6c4a7d0828
So far, this kernel doesn't deal with edge. AUGEM: Automatically Generate High Performance Dense Linear Algebra Kernels on x86 CPUs. Qian Wang, Xianyi Zhang, Yunquan Zhang, and Qing Yi. In the International Conference for High Performance Computing, Networking, Storage and Analysis (SC'13). Denver, CO. Nov, 2013.
320 lines
7.5 KiB
C
320 lines
7.5 KiB
C
/*********************************************************************/
|
|
/* Copyright 2009, 2010 The University of Texas at Austin. */
|
|
/* All rights reserved. */
|
|
/* */
|
|
/* Redistribution and use in source and binary forms, with or */
|
|
/* without modification, are permitted provided that the following */
|
|
/* conditions are met: */
|
|
/* */
|
|
/* 1. Redistributions of source code must retain the above */
|
|
/* copyright notice, this list of conditions and the following */
|
|
/* disclaimer. */
|
|
/* */
|
|
/* 2. 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. */
|
|
/* */
|
|
/* THIS SOFTWARE IS PROVIDED BY THE UNIVERSITY OF TEXAS AT */
|
|
/* AUSTIN ``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 UNIVERSITY OF TEXAS AT */
|
|
/* AUSTIN 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. */
|
|
/* */
|
|
/* The views and conclusions contained in the software and */
|
|
/* documentation are those of the authors and should not be */
|
|
/* interpreted as representing official policies, either expressed */
|
|
/* or implied, of The University of Texas at Austin. */
|
|
/*********************************************************************/
|
|
|
|
#include "common.h"
|
|
|
|
static FLOAT dm1 = -1.;
|
|
|
|
#ifdef CONJ
|
|
#define GEMM_KERNEL GEMM_KERNEL_R
|
|
#else
|
|
#define GEMM_KERNEL GEMM_KERNEL_N
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_M == 1
|
|
#define GEMM_UNROLL_M_SHIFT 0
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_M == 2
|
|
#define GEMM_UNROLL_M_SHIFT 1
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_M == 4
|
|
#define GEMM_UNROLL_M_SHIFT 2
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_M == 6
|
|
#define GEMM_UNROLL_M_SHIFT 2
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_M == 8
|
|
#define GEMM_UNROLL_M_SHIFT 3
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_M == 16
|
|
#define GEMM_UNROLL_M_SHIFT 4
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_N == 1
|
|
#define GEMM_UNROLL_N_SHIFT 0
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_N == 2
|
|
#define GEMM_UNROLL_N_SHIFT 1
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_N == 4
|
|
#define GEMM_UNROLL_N_SHIFT 2
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_N == 8
|
|
#define GEMM_UNROLL_N_SHIFT 3
|
|
#endif
|
|
|
|
#if GEMM_DEFAULT_UNROLL_N == 16
|
|
#define GEMM_UNROLL_N_SHIFT 4
|
|
#endif
|
|
|
|
#ifndef COMPLEX
|
|
|
|
static inline void solve(BLASLONG m, BLASLONG n, FLOAT *a, FLOAT *b, FLOAT *c, BLASLONG ldc) {
|
|
|
|
FLOAT aa, bb;
|
|
|
|
int i, j, k;
|
|
|
|
for (i = 0; i < n; i++) {
|
|
|
|
bb = *(b + i);
|
|
|
|
for (j = 0; j < m; j ++) {
|
|
aa = *(c + j + i * ldc);
|
|
aa *= bb;
|
|
*a = aa;
|
|
*(c + j + i * ldc) = aa;
|
|
a ++;
|
|
|
|
for (k = i + 1; k < n; k ++){
|
|
*(c + j + k * ldc) -= aa * *(b + k);
|
|
}
|
|
|
|
}
|
|
b += n;
|
|
}
|
|
}
|
|
|
|
#else
|
|
|
|
static inline void solve(BLASLONG m, BLASLONG n, FLOAT *a, FLOAT *b, FLOAT *c, BLASLONG ldc) {
|
|
|
|
FLOAT aa1, aa2;
|
|
FLOAT bb1, bb2;
|
|
FLOAT cc1, cc2;
|
|
|
|
int i, j, k;
|
|
|
|
ldc *= 2;
|
|
|
|
for (i = 0; i < n; i++) {
|
|
|
|
bb1 = *(b + i * 2 + 0);
|
|
