457 lines
17 KiB
C++
457 lines
17 KiB
C++
/*M///////////////////////////////////////////////////////////////////////////////////////
|
|
//
|
|
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
|
//
|
|
// By downloading, copying, installing or using the software you agree to this license.
|
|
// If you do not agree to this license, do not download, install,
|
|
// copy or use the software.
|
|
//
|
|
//
|
|
// License Agreement
|
|
// For Open Source Computer Vision Library
|
|
//
|
|
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
|
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
|
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
|
// Third party copyrights are property of their respective owners.
|
|
//
|
|
// Redistribution and use in source and binary forms, with or without modification,
|
|
// are permitted provided that the following conditions are met:
|
|
//
|
|
// * Redistribution's of source code must retain the above copyright notice,
|
|
// this list of conditions and the following disclaimer.
|
|
//
|
|
// * Redistribution's 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.
|
|
//
|
|
// * The name of the copyright holders may not 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 the Intel Corporation or contributors 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.
|
|
//
|
|
//M*/
|
|
|
|
#pragma once
|
|
|
|
#ifndef OPENCV_CUDEV_BLOCK_REDUCE_KEY_VAL_DETAIL_HPP
|
|
#define OPENCV_CUDEV_BLOCK_REDUCE_KEY_VAL_DETAIL_HPP
|
|
|
|
#include "../../common.hpp"
|
|
#include "../../util/tuple.hpp"
|
|
#include "../../util/type_traits.hpp"
|
|
#include "../../warp/warp.hpp"
|
|
|
|
namespace cv { namespace cudev {
|
|
|
|
namespace block_reduce_key_val_detail
|
|
{
|
|
// GetType
|
|
|
|
template <typename T> struct GetType;
|
|
|
|
template <typename T> struct GetType<T*>
|
|
{
|
|
typedef T type;
|
|
};
|
|
|
|
template <typename T> struct GetType<volatile T*>
|
|
{
|
|
typedef T type;
|
|
};
|
|
|
|
template <typename T> struct GetType<T&>
|
|
{
|
|
typedef T type;
|
|
};
|
|
|
|
// For
|
|
|
|
template <int I, int N> struct For
|
|
{
|
|
template <class PointerTuple, class ReferenceTuple>
|
|
__device__ static void loadToSmem(const PointerTuple& smem, const ReferenceTuple& data, uint tid)
|
|
{
|
|
get<I>(smem)[tid] = get<I>(data);
|
|
|
|
For<I + 1, N>::loadToSmem(smem, data, tid);
|
|
}
|
|
|
|
template <class PointerTuple, class ReferenceTuple>
|
|
__device__ static void loadFromSmem(const PointerTuple& smem, const ReferenceTuple& data, uint tid)
|
|
{
|
|
get<I>(data) = get<I>(smem)[tid];
|
|
|
|
For<I + 1, N>::loadFromSmem(smem, data, tid);
|
|
}
|
|
|
|
template <class PointerTuple, class ReferenceTuple>
|
|
__device__ static void copy(const PointerTuple& svals, const ReferenceTuple& val, uint tid, uint delta)
|
|
{
|
|
get<I>(svals)[tid] = get<I>(val) = get<I>(svals)[tid + delta];
|
|
|
|
For<I + 1, N>::copy(svals, val, tid, delta);
|
|
}
|
|
|
|
template <class KeyPointerTuple, class KeyReferenceTuple, class ValPointerTuple, class ValReferenceTuple, class CmpTuple>
|
|
__device__ static void merge(const KeyPointerTuple& skeys, const KeyReferenceTuple& key,
|
|
const ValPointerTuple& svals, const ValReferenceTuple& val,
|
|
const CmpTuple& cmp,
|
|
uint tid, uint delta)
|
|
{
|
|
typename GetType<typename tuple_element<I, KeyPointerTuple>::type>::type reg = get<I>(skeys)[tid + delta];
|
|
|
|
if (get<I>(cmp)(reg, get<I>(key)))
|
|
{
|
|
get<I>(skeys)[tid] = get<I>(key) = reg;
|
|
