setMemoryRessource(eMemoryResource::Device);
  Int32 nb_loop = 1000;
  Int32 nb_value = 1000000;
  Int64 expected_value = 1000004000000;
 
  nb_value = 10000000;
  expected_value = 100000040000000;
  nb_loop = 100;
 
  Int32 nb_thread = 256;
  Int32 nb_part = 1;
  if (!queue.isAcceleratorPolicy()) {
    if (arccoreIsDebug())
      nb_loop /= 20;
    else
      nb_loop /= 4;
    if (nb_loop == 0)
      nb_loop = 1;
  }
  //nb_loop = 1;
 
  std::cout << "Using accelerator policy name=" << queue.executionPolicy() << "\n";
  std::cout << "Sizeof (ReducerSum2<Int64>) = " << sizeof(ReducerSum2<Int64>) << " nb_loop=" << nb_loop << "\n";
 
  eMemoryResource mem = queue.memoryResource();
  NumArray<Int64, MDDim1> host_c(eMemoryResource::Host);
  host_c.resize(nb_value);
  {
    for (Int32 i = 0; i < nb_value; ++i) {
      host_c(i) = (i + 2) + (i + 3);
    };
  }
 
  NumArray<Int64, MDDim1> c(mem);
  c.copy(host_c);
 
  nb_part = 1;
  for (Int32 k = 1; k < 5; ++k) {
    {
      Int64 v = _testReduceDirect(queue, c, nb_thread, nb_value, nb_part, nb_loop, false);
      Int64 v2 = v / nb_loop;
      ASSERT_EQ(v2, expected_value);
    }
    {
      Int64 v = _testReduceDirect(queue, c, nb_thread, nb_value, nb_part, nb_loop, true);
      Int64 v2 = v / nb_loop;
      ASSERT_EQ(v2, expected_value);
    }
    nb_part *= 2;
  }
 
  nb_part = 1;
  for (Int32 k = 1; k < 5; ++k) {
    {
      Int64 v = _testReduceGridStride(queue, c, nb_thread, nb_value, nb_part, nb_loop, false);
      Int64 v2 = v / nb_loop;
      ASSERT_EQ(v2, expected_value);
    }
    {
      Int64 v = _testReduceGridStride(queue, c, nb_thread, nb_value, nb_part, nb_loop, true);
      Int64 v2 = v / nb_loop;
      ASSERT_EQ(v2, expected_value);
    }
    nb_part *= 2;
  }
  // Teste la réduction avec un tableau vide (dans ce cas le kernel n'est pas lancé)
  {
    Int64 v = _testReduceEmpty(queue, {});
    ASSERT_EQ(v, 0);
  }
}
 
ARCCORE_TEST_DO_TEST_ACCELERATOR(ArccoreAccelerator, TestReduceDirect, _doTestReduceDirect);
 
/*---------------------------------------------------------------------------*/
/*---------------------------------------------------------------------------*/