floraachy/OpenBLAS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
009864edde4d4e5e2276ca6ab2bfac4d087341cb
OpenBLAS/driver/level3/level3.c
Rajalakshmi Srinivasaraghavan 7eb55504b1 RFC : Add half precision gemm for bfloat16 in OpenBLAS 
This patch adds support for bfloat16 data type matrix multiplication kernel.
For architectures that don't support bfloat16, it is defined as unsigned short
(2 bytes).  Default unroll sizes can be changed as per architecture as done for
SGEMM and for now 8 and 4 are used for M and N.  Size of ncopy/tcopy can be
changed as per architecture requirement and for now, size 2 is used.

Added shgemm in kernel/power/KERNEL.POWER9 and tested in powerpc64le and
powerpc64.  For reference, added a small test compare_sgemm_shgemm.c to compare
sgemm and shgemm output.

This patch does not cover OpenBLAS test, benchmark and lapack tests for shgemm.
Complex type implementation can be discussed and added once this is approved.
2020-04-14 14:55:08 -05:00

413 lines
13 KiB
C
Raw Blame History

/*********************************************************************/
/* 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. */
/*********************************************************************/
/* This file is a template for level 3 operation */
#ifndef BETA_OPERATION
#if !defined(XDOUBLE) || !defined(QUAD_PRECISION)
#ifndef COMPLEX
#define BETA_OPERATION(M_FROM, M_TO, N_FROM, N_TO, BETA, C, LDC) \
GEMM_BETA((M_TO) - (M_FROM), (N_TO - N_FROM), 0, \
BETA[0], NULL, 0, NULL, 0, \
(FLOAT *)(C) + ((M_FROM) + (N_FROM) * (LDC)) * COMPSIZE, LDC)
#else
#define BETA_OPERATION(M_FROM, M_TO, N_FROM, N_TO, BETA, C, LDC) \
GEMM_BETA((M_TO) - (M_FROM), (N_TO - N_FROM), 0, \
BETA[0], BETA[1], NULL, 0, NULL, 0, \
(FLOAT *)(C) + ((M_FROM) + (N_FROM) * (LDC)) * COMPSIZE, LDC)
#endif
#else
#define BETA_OPERATION(M_FROM, M_TO, N_FROM, N_TO, BETA, C, LDC) \
GEMM_BETA((M_TO) - (M_FROM), (N_TO - N_FROM), 0, \
BETA, NULL, 0, NULL, 0, \
(FLOAT *)(C) + ((M_FROM) + (N_FROM) * (LDC)) * COMPSIZE, LDC)
#endif
#endif
#ifndef ICOPY_OPERATION
#if defined(NN) || defined(NT) || defined(NC) || defined(NR) || \
defined(RN) || defined(RT) || defined(RC) || defined(RR)
#define ICOPY_OPERATION(M, N, A, LDA, X, Y, BUFFER) GEMM_ITCOPY(M, N, (IFLOAT *)(A) + ((Y) + (X) * (LDA)) * COMPSIZE, LDA, BUFFER);
#else
#define ICOPY_OPERATION(M, N, A, LDA, X, Y, BUFFER) GEMM_INCOPY(M, N, (IFLOAT *)(A) + ((X) + (Y) * (LDA)) * COMPSIZE, LDA, BUFFER);
#endif
#endif
#ifndef OCOPY_OPERATION
#if defined(NN) || defined(TN) || defined(CN) || defined(RN) || \
defined(NR) || defined(TR) || defined(CR) || defined(RR)
#define OCOPY_OPERATION(M, N, A, LDA, X, Y, BUFFER) GEMM_ONCOPY(M, N, (IFLOAT *)(A) + ((X) + (Y) * (LDA)) * COMPSIZE, LDA, BUFFER);
#else
#define OCOPY_OPERATION(M, N, A, LDA, X, Y, BUFFER) GEMM_OTCOPY(M, N, (IFLOAT *)(A) + ((Y) + (X) * (LDA)) * COMPSIZE, LDA, BUFFER);
#endif
#endif
#ifndef KERNEL_FUNC
#if defined(NN) || defined(NT) || defined(TN) || defined(TT)
#define KERNEL_FUNC GEMM_KERNEL_N
#endif
#if defined(CN) || defined(CT) || defined(RN) || defined(RT)
#define KERNEL_FUNC GEMM_KERNEL_L
#endif
#if defined(NC) || defined(TC) || defined(NR) || defined(TR)
