cub_schmoo.cpp

/******************************************************************************
 * Copyright (c) 2011, Duane Merrill.  All rights reserved.
 * Copyright (c) 2011-2018, NVIDIA CORPORATION.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *     * Redistributions of source code must retain the above copyright
 *       notice, this list of conditions and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above copyright
 *       notice, this list of conditions and the following disclaimer in the
 *       documentation and/or other materials provided with the distribution.
 *     * Neither the name of the NVIDIA CORPORATION nor the
 *       names of its contributors may be used to endorse or promote products
 *       derived from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 ******************************************************************************/

/******************************************************************************
 * Evaluates different tuning configurations of DeviceReduce.
 *
 * The best way to use this program:
 * (1) Find the best all-around single-block tune for a given arch.
 *     For example, 100 samples [1 ..512], 100 timing iterations per config per sample:
 *         ./bin/tune_device_reduce_sm200_nvvm_5.0_abi_i386 --i=100 --s=100 --n=512 --single --device=0
 * (2) Update the single tune in device_reduce.cuh
 * (3) Find the best all-around multi-block tune for a given arch.
 *     For example, 100 samples [single-block tile-size ..  50,331,648], 100 timing iterations per config per sample:
 *         ./bin/tune_device_reduce_sm200_nvvm_5.0_abi_i386 --i=100 --s=100 --device=0
 * (4) Update the multi-block tune in device_reduce.cuh
 *
 ******************************************************************************/

// Ensure printing of CUDA runtime errors to console
#define CUB_STDERR

#include <vector>
#include <algorithm>
#include <stdio.h>
#include <cub/cub.cuh>
#include "../test/test_util.h"

using namespace cub;
using namespace std;


//---------------------------------------------------------------------
// Globals, constants and typedefs
//---------------------------------------------------------------------

#ifndef TUNE_ARCH
#define TUNE_ARCH 100
#endif

int     g_max_items         = 48 * 1024 * 1024;
int     g_samples           = 100;
int     g_timing_iterations        = 2;
bool    g_verbose           = false;
bool    g_single            = false;
bool    g_verify            = true;
CachingDeviceAllocator  g_allocator;


//---------------------------------------------------------------------
// Host utility subroutines
//---------------------------------------------------------------------

/**
 * Initialize problem
 */
template <typename T>
void Initialize(
    GenMode         gen_mode,
    T               *h_in,
    int             num_items)
{
    for (int i = 0; i < num_items; ++i)
    {
        InitValue(gen_mode, h_in[i], i);
    }
}

/**
 * Sequential reduction
 */
template <typename T, typename ReductionOp>
T Reduce(
    T               *h_in,
    ReductionOp     reduction_op,
    int             num_items)
{
    T retval = h_in[0];
    for (int i = 1; i < num_items; ++i)
        retval = reduction_op(retval, h_in[i]);

    return retval;
}



//---------------------------------------------------------------------
// Full tile test generation
//---------------------------------------------------------------------


/**
 * Wrapper structure for generating and
 * running different tuning configurations
 */
template <
    typename T,
    typename OffsetT,
    typename ReductionOp>
struct Schmoo
{