get<I>(svals)[tid] = get<I>(val) = get<I>(svals)[tid + delta];
|
|
}
|
|
|
|
For<I + 1, N>::merge(skeys, key, svals, val, cmp, tid, delta);
|
|
}
|
|
};
|
|
|
|
template <int N> struct For<N, N>
|
|
{
|
|
template <class PointerTuple, class ReferenceTuple>
|
|
__device__ static void loadToSmem(const PointerTuple&, const ReferenceTuple&, uint)
|
|
{
|
|
}
|
|
|
|
template <class PointerTuple, class ReferenceTuple>
|
|
__device__ static void loadFromSmem(const PointerTuple&, const ReferenceTuple&, uint)
|
|
{
|
|
}
|
|
|
|
template <class PointerTuple, class ReferenceTuple>
|
|
__device__ static void copy(const PointerTuple&, const ReferenceTuple&, uint, uint)
|
|
{
|
|
}
|
|
|
|
template <class KeyPointerTuple, class KeyReferenceTuple, class ValPointerTuple, class ValReferenceTuple, class CmpTuple>
|
|
__device__ static void merge(const KeyPointerTuple&, const KeyReferenceTuple&,
|
|
const ValPointerTuple&, const ValReferenceTuple&,
|
|
const CmpTuple&,
|
|
uint, uint)
|
|
{
|
|
}
|
|
};
|
|
|
|
// loadToSmem / loadFromSmem
|
|
|
|
template <typename T>
|
|
__device__ __forceinline__ void loadToSmem(volatile T* smem, T& data, uint tid)
|
|
{
|
|
smem[tid] = data;
|
|
}
|
|
|
|
template <typename T>
|
|
__device__ __forceinline__ void loadFromSmem(volatile T* smem, T& data, uint tid)
|
|
{
|
|
data = smem[tid];
|
|
}
|
|
|
|
#if (CUDART_VERSION < 12040)
|
|
template <typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
|
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9>
|
|
__device__ __forceinline__ void loadToSmem(const tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& smem,
|
|
const tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& data,
|
|
uint tid)
|
|
{
|
|
For<0, tuple_size<tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9> >::value>::loadToSmem(smem, data, tid);
|
|
}
|
|
|
|
template <typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
|
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9>
|
|
__device__ __forceinline__ void loadFromSmem(const tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& smem,
|
|
const tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& data,
|
|
uint tid)
|
|
{
|
|
For<0, tuple_size<tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9> >::value>::loadFromSmem(smem, data, tid);
|
|
}
|
|
|
|
// copyVals
|
|
|
|
template <typename V>
|
|
__device__ __forceinline__ void copyVals(volatile V* svals, V& val, uint tid, uint delta)
|
|
{
|
|
svals[tid] = val = svals[tid + delta];
|
|
}
|
|
|
|
template <typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
|
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9>
|
|
__device__ __forceinline__ void copyVals(const tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& svals,
|
|
const tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
|
uint tid, uint delta)
|
|
{
|
|
For<0, tuple_size<tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9> >::value>::copy(svals, val, tid, delta);
|
|
}
|
|
|
|
// merge
|
|
|
|
template <typename K, typename V, class Cmp>
|
|
__device__ void merge(volatile K* skeys, K& key, volatile V* svals, V& val, const Cmp& cmp, uint tid, uint delta)
|
|
{
|
|
K reg = skeys[tid + delta];
|
|
|
|
if (cmp(reg, key))
|
|
{
|
|
skeys[tid] = key = reg;
|
|
copyVals(svals, val, tid, delta);
|
|
}
|
|
}
|
|
|
|
template <typename K,
|
|
typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
|
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9,
|
|
class Cmp>
|
|
__device__ void merge(volatile K* skeys, K& key,
|