#define KERNEL_FUNC GEMM_KERNEL_R
#endif
#if defined(CC) || defined(CR) || defined(RC) || defined(RR)
#define KERNEL_FUNC GEMM_KERNEL_B
#endif
#endif
#ifndef KERNEL_OPERATION
#if !defined(XDOUBLE) || !defined(QUAD_PRECISION)
#ifndef COMPLEX
#define KERNEL_OPERATION(M, N, K, ALPHA, SA, SB, C, LDC, X, Y) \
KERNEL_FUNC(M, N, K, ALPHA[0], SA, SB, (FLOAT *)(C) + ((X) + (Y) * LDC) * COMPSIZE, LDC)
#else
#define KERNEL_OPERATION(M, N, K, ALPHA, SA, SB, C, LDC, X, Y) \
KERNEL_FUNC(M, N, K, ALPHA[0], ALPHA[1], SA, SB, (FLOAT *)(C) + ((X) + (Y) * LDC) * COMPSIZE, LDC)
#endif
#else
#define KERNEL_OPERATION(M, N, K, ALPHA, SA, SB, C, LDC, X, Y) \
KERNEL_FUNC(M, N, K, ALPHA, SA, SB, (FLOAT *)(C) + ((X) + (Y) * LDC) * COMPSIZE, LDC)
#endif
#endif
#ifndef FUSED_KERNEL_OPERATION
#if defined(NN) || defined(TN) || defined(CN) || defined(RN) || \
defined(NR) || defined(TR) || defined(CR) || defined(RR)
#ifndef COMPLEX
#define FUSED_KERNEL_OPERATION(M, N, K, ALPHA, SA, SB, B, LDB, C, LDC, I, J, L) \
FUSED_GEMM_KERNEL_N(M, N, K, ALPHA[0], SA, SB, \
(FLOAT *)(B) + ((L) + (J) * LDB) * COMPSIZE, LDB, (FLOAT *)(C) + ((I) + (J) * LDC) * COMPSIZE, LDC)
#else
#define FUSED_KERNEL_OPERATION(M, N, K, ALPHA, SA, SB, B, LDB, C, LDC, I, J, L) \
FUSED_GEMM_KERNEL_N(M, N, K, ALPHA[0], ALPHA[1], SA, SB, \
(FLOAT *)(B) + ((L) + (J) * LDB) * COMPSIZE, LDB, (FLOAT *)(C) + ((I) + (J) * LDC) * COMPSIZE, LDC)
#endif
#else
#ifndef COMPLEX
#define FUSED_KERNEL_OPERATION(M, N, K, ALPHA, SA, SB, B, LDB, C, LDC, I, J, L) \
FUSED_GEMM_KERNEL_T(M, N, K, ALPHA[0], SA, SB, \
(FLOAT *)(B) + ((J) + (L) * LDB) * COMPSIZE, LDB, (FLOAT *)(C) + ((I) + (J) * LDC) * COMPSIZE, LDC)
#else
#define FUSED_KERNEL_OPERATION(M, N, K, ALPHA, SA, SB, B, LDB, C, LDC, I, J, L) \
FUSED_GEMM_KERNEL_T(M, N, K, ALPHA[0], ALPHA[1], SA, SB, \
(FLOAT *)(B) + ((J) + (L) * LDB) * COMPSIZE, LDB, (FLOAT *)(C) + ((I) + (J) * LDC) * COMPSIZE, LDC)
#endif
#endif
#endif
#ifndef A
#define A args -> a
#endif
#ifndef LDA
#define LDA args -> lda
#endif
#ifndef B
#define B args -> b
#endif
#ifndef LDB
#define LDB args -> ldb
#endif
#ifndef C
#define C args -> c
#endif
#ifndef LDC
#define LDC args -> ldc
#endif
#ifndef M
#define M args -> m
#endif
#ifndef N
#define N args -> n
#endif
#ifndef K
#define K args -> k
#endif
#ifdef TIMING
#define START_RPCC() rpcc_counter = rpcc()
#define STOP_RPCC(COUNTER) COUNTER += rpcc() - rpcc_counter
#else
#define START_RPCC()
#define STOP_RPCC(COUNTER)
#endif
int CNAME(blas_arg_t *args, BLASLONG *range_m, BLASLONG *range_n,
XFLOAT *sa, XFLOAT *sb, BLASLONG dummy){
BLASLONG k, lda, ldb, ldc;
FLOAT *alpha, *beta;
IFLOAT *a, *b;
FLOAT *c;
BLASLONG m_from, m_to, n_from, n_to;
BLASLONG ls, is, js;
BLASLONG min_l, min_i, min_j;
#if !defined(FUSED_GEMM) || defined(TIMING)
BLASLONG jjs, min_jj;
#endif
BLASLONG l1stride, gemm_p, l2size;
#if defined(XDOUBLE) && defined(QUAD_PRECISION)
xidouble xalpha;
#endif
#ifdef TIMING
unsigned long long rpcc_counter;
unsigned long long innercost = 0;
unsigned long long outercost = 0;
unsigned long long kernelcost = 0;
double total;
#endif
k = K;
a = (IFLOAT *)A;
b = (IFLOAT *)B;
c = (FLOAT *)C;
lda = LDA;
ldb = LDB;
ldc = LDC;
