cu-sp-matrix.cc
7.64 KB
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
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
// cudamatrix/cu-sp-matrix.cc
// Copyright 2013 Karel Vesely
// 2014-2015 Johns Hopkins University (author: Daniel Povey)
// See ../../COPYING for clarification regarding multiple authors
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// THIS CODE IS PROVIDED *AS IS* BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED
// WARRANTIES OR CONDITIONS OF TITLE, FITNESS FOR A PARTICULAR PURPOSE,
// MERCHANTABLITY OR NON-INFRINGEMENT.
// See the Apache 2 License for the specific language governing permissions and
// limitations under the License.
#if HAVE_CUDA == 1
#include <cuda_runtime_api.h>
#include <cublas_v2.h>
#endif
#include "base/timer.h"
#include "cudamatrix/cu-common.h"
#include "cudamatrix/cu-vector.h"
#include "cudamatrix/cu-device.h"
#include "cudamatrix/cu-kernels.h"
#include "cudamatrix/cu-math.h"
#include "cudamatrix/cu-sp-matrix.h"
#include "cudamatrix/cu-matrix.h"
#include "cudamatrix/cublas-wrappers.h"
namespace kaldi {
template<typename Real>
void CuSpMatrix<Real>::CopyFromMat(const CuMatrixBase<Real> &M,
SpCopyType copy_type) {
KALDI_ASSERT(this->num_rows_ == M.NumRows() &&
this->num_rows_ == M.NumCols());
if (this->num_rows_ == 0)
return;
#if HAVE_CUDA == 1
if (CuDevice::Instantiate().Enabled()) {
CuTimer tim;
MatrixIndexT D = this->NumRows();
if (D == 0)
return;
switch (copy_type) {
case kTakeMeanAndCheck:
KALDI_ERR << "kTakeMeanAndCheck not supported!";
// The grid/block dimensions have been very roughly tuned for the
// individual cases.
case kTakeMean:
{
dim3 dimBlock(CU2DBLOCK, CU2DBLOCK);
dim3 dimGrid(n_blocks(D, CU2DBLOCK), n_blocks(D, CU2DBLOCK));
cuda_take_mean(dimGrid, dimBlock, M.Data(), this->data_, M.Dim());
CU_SAFE_CALL(cudaGetLastError());
}
break;
case kTakeLower:
{
int32 block_size = std::min(CU1DBLOCK, this->num_rows_);
dim3 dimBlock(1, block_size);
dim3 dimGrid(D, n_blocks(D, block_size));
cuda_take_lower(dimGrid, dimBlock, M.Data(), this->data_, M.Dim());
CU_SAFE_CALL(cudaGetLastError());
}
break;
case kTakeUpper:
{
dim3 dimBlock(CU2DBLOCK, CU2DBLOCK);
dim3 dimGrid(n_blocks(D, CU2DBLOCK), n_blocks(D, CU2DBLOCK));
cuda_take_upper(dimGrid, dimBlock, M.Data(), this->data_, M.Dim());
CU_SAFE_CALL(cudaGetLastError());
}
break;
default:
KALDI_ASSERT("Invalid argument to CuSpMatrix::CopyFromMat");
}
CuDevice::Instantiate().AccuProfile("CuSpMatrix::CopyFromMat(from CuMatrixBase)", tim);
} else
#endif
{
Mat().CopyFromMat(M.Mat(), copy_type);
}
}
template<typename Real>
void CuSpMatrix<Real>::Invert() {
#if HAVE_CUDA == 1
if (CuDevice::Instantiate().Enabled()) {
CuMatrix<Real> mat(this->num_rows_, this->num_rows_);
mat.CopyFromSp(*this);
mat.SymInvertPosDef();
this->CopyFromMat(mat);
} else
#endif
{ // Use inversion of CPU-based SpMatrix.
Mat().Invert();
}
}
template<typename Real>
void CuSpMatrix<Real>::AddVec2(const Real alpha, const CuVectorBase<Real> &v) {
KALDI_ASSERT(v.Dim() == this->NumRows());
#if HAVE_CUDA == 1
if (CuDevice::Instantiate().Enabled()) {
if (this->num_rows_ == 0) return;
CuTimer tim;
size_t nr = this->num_rows_;
dim3 dimBlock(CU2DBLOCK, CU2DBLOCK);
dim3 dimGrid(n_blocks(nr, CU2DBLOCK), n_blocks(nr, CU2DBLOCK));
CUBLAS_SAFE_CALL(cublas_spr(GetCublasHandle(), CUBLAS_FILL_MODE_UPPER, this->num_rows_, alpha, v.Data(),
1, this->Data()));
CuDevice::Instantiate().AccuProfile("CuSpMatrix::AddVec2", tim);
} else
#endif
{
Mat().AddVec2(alpha, v.Vec());
}
}
template<typename Real>
void CuSpMatrix<Real>::AddMat2(const Real alpha, const CuMatrixBase<Real> &M,
MatrixTransposeType transM, const Real beta) {
KALDI_ASSERT((transM == kNoTrans && this->NumRows() == M.NumRows())
|| (transM == kTrans && this->NumRows() == M.NumCols()));
#if HAVE_CUDA == 1
if (CuDevice::Instantiate().Enabled()) {
if (this->num_rows_ == 0) return;
CuTimer tim;
MatrixIndexT this_dim = this->NumRows(),
m_other_dim = (transM == kNoTrans ? M.NumCols() : M.NumRows());
if (this_dim == 0) return;
if (alpha == 0.0) {
if (beta != 1.0) this->Scale(beta);
return;
}
cublasOperation_t trans = (transM == kTrans ? CUBLAS_OP_N : CUBLAS_OP_T);
CuMatrix<Real> tmp_mat(*this);
cublas_syrk(GetCublasHandle(), CUBLAS_FILL_MODE_UPPER, trans, this_dim, m_other_dim, alpha, M.Data(),
M.Stride(), beta, tmp_mat.Data(), tmp_mat.Stride());
this->CopyFromMat(tmp_mat, kTakeLower);
CuDevice::Instantiate().AccuProfile("CuSpMatrix::AddMat2", tim);
} else
#endif
{
Mat().AddMat2(alpha, M.Mat(), transM, beta);
}
}
/**
* C++ templatd wrapper of ANSI-C CUBLAS function GEMM (matrix multiply)
*/
template<typename Real, typename OtherReal>
Real TraceSpSp(const CuSpMatrix<Real> &A, const CuSpMatrix<OtherReal> &B) {
KALDI_ASSERT(A.NumRows() == B.NumRows());
#if HAVE_CUDA == 1
if (CuDevice::Instantiate().Enabled()) {
if (A.NumRows() == 0) return 0.0;
MatrixIndexT nr = A.NumRows(), size = nr * (nr+1) / 2;
CuVector<Real> Adiag(nr, kUndefined);
CuVector<OtherReal> Bdiag(nr, kUndefined);
Adiag.CopyDiagFromPacked(A);
Bdiag.CopyDiagFromPacked(B);
CuSubVector<Real> Aall(A.Data(), size);
CuSubVector<OtherReal> Ball(B.Data(), size);
// Below, we subtrace VecVec(Adiag, Bdiag) to remove double-counting
// on the diagonal.
return 2.0 * VecVec(Aall, Ball) - VecVec(Adiag, Bdiag);
} else
#endif
{
return TraceSpSp(A.Mat(), B.Mat());
}
}
template
float TraceSpSp(const CuSpMatrix<float> &A, const CuSpMatrix<float> &B);
template
float TraceSpSp(const CuSpMatrix<float> &A, const CuSpMatrix<double> &B);
template
double TraceSpSp(const CuSpMatrix<double> &A, const CuSpMatrix<float> &B);
template
double TraceSpSp(const CuSpMatrix<double> &A, const CuSpMatrix<double> &B);
template<typename Real>
bool CuSpMatrix<Real>::ApproxEqual(const CuSpMatrix<Real> &B, Real tol) const {
KALDI_ASSERT(this->NumRows() == B.NumRows());
CuSpMatrix<Real> diff(*this);
diff.AddSp(-1.0, B);
Real a = this->FrobeniusNorm(), b = B.FrobeniusNorm(),
d = diff.FrobeniusNorm();
return (d <= tol * std::max(a, b));
}
template<typename Real>
bool CuSpMatrix<Real>::IsUnit(Real tol) const {
// want to return:
//FrobeniusNorm(*this - I) <= tol * NumRows(), i.e.:
//sqrt (trace((*this - I)(*this-I)) <= tol * NumRows()
// trace((*this - I)(*this - I)) <= tol * NumRows()
// trace(*this * *this) + trace(I) - 2 * trace(*this) <= tol * NumRows()
// trace(*this * *this) + dim - 2*this.Trace() <= tol * NumRows()
// Note: we could do this more efficiently still, by slightly changing the
// definition of IsUnit and getting rid of the extra stuff inside TraceSpSp
// that corrects for the diagonal being counted twice.
return (TraceSpSp(*this, *this) + this->NumRows() - 2.0 * this->Trace() <=
tol * this->NumRows());
}
template <class Real>
CuSpMatrix<Real>& CuSpMatrix<Real>::operator = (const CuSpMatrix<Real> &in) {
this->Resize(in.NumRows(), kUndefined);
this->CopyFromPacked(in);
return *this;
}
template class CuSpMatrix<float>;
template class CuSpMatrix<double>;
} // namespace