|
const tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& svals,
|
|
const tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
|
const Cmp& cmp, uint tid, uint delta)
|
|
{
|
|
K reg = skeys[tid + delta];
|
|
|
|
if (cmp(reg, key))
|
|
{
|
|
skeys[tid] = key = reg;
|
|
copyVals(svals, val, tid, delta);
|
|
}
|
|
}
|
|
|
|
template <typename KP0, typename KP1, typename KP2, typename KP3, typename KP4, typename KP5, typename KP6, typename KP7, typename KP8, typename KP9,
|
|
typename KR0, typename KR1, typename KR2, typename KR3, typename KR4, typename KR5, typename KR6, typename KR7, typename KR8, typename KR9,
|
|
typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
|
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9,
|
|
class Cmp0, class Cmp1, class Cmp2, class Cmp3, class Cmp4, class Cmp5, class Cmp6, class Cmp7, class Cmp8, class Cmp9>
|
|
__device__ __forceinline__ void merge(const tuple<KP0, KP1, KP2, KP3, KP4, KP5, KP6, KP7, KP8, KP9>& skeys,
|
|
const tuple<KR0, KR1, KR2, KR3, KR4, KR5, KR6, KR7, KR8, KR9>& key,
|
|
const tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& svals,
|
|
const tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
|
const tuple<Cmp0, Cmp1, Cmp2, Cmp3, Cmp4, Cmp5, Cmp6, Cmp7, Cmp8, Cmp9>& cmp,
|
|
uint tid, uint delta)
|
|
{
|
|
For<0, tuple_size<tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9> >::value>::merge(skeys, key, svals, val, cmp, tid, delta);
|
|
}
|
|
#else
|
|
template <typename... VP, typename... VR>
|
|
__device__ __forceinline__ void loadToSmem(const tuple<VP...>& smem, const tuple<VR...>& data, uint tid)
|
|
{
|
|
For<0, tuple_size<tuple<VP...> >::value>::loadToSmem(smem, data, tid);
|
|
}
|
|
|
|
template <typename... VP, typename... VR>
|
|
__device__ __forceinline__ void loadFromSmem(const tuple<VP...>& smem, const tuple<VR...>& data, uint tid)
|
|
{
|
|
For<0, tuple_size<tuple<VP...> >::value>::loadFromSmem(smem, data, tid);
|
|
}
|
|
|
|
// copyVals
|
|
|
|
template <typename V>
|
|
__device__ __forceinline__ void copyVals(volatile V* svals, V& val, uint tid, uint delta)
|
|
{
|
|
svals[tid] = val = svals[tid + delta];
|
|
}
|
|
|
|
template <typename... VP, typename... VR>
|
|
__device__ __forceinline__ void copyVals(const tuple<VP...>& svals, const tuple<VR...>& val, uint tid, uint delta)
|
|
{
|
|
For<0, tuple_size<tuple<VP...> >::value>::copy(svals, val, tid, delta);
|
|
}
|
|
|
|
// merge
|
|
|
|
template <typename K, typename V, class Cmp>
|
|
__device__ void merge(volatile K* skeys, K& key, volatile V* svals, V& val, const Cmp& cmp, uint tid, uint delta)
|
|
{
|
|
K reg = skeys[tid + delta];
|
|
|
|
if (cmp(reg, key))
|
|
{
|
|
skeys[tid] = key = reg;
|
|
copyVals(svals, val, tid, delta);
|
|
}
|
|
}
|
|
|
|
template <typename K, typename... VP, typename... VR, class Cmp>
|
|
__device__ void merge(volatile K* skeys, K& key, const tuple<VP...>& svals, const tuple<VR...>& val, const Cmp& cmp, uint tid, uint delta)
|
|
{
|
|
K reg = skeys[tid + delta];
|
|
|
|
if (cmp(reg, key))
|
|
{
|
|
skeys[tid] = key = reg;
|
|
copyVals(svals, val, tid, delta);
|
|
}
|
|
}
|
|
|
|
template <typename... KP, typename... KR, typename... VP, typename... VR, class... Cmp>
|
|
__device__ __forceinline__ void merge(const tuple<KP...>& skeys, const tuple<KR...>& key,
|
|
const tuple<VP...>& svals, const tuple<VR...>& val,
|
|
const tuple<Cmp...>& cmp, uint tid, uint delta)
|
|
{
|
|
For<0, tuple_size<tuple<VP...> >::value>::merge(skeys, key, svals, val, cmp, tid, delta);