alpha = (FLOAT *)args -> alpha;
beta = (FLOAT *)args -> beta;
m_from = 0;
m_to = M;
if (range_m) {
m_from = *(((BLASLONG *)range_m) + 0);
m_to = *(((BLASLONG *)range_m) + 1);
}
n_from = 0;
n_to = N;
if (range_n) {
n_from = *(((BLASLONG *)range_n) + 0);
n_to = *(((BLASLONG *)range_n) + 1);
}
if (beta) {
#if !defined(XDOUBLE) || !defined(QUAD_PRECISION)
#ifndef COMPLEX
if (beta[0] != ONE
#else
if ((beta[0] != ONE) || (beta[1] != ZERO)
#endif
#else
if (((beta[0].x[1] != 0x3fff000000000000UL) || beta[0].x[0] != 0)
#ifdef COMPLEX
&&(((beta[1].x[0] | beta[1].x[1]) << 1) != 0)
#endif
#endif
) {
#if defined(XDOUBLE) && defined(QUAD_PRECISION)
xidouble xbeta;
qtox(&xbeta, beta);
#endif
BETA_OPERATION(m_from, m_to, n_from, n_to, beta, c, ldc);
}
}
if ((k == 0) || (alpha == NULL)) return 0;
#if !defined(XDOUBLE) || !defined(QUAD_PRECISION)
if ( alpha[0] == ZERO
#ifdef COMPLEX
&& alpha[1] == ZERO
#endif
) return 0;
#else
if (((alpha[0].x[0] | alpha[0].x[1]
#ifdef COMPLEX
| alpha[1].x[0] | alpha[1].x[1]
#endif
) << 1) == 0) return 0;
#endif
#if defined(XDOUBLE) && defined(QUAD_PRECISION)
qtox(&xalpha, alpha);
#endif
l2size = GEMM_P * GEMM_Q;
#if 0
fprintf(stderr, "GEMM(Single): M_from : %ld M_to : %ld N_from : %ld N_to : %ld k : %ld\n", m_from, m_to, n_from, n_to, k);
fprintf(stderr, "GEMM(Single):: P = %4ld Q = %4ld R = %4ld\n", (BLASLONG)GEMM_P, (BLASLONG)GEMM_Q, (BLASLONG)GEMM_R);
// fprintf(stderr, "GEMM: SA .. %p SB .. %p\n", sa, sb);
// fprintf(stderr, "A = %p B = %p C = %p\n\tlda = %ld ldb = %ld ldc = %ld\n", a, b, c, lda, ldb, ldc);
#endif
#ifdef TIMING
innercost = 0;
outercost = 0;
kernelcost = 0;
#endif
for(js = n_from; js < n_to; js += GEMM_R){
min_j = n_to - js;
if (min_j > GEMM_R) min_j = GEMM_R;
for(ls = 0; ls < k; ls += min_l){
min_l = k - ls;
if (min_l >= GEMM_Q * 2) {
// gemm_p = GEMM_P;
min_l = GEMM_Q;
} else {
if (min_l > GEMM_Q) {
min_l = ((min_l / 2 + GEMM_UNROLL_M - 1)/GEMM_UNROLL_M) * GEMM_UNROLL_M;
}
gemm_p = ((l2size / min_l + GEMM_UNROLL_M - 1)/GEMM_UNROLL_M) * GEMM_UNROLL_M;
while (gemm_p * min_l > l2size) gemm_p -= GEMM_UNROLL_M;
}
/* First, we have to move data A to L2 cache */
min_i = m_to - m_from;
l1stride = 1;
if (min_i >= GEMM_P * 2) {
min_i = GEMM_P;
} else {
if (min_i > GEMM_P) {
min_i = ((min_i / 2 + GEMM_UNROLL_M - 1)/GEMM_UNROLL_M) * GEMM_UNROLL_M;
} else {
l1stride = 0;
}
}
START_RPCC();
ICOPY_OPERATION(min_l, min_i, a, lda, ls, m_from, sa);
STOP_RPCC(innercost);
#if defined(FUSED_GEMM) && !defined(TIMING)
FUSED_KERNEL_OPERATION(min_i, min_j, min_l, alpha,
sa, sb, b, ldb, c, ldc, m_from, js, ls);
#else
for(jjs = js; jjs < js + min_j; jjs += min_jj){
min_jj = min_j + js - jjs;
#ifdef SKYLAKEX
/* the current AVX512 s/d/c/z GEMM kernel requires n>=6*GEMM_UNROLL_N to achieve best performance */
if (min_jj >= 6*GEMM_UNROLL_N) min_jj = 6*GEMM_UNROLL_N;
#else
if (min_jj >= 3*GEMM_UNROLL_N) min_jj = 3*GEMM_UNROLL_N;
else
if (min_jj >= 2*GEMM_UNROLL_N) min_jj = 2*GEMM_UNROLL_N;
else
if (min_jj > GEMM_UNROLL_N) min_jj = GEMM_UNROLL_N;
#endif
START_RPCC();
OCOPY_OPERATION(min_l, min_jj, b, ldb, ls, jjs,
sb + min_l * (jjs - js) * COMPSIZE * l1stride);
STOP_RPCC(outercost);
START_RPCC();
#if !defined(XDