1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
|
/****************************************************************/
/* Parallel Combinatorial BLAS Library (for Graph Computations) */
/* version 1.6 -------------------------------------------------*/
/* date: 6/15/2017 ---------------------------------------------*/
/* authors: Ariful Azad, Aydin Buluc --------------------------*/
/****************************************************************/
/*
Copyright (c) 2010-2017, The Regents of the University of California
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
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
*/
#ifndef _SP_IMPL_H_
#define _SP_IMPL_H_
#include <iostream>
#include <vector>
#include "PreAllocatedSPA.h"
#include "Deleter.h"
namespace combblas {
template <class IT, class NT>
class Dcsc;
template <class IT, class NT>
class Csc;
template <class SR, class IT, class NUM, class IVT, class OVT>
struct SpImpl;
//! Overload #1: DCSC
template <class SR, class IT, class NUM, class IVT, class OVT>
void SpMXSpV(const Dcsc<IT,NUM> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector< OVT > & numy, PreAllocatedSPA<OVT> & SPA)
{
// ignoring SPA for now. However, a branching similar to the CSC case can be implemented
SpImpl<SR,IT,NUM,IVT,OVT>::SpMXSpV(Adcsc, mA, indx, numx, veclen, indy, numy); // don't touch this
};
//! Overload #2: DCSC
template <class SR, class IT, class NUM, class IVT, class OVT>
void SpMXSpV(const Dcsc<IT,NUM> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
int32_t * indy, OVT * numy, int * cnts, int * dspls, int p_c)
{
SpImpl<SR,IT,NUM,IVT,OVT>::SpMXSpV(Adcsc, mA, indx, numx, veclen, indy, numy, cnts, dspls,p_c); // don't touch this
};
//! Overload #3: DCSC
template <class SR, class IT, class NUM, class IVT, class OVT>
void SpMXSpV_ForThreading(const Dcsc<IT,NUM> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector< OVT > & numy, int32_t offset)
{
SpImpl<SR,IT,NUM,IVT,OVT>::SpMXSpV_ForThreading(Adcsc, mA, indx, numx, veclen, indy, numy, offset); // don't touch this
};
//! Overload #4: DCSC w/ preallocated SPA
template <class SR, class IT, class NUM, class IVT, class OVT>
void SpMXSpV_ForThreading(const Dcsc<IT,NUM> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector< OVT > & numy, int32_t offset, std::vector<OVT> & localy, BitMap & isthere, std::vector<uint32_t> & nzinds)
{
SpImpl<SR,IT,NUM,IVT,OVT>::SpMXSpV_ForThreading(Adcsc, mA, indx, numx, veclen, indy, numy, offset, localy, isthere, nzinds);
};
/*
The following two functions are base CSC implementation. All overloaded function calls will be routed to these functions.
*/
// all CSC will fall to this
template <typename SR, typename IT, typename NUM, typename IVT, typename OVT>
void SpMXSpV_HeapSort(const Csc<IT,NUM> & Acsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen, std::vector<int32_t> & indy, std::vector<OVT> & numy, int32_t offset);
// all PreAllocatedSPA will fall to this
template <class SR, class IT, class NUM, class IVT, class OVT>
void SpMXSpV_Bucket(const Csc<IT,NUM> & Acsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,std::vector<int32_t> & indy, std::vector< OVT > & numy, PreAllocatedSPA<OVT> & SPA);
//! Overload #1: CSC
template <class SR, class IT, class NUM, class IVT, class OVT>
void SpMXSpV(const Csc<IT,NUM> & Acsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
int32_t * indy, OVT * numy, int * cnts, int * dspls, int p_c)
{
std::cout << "Optbuf enabled version is not yet supported with CSC matrices" << std::endl;
};
//! Overload #2: CSC
template <class SR, class IT, class NUM, class IVT, class OVT>
void SpMXSpV(const Csc<IT,NUM> & Acsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector< OVT > & numy, PreAllocatedSPA<OVT> & SPA)
{
if(SPA.initialized)
SpMXSpV_Bucket<SR>(Acsc, mA, indx, numx, veclen, indy, numy, SPA);
else
SpMXSpV_HeapSort<SR>(Acsc, mA, indx, numx, veclen, indy, numy, 0);
};
//! Overload #3: CSC
template <class SR, class IT, class NUM, class IVT, class OVT>
void SpMXSpV_ForThreading(const Csc<IT,NUM> & Acsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector< OVT > & numy, int32_t offset)
{
SpMXSpV_HeapSort<SR>(Acsc, mA, indx, numx, veclen, indy, numy, offset);
};
//! Overload #4: CSC w/ preallocated SPA
template <class SR, class IT, class NUM, class IVT, class OVT>
void SpMXSpV_ForThreading(const Csc<IT,NUM> & Acsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector< OVT > & numy, int32_t offset, std::vector<OVT> & localy, BitMap & isthere, std::vector<uint32_t> & nzinds)
{
SpMXSpV_HeapSort<SR>(Acsc, mA, indx, numx, veclen, indy, numy, offset);
// We can eventually call SpMXSpV_HeapMerge or SpMXSpV_SPA (not implemented for CSC yet)
};
/**
* IT: The sparse matrix index type. Sparse vector index type is fixed to be int32_t
* It is the caller function's (inside ParFriends/Friends) job to convert any different types
* and ensure correctness. Rationale is efficiency, and the fact that we know for sure
* that 32-bit LOCAL indices are sufficient for all reasonable concurrencies and data sizes (as of 2011)
* \todo: As of 2015, this might not be true!!! (ABAB)
**/
template <class SR, class IT, class NUM, class IVT, class OVT>
struct SpImpl
{
static void SpMXSpV(const Dcsc<IT,NUM> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector< OVT > & numy); // specialize this
static void SpMXSpV(const Dcsc<IT,NUM> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
int32_t * indy, OVT * numy, int * cnts, int * dspls, int p_c)
{
std::cout << "Optbuf enabled version is not yet supported with general (non-boolean) matrices" << std::endl;
};
static void SpMXSpV_ForThreading(const Dcsc<IT,NUM> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector<OVT> & numy, int32_t offset)
{
std::cout << "Threaded version is not yet supported with general (non-boolean) matrices" << std::endl;
};
static void SpMXSpV_ForThreading(const Dcsc<IT,NUM> & Acsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector<OVT> & numy, int32_t offset, std::vector<OVT> & localy, BitMap & isthere, std::vector<uint32_t> & nzinds)
{
std::cout << "Threaded version is not yet supported with general (non-boolean) matrices" << std::endl;
};
};
template <class SR, class IT, class IVT, class OVT>
struct SpImpl<SR,IT,bool, IVT, OVT> // specialization
{
static void SpMXSpV(const Dcsc<IT,bool> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector< OVT > & numy);
static void SpMXSpV(const Dcsc<IT,bool> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
int32_t * indy, OVT * numy, int * cnts, int * dspls, int p_c);
//! Dcsc and vector index types do not need to match
static void SpMXSpV_ForThreading(const Dcsc<IT,bool> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector<OVT> & numy, int32_t offset);
//! Dcsc and vector index types do not need to match
static void SpMXSpV_ForThreading(const Dcsc<IT,bool> & Adcsc, int32_t mA, const int32_t * indx, const IVT * numx, int32_t veclen,
std::vector<int32_t> & indy, std::vector<OVT> & numy, int32_t offset, std::vector<OVT> & localy, BitMap & isthere, std::vector<uint32_t> & nzinds);
};
}
#include "SpImpl.cpp"
#endif
|
