fflas-ffpack-common 2.2.2-5 / usr / include / fflas-ffpack / fflas / fflas_fgemm / fgemm_classical.inl

This file is indexed.

/usr/include/fflas-ffpack/fflas/fflas_fgemm/fgemm_classical.inl is in fflas-ffpack-common 2.2.2-5.

This file is owned by root:root, with mode 0o644.

The actual contents of the file can be viewed below. 

  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
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
/* -*- mode: C++; tab-width: 8; indent-tabs-mode: t; c-basic-offset: 8 -*- */
// vim:sts=8:sw=8:ts=8:noet:sr:cino=>s,f0,{0,g0,(0,\:0,t0,+0,=s
/*
 * Copyright (C) 2008, 2014 the FFLAS-FFPACK group
 *
 * Written by Clement Pernet <Clement.Pernet@imag.fr>
 *            Brice Boyer (briceboyer) <boyer.brice@gmail.com>
 *
 *
 * ========LICENCE========
 * This file is part of the library FFLAS-FFPACK.
 *
 * FFLAS-FFPACK is free software: you can redistribute it and/or modify
 * it under the terms of the  GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
 * ========LICENCE========
 *.
 */

/** @file fflas_fgemm/fgemm_classical.inl
 * @brief Classical \f$2n^3\$f matrix multiplication.
 * @warning The domain is supposed to be a field since some divisions are required for efficiency purposes
 * An alternative has to be written for finite rings if necessary
 */

#ifndef __FFLASFFPACK_fflas_fflas_fgemm_classical_INL
#define __FFLASFFPACK_fflas_fflas_fgemm_classical_INL

#include <cmath>

#include "fflas-ffpack/field/field-traits.h"
#ifdef __FFLASFFPACK_HAVE_SSE4_1_INSTRUCTIONS
#include "fflas-ffpack/fflas/fflas_igemm/igemm.h"
#endif
#include "fflas-ffpack/utils/Matio.h"
namespace FFLAS {

	// F is a field supporting delayed reductions
	template<class Field>
	inline void fgemm (const Field & F,
                           const FFLAS_TRANSPOSE ta,
                           const FFLAS_TRANSPOSE tb,
                           const size_t m, const size_t n,const size_t k,
                           const typename Field::Element alpha,
                           typename Field::ConstElement_ptr A, const size_t lda,
                           typename Field::ConstElement_ptr B, const size_t ldb,
                           const typename Field::Element beta,
                           typename Field::Element_ptr C, const size_t ldc,
                           MMHelper<Field, MMHelperAlgo::Classic, ModeCategories::LazyTag> & H)
	{		
                // Input matrices are unreduced: need to figure out the best option between:
                // - reducing them
                // - making possibly more blocks (smaller kmax)
		typedef MMHelper<Field, MMHelperAlgo::Classic, ModeCategories::LazyTag> HelperType;
		typename HelperType::DelayedField::Element alphadf, betadf;
		betadf = beta;
		
		if (F.isMOne (alpha)) {
			alphadf = -H.delayedField.one;
		} else {
			alphadf = F.one;
			if (! F.isOne( alpha)) {
				    // Compute y = A*x + beta/alpha.y
				    // and after y *= alpha
				FFLASFFPACK_check(!F.isZero(alpha));
				typename Field::Element betadalpha;
				F.init(betadalpha);
				F.div (betadalpha, beta, alpha);
				betadf = betadalpha;
			}
		}

		if (F.isMOne(betadf)) betadf = -F.one;

		size_t kmax = H.MaxDelayedDim (betadf);
		H.checkA(F,ta, m,k,A,lda);
		H.checkB(F,tb, k,n,B,ldb);
		if (kmax <=  k/2 || H.Aunfit() || H.Bunfit() ){
                        // Might as well reduce inputs
                        if (H.Amin < H.FieldMin || H.Amax>H.FieldMax){
				H.initA();
				freduce_constoverride (F, (ta==FflasNoTrans)?m:k, (ta==FflasNoTrans)?k:m, A, lda);
			}
			if (H.Bmin < H.FieldMin || H.Bmax>H.FieldMax){
				H.initB();
				freduce_constoverride (F, (tb==FflasNoTrans)?k:n, (tb==FflasNoTrans)?n:k, B, ldb);
			}
			if (H.Cmin < H.FieldMin || H.Cmax>H.FieldMax){
				H.initC();
				freduce (F, m, n, C, ldc);
			}
			kmax = H.MaxDelayedDim (betadf);
		}
		
		if (!kmax){
			MMHelper<Field, MMHelperAlgo::Classic, ModeCategories::DefaultTag> HG(H);
			H.initOut();
			return fgemm (F, ta, tb, m,n,k,alpha, A, lda, B, ldb, beta, C, ldc, HG);
		}

		size_t k2 = std::min(k,kmax);
		size_t nblock = k / kmax;
		size_t remblock = k % kmax;
		if (!remblock) {
			remblock = kmax;
			--nblock;
		}
		size_t shiftA, shiftB;
		if (ta == FflasTrans) shiftA = k2*lda;
		else shiftA = k2;
		if (tb == FflasTrans) shiftB = k2;
		else shiftB = k2*ldb;

		typedef MMHelper<typename HelperType::DelayedField, MMHelperAlgo::Classic, ModeCategories::DefaultBoundedTag> DelayedHelper_t;
		DelayedHelper_t Hfp(H);
		typedef typename HelperType::DelayedField::Element DFElt;
		typedef typename HelperType::DelayedField::Element_ptr DFElt_ptr;
		typedef typename HelperType::DelayedField::ConstElement_ptr DFCElt_ptr;
		
		fgemm (H.delayedField, ta, tb, m, n, remblock, alphadf, 
		       (DFCElt_ptr)A +nblock*shiftA, lda,
		       (DFCElt_ptr)B +nblock*shiftB, ldb, betadf, 
		       (DFElt_ptr)C, ldc, Hfp);

		for (size_t i = 0; i < nblock; ++i) {
			freduce (F, m, n, C, ldc);
			Hfp.initC();
			fgemm (H.delayedField, ta, tb, m, n, k2, alphadf, 
			       (DFCElt_ptr)A +i*shiftA, lda,
			       (DFCElt_ptr)B +i*shiftB, ldb, F.one, 
			       (DFElt_ptr)C, ldc, Hfp);
		}

                if (!F.isOne(alpha) && !F.isMOne(alpha)){
			DFElt al; F.convert(al, alpha);
			if (al<0) al = -al;
			// This cast is needed when Outmin base type is int8/16_t,
			// getting -Outmin returns a int, not the same base type.
			if (std::max(static_cast<const decltype(Hfp.Outmin)&>(-Hfp.Outmin), Hfp.Outmax)
			    >Hfp.MaxStorableValue/al){
				freduce (F, m, n, C, ldc);
				Hfp.initOut();
			}

			fscalin(H.delayedField, m,n,alpha,(typename DelayedHelper_t::DelayedField_t::Element_ptr)C,ldc);

			if (alpha>0){
				H.Outmin = (const DFElt)(alpha) * Hfp.Outmin;
				H.Outmax = (const DFElt)alpha * Hfp.Outmax;
			} else {
				H.Outmin = (const DFElt)alpha * Hfp.Outmax;
				H.Outmax = (const DFElt)alpha * Hfp.Outmin;
			}
		}else {
			H.Outmin = Hfp.Outmin;
			H.Outmax = Hfp.Outmax;
		}
		H.checkOut(F,m,n,C,ldc);
	}
} // FFLAS

namespace FFLAS {

	// Classic multiplication over a generic finite field
	template  < class Field>
	inline void fgemm (const Field& F,
			   const FFLAS_TRANSPOSE ta,
			   const FFLAS_TRANSPOSE tb,
			   const size_t m, const size_t n,const size_t k,
			   const typename Field::Element alpha,
			   typename Field::ConstElement_ptr A, const size_t lda,
			   typename Field::ConstElement_ptr B, const size_t ldb,
			   const typename Field::Element beta,
			   typename Field::Element_ptr C, const size_t ldc,
			   MMHelper<Field, MMHelperAlgo::Classic, ModeCategories::DefaultTag> & H)
	{
		if (F.isZero (alpha)) {
		    fscalin(F, m, n, beta, C, ldc);
		    return;
		}
                // Standard algorithm is performed over the Field, without conversion
		if (F.isZero (beta))
                        fzero (F, m, n, C, ldc);
		else {
			typename Field::Element betadivalpha;
			F.init(betadivalpha);
			F.div (betadivalpha, beta, alpha);
			fscalin(F,m,n,betadivalpha,C,ldc);
		}
		if (ta == FflasNoTrans)
			if (tb == FflasNoTrans)
				for (size_t i = 0; i < m; ++i)
					for (size_t l = 0; l < k; ++l)
						for (size_t j = 0; j < n; ++j)
							F.axpyin (*(C+i*ldc+j), *(A+i*lda+l), *(B+l*ldb+j));
			else
				for (size_t i = 0; i < m; ++i)
					for (size_t j = 0; j < n; ++j)
						for (size_t l = 0; l < k; ++l)
							F.axpyin (*(C+i*ldc+j), *(A+i*lda+l), *(B+j*ldb+l));
		else
			if (tb == FflasNoTrans)
				for (size_t i = 0; i < m; ++i)
					for (size_t l = 0; l < k; ++l)
						for (size_t j = 0; j < n; ++j)
							F.axpyin (*(C+i*ldc+j), *(A+l*lda+i), *(B+l*ldb+j));
			else
				for (size_t i = 0; i < m; ++i)
					for (size_t j = 0; j < n; ++j)
						for (size_t l = 0; l < k; ++l)
							F.axpyin (*(C+i*ldc+j), *(A+l*lda+i), *(B+j*ldb+l));
		fscalin(F,m,n,alpha,C,ldc);
	}
	template  < class Field>
	inline void fgemm (const Field& F,
			   const FFLAS_TRANSPOSE ta,
			   const FFLAS_TRANSPOSE tb,
			   const size_t m, const size_t n,const size_t k,
			   const typename Field::Element alpha,
			   typename Field::ConstElement_ptr A, const size_t lda,
			   typename Field::ConstElement_ptr B, const size_t ldb,
			   const typename Field::Element beta,
			   typename Field::Element_ptr C, const size_t ldc,
			   MMHelper<Field, MMHelperAlgo::Classic, ModeCategories::DefaultBoundedTag> & H)
	{
		MMHelper<Field, MMHelperAlgo::Classic, ModeCategories::DefaultTag>  Hd(F,0);
		fgemm (F,ta,tb,m,n,k,alpha,A,lda,B,ldb,beta,C,ldc,Hd);
		H.setOutBounds (k,alpha,beta);
	}

	inline void fgemm (const Givaro::DoubleDomain& F,
			   const FFLAS_TRANSPOSE ta,
			   const FFLAS_TRANSPOSE tb,
			   const size_t m, const size_t n,const size_t k,
			   const Givaro::DoubleDomain::Element alpha,
			   Givaro::DoubleDomain::ConstElement_ptr Ad, const size_t lda,
			   Givaro::DoubleDomain::ConstElement_ptr Bd, const size_t ldb,
			   const Givaro::DoubleDomain::Element beta,
			   Givaro::DoubleDomain::Element_ptr Cd, const size_t ldc,
			   MMHelper<Givaro::DoubleDomain, MMHelperAlgo::Classic, ModeCategories::DefaultTag> &H)
	{
		FFLASFFPACK_check(lda);
		FFLASFFPACK_check(ldb);
		FFLASFFPACK_check(ldc);

      		cblas_dgemm (CblasRowMajor, (CBLAS_TRANSPOSE) ta, (CBLAS_TRANSPOSE) tb,
			     (int)m, (int)n, (int)k, (Givaro::DoubleDomain::Element) alpha,
			     Ad, (int)lda, Bd, (int)ldb, (Givaro::DoubleDomain::Element) beta, Cd, (int)ldc);
	}

	inline void fgemm (const Givaro::FloatDomain& F,
			   const FFLAS_TRANSPOSE ta,
			   const FFLAS_TRANSPOSE tb,
			   const size_t m, const size_t n,const size_t k,
			   const Givaro::FloatDomain::Element alpha,
			   Givaro::FloatDomain::ConstElement_ptr Ad, const size_t lda,
			   Givaro::FloatDomain::ConstElement_ptr Bd, const size_t ldb,
			   const Givaro::FloatDomain::Element beta,
			   Givaro::FloatDomain::Element_ptr Cd, const size_t ldc,
			   MMHelper<Givaro::FloatDomain, MMHelperAlgo::Classic,ModeCategories::DefaultTag> & H)
	{
		FFLASFFPACK_check(lda);
		FFLASFFPACK_check(ldb);
		FFLASFFPACK_check(ldc);

		cblas_sgemm (CblasRowMajor, (CBLAS_TRANSPOSE) ta, (CBLAS_TRANSPOSE) tb,
			     (int)m, (int)n, (int)k, (Givaro::FloatDomain::Element) alpha,
			     Ad, (int)lda, Bd, (int)ldb, (Givaro::FloatDomain::Element) beta,Cd, (int)ldc);
	}

	inline void fgemm (const Givaro::ZRing<int64_t>& F,
			   const FFLAS_TRANSPOSE ta,
			   const FFLAS_TRANSPOSE tb,
			   const size_t m, const size_t n,const size_t k,
			   const int64_t alpha,
			   const int64_t * Ad, const size_t lda,
			   const int64_t * Bd, const size_t ldb,
			   const int64_t beta,
			   int64_t * Cd, const size_t ldc,
			   MMHelper<Givaro::ZRing<int64_t>, MMHelperAlgo::Classic, ModeCategories::DefaultTag> & H)
	{
		FFLASFFPACK_check(lda);
		FFLASFFPACK_check(ldb);
		FFLASFFPACK_check(ldc);
		
#if defined (__FFLASFFPACK_HAVE_SSE4_1_INSTRUCTIONS)
		igemm_ (FflasRowMajor, ta, tb, (int)m, (int)n, (int)k, alpha, Ad, (int)lda, Bd, (int)ldb, beta, Cd, (int)ldc);
#else
		for (size_t i=0; i<m; i++){
			for (size_t j=0; j<n; j++)
				Cd[i*ldc+j] *= beta;
			for (size_t l=0; l<k; l++){
				int64_t a = alpha* ((ta==FflasNoTrans) ? Ad[i*lda+l] : Ad[i+l*lda]);
				for (size_t j=0; j<n; j++)
					Cd[i*ldc+j] += a*((tb==FflasNoTrans) ? Bd[l*ldb+j] : Bd[l+j*ldb]);
			}
		}
#endif
	}
} // FFLAS

#endif // __FFLASFFPACK_fflas_fflas_fgemm_classical_INL

APT Browse - Built by Thomas Orozco.