bb2 = *(b + i * 2 + 1);
|
|
|
|
for (j = 0; j < m; j ++) {
|
|
aa1 = *(c + j * 2 + 0 + i * ldc);
|
|
aa2 = *(c + j * 2 + 1 + i * ldc);
|
|
|
|
#ifndef CONJ
|
|
cc1 = aa1 * bb1 - aa2 * bb2;
|
|
cc2 = aa1 * bb2 + aa2 * bb1;
|
|
#else
|
|
cc1 = aa1 * bb1 + aa2 * bb2;
|
|
cc2 = -aa1 * bb2 + aa2 * bb1;
|
|
#endif
|
|
|
|
*(a + 0) = cc1;
|
|
*(a + 1) = cc2;
|
|
*(c + j * 2 + 0 + i * ldc) = cc1;
|
|
*(c + j * 2 + 1 + i * ldc) = cc2;
|
|
a += 2;
|
|
|
|
for (k = i + 1; k < n; k ++){
|
|
#ifndef CONJ
|
|
*(c + j * 2 + 0 + k * ldc) -= cc1 * *(b + k * 2 + 0) - cc2 * *(b + k * 2 + 1);
|
|
*(c + j * 2 + 1 + k * ldc) -= cc1 * *(b + k * 2 + 1) + cc2 * *(b + k * 2 + 0);
|
|
#else
|
|
*(c + j * 2 + 0 + k * ldc) -= cc1 * *(b + k * 2 + 0) + cc2 * *(b + k * 2 + 1);
|
|
*(c + j * 2 + 1 + k * ldc) -= - cc1 * *(b + k * 2 + 1) + cc2 * *(b + k * 2 + 0);
|
|
#endif
|
|
}
|
|
|
|
}
|
|
b += n * 2;
|
|
}
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
int CNAME(BLASLONG m, BLASLONG n, BLASLONG k, FLOAT dummy1,
|
|
#ifdef COMPLEX
|
|
FLOAT dummy2,
|
|
#endif
|
|
FLOAT *a, FLOAT *b, FLOAT *c, BLASLONG ldc, BLASLONG offset){
|
|
|
|
FLOAT *aa, *cc;
|
|
BLASLONG kk;
|
|
BLASLONG i, j, jj;
|
|
|
|
#if 0
|
|
fprintf(stderr, "TRSM RN KERNEL m = %3ld n = %3ld k = %3ld offset = %3ld\n",
|
|
m, n, k, offset);
|
|
#endif
|
|
|
|
jj = 0;
|
|
j = (n >> GEMM_UNROLL_N_SHIFT);
|
|
kk = -offset;
|
|
|
|
while (j > 0) {
|
|
|
|
aa = a;
|
|
cc = c;
|
|
|
|
i = (m >> GEMM_UNROLL_M_SHIFT);
|
|
|
|
if (i > 0) {
|
|
do {
|
|
if (kk > 0) {
|
|
GEMM_KERNEL(GEMM_UNROLL_M, GEMM_UNROLL_N, kk, dm1,
|
|
#ifdef COMPLEX
|
|
ZERO,
|
|
#endif
|
|
aa, b, cc, ldc);
|
|
}
|
|
|
|
solve(GEMM_UNROLL_M, GEMM_UNROLL_N,
|
|
aa + kk * GEMM_UNROLL_M * COMPSIZE,
|
|
b + kk * GEMM_UNROLL_N * COMPSIZE,
|
|
cc, ldc);
|
|
|
|
aa += GEMM_UNROLL_M * k * COMPSIZE;
|
|
cc += GEMM_UNROLL_M * COMPSIZE;
|
|
i --;
|
|
} while (i > 0);
|
|
}
|
|
|
|
|
|
if (m & (GEMM_UNROLL_M - 1)) {
|
|
i = (GEMM_UNROLL_M >> 1);
|
|
while (i > 0) {
|
|
if (m & i) {
|
|
if (kk > 0) {
|
|
GEMM_KERNEL(i, GEMM_UNROLL_N, kk, dm1,
|
|
#ifdef COMPLEX
|
|
ZERO,
|
|
#endif
|
|
aa, b, cc, ldc);
|
|
}
|
|
solve(i, GEMM_UNROLL_N,
|
|
aa + kk * i * COMPSIZE,
|
|
b + kk * GEMM_UNROLL_N * COMPSIZE,
|
|
cc, ldc);
|
|
|
|
aa += i * k * COMPSIZE;
|
|
cc += i * COMPSIZE;
|
|
}
|
|
i >>= 1;
|
|
}
|
|
}
|
|
|
|
kk += GEMM_UNROLL_N;
|
|
b += GEMM_UNROLL_N * k * COMPSIZE;
|
|
c += GEMM_UNROLL_N * ldc * COMPSIZE;
|
|
j --;
|
|
jj += GEMM_UNROLL_M;
|
|
}
|
|
|
|
if (n & (GEMM_UNROLL_N - 1)) {
|
|
|
|
j = (GEMM_UNROLL_N >> 1);
|
|
while (j > 0) {
|
|
if (n & j) {
|
|
|
|
aa = a;
|
|
cc = c;
|
|
|
|
i = (m >> GEMM_UNROLL_M_SHIFT);
|
|
|
|
while (i > 0) {
|
|
if (kk > 0) {
|
|
GEMM_KERNEL(GEMM_UNROLL_M, j, kk, dm1,
|
|
#ifdef COMPLEX
|
|
ZERO,
|
|
#endif
|
|
aa,
|
|
b,
|
|
cc,
|
|
ldc);
|
|
}
|
|
|
|
solve(GEMM_UNROLL_M, j,
|
|
aa + kk * GEMM_UNROLL_M * COMPSIZE,
|
|
b + kk * j * COMPSIZE, cc, ldc);
|
|
|
|
aa += GEMM_UNROLL_M * k * COMPSIZE;
|
|
cc += GEMM_UNROLL_M * COMPSIZE;
|
|
i --;
|
|
}
|
|
|
|
if (m & (GEMM_UNROLL_M - 1)) {
|
|
i = (GEMM_UNROLL_M >> 1);
|
|
while (i > 0) {
|
|
if (m & i) {
|
|
if (kk > 0) {
|
|
GEMM_KERNEL(i, j, kk, dm1,
|
|
#ifdef COMPLEX
|
|
ZERO,
|
|
#endif
|
|
aa,
|
|
b,
|
|
cc,
|
|
ldc);
|
|
}
|
|
|
|
solve(i, j,
|
|
aa + kk * i * COMPSIZE,
|
|
b + kk * j * COMPSIZE, cc, ldc);
|
|
|
|
aa += i * k * COMPSIZE;
|
|
cc += i * COMPSIZE;
|
|
}
|
|
i >>= 1;
|
|
}
|
|
}
|
|
|
|
b += j * k * COMPSIZE;
|
|
c += j * ldc * COMPSIZE;
|
|
kk += j;
|
|
}
|
|
j >>= 1;
|
|
}
|
|
}
|
|
|
|
return 0;
|
|
}
|