|
|
}
|
|
#endif
|
|
|
|
// Generic
|
|
|
|
template <int N> struct Generic
|
|
{
|
|
template <class KP, class KR, class VP, class VR, class Cmp>
|
|
__device__ static void reduce(KP skeys, KR key, VP svals, VR val, uint tid, Cmp cmp)
|
|
{
|
|
loadToSmem(skeys, key, tid);
|
|
loadValsToSmem(svals, val, tid);
|
|
if (N >= 32)
|
|
__syncthreads();
|
|
|
|
if (N >= 2048)
|
|
{
|
|
if (tid < 1024)
|
|
merge(skeys, key, svals, val, cmp, tid, 1024);
|
|
|
|
__syncthreads();
|
|
}
|
|
if (N >= 1024)
|
|
{
|
|
if (tid < 512)
|
|
merge(skeys, key, svals, val, cmp, tid, 512);
|
|
|
|
__syncthreads();
|
|
}
|
|
if (N >= 512)
|
|
{
|
|
if (tid < 256)
|
|
merge(skeys, key, svals, val, cmp, tid, 256);
|
|
|
|
__syncthreads();
|
|
}
|
|
if (N >= 256)
|
|
{
|
|
if (tid < 128)
|
|
merge(skeys, key, svals, val, cmp, tid, 128);
|
|
|
|
__syncthreads();
|
|
}
|
|
if (N >= 128)
|
|
{
|
|
if (tid < 64)
|
|
merge(skeys, key, svals, val, cmp, tid, 64);
|
|
|
|
__syncthreads();
|
|
}
|
|
if (N >= 64)
|
|
{
|
|
if (tid < 32)
|
|
merge(skeys, key, svals, val, cmp, tid, 32);
|
|
}
|
|
|
|
if (tid < 16)
|
|
{
|
|
merge(skeys, key, svals, val, cmp, tid, 16);
|
|
merge(skeys, key, svals, val, cmp, tid, 8);
|
|
merge(skeys, key, svals, val, cmp, tid, 4);
|
|
merge(skeys, key, svals, val, cmp, tid, 2);
|
|
merge(skeys, key, svals, val, cmp, tid, 1);
|
|
}
|
|
}
|
|
};
|
|
|
|
// Unroll
|
|
|
|
template <int I, class KP, class KR, class VP, class VR, class Cmp> struct Unroll
|
|
{
|
|
__device__ static void loop(KP skeys, KR key, VP svals, VR val, uint tid, Cmp cmp)
|
|
{
|
|
merge(skeys, key, svals, val, cmp, tid, I);
|
|
Unroll<I / 2, KP, KR, VP, VR, Cmp>::loop(skeys, key, svals, val, tid, cmp);
|
|
}
|
|
};
|
|
|
|
template <class KP, class KR, class VP, class VR, class Cmp> struct Unroll<0, KP, KR, VP, VR, Cmp>
|
|
{
|
|
__device__ __forceinline__ static void loop(KP, KR, VP, VR, uint, Cmp)
|
|
{
|
|
}
|
|
};
|
|
|
|
// WarpOptimized
|
|
|
|
template <int N> struct WarpOptimized
|
|
{
|
|
template <class KP, class KR, class VP, class VR, class Cmp>
|
|
__device__ static void reduce(KP skeys, KR key, VP svals, VR val, uint tid, Cmp cmp)
|
|
{
|
|
loadToSmem(skeys, key, tid);
|
|
loadToSmem(svals, val, tid);
|
|
|
|
if (tid < N / 2)
|
|
Unroll<N / 2, KP, KR, VP, VR, Cmp>::loop(skeys, key, svals, val, tid, cmp);
|
|
}
|
|
};
|
|
|
|
// GenericOptimized32
|
|
|
|
template <uint N> struct GenericOptimized32
|
|
{
|
|
enum { M = N / 32 };
|
|
|
|
template <class KP, class KR, class VP, class VR, class Cmp>
|
|
__device__ static void reduce(KP skeys, KR key, VP svals, VR val, uint tid, Cmp cmp)
|
|
{
|
|
const uint laneId = Warp::laneId();
|
|
|
|
loadToSmem(skeys, key, tid);
|
|
loadToSmem(svals, val, tid);
|
|
|
|
if (laneId < 16)
|
|
Unroll<16, KP, KR, VP, VR, Cmp>::loop(skeys, key, svals, val, tid, cmp);
|
|
|
|
__syncthreads();
|
|
|
|
if (laneId == 0)
|
|
{
|
|
loadToSmem(skeys, key, tid / 32);
|
|
loadToSmem(svals, val, tid / 32);
|
|
}
|
|
|
|
__syncthreads();
|
|
|
|
loadFromSmem(skeys, key, tid);
|
|
|
|
if (tid < 32)
|
|
{
|
|
Unroll<M / 2, KP, KR, VP, VR, Cmp>::loop(skeys, key, svals, val, tid, cmp);
|
|
}
|
|
}
|
|
};
|
|
|
|
template <int N> struct Dispatcher
|
|
{
|
|
typedef typename SelectIf<
|
|
(N <= 32) && IsPowerOf2<N>::value,
|
|
WarpOptimized<N>,
|
|
typename SelectIf<
|
|
(N <= 1024) && IsPowerOf2<N>::value,
|
|
GenericOptimized32<N>,
|
|
Generic<N>
|
|
>::type
|
|
>::type reductor;
|
|
};
|
|
}
|
|
|
|
}}
|
|
|
|
#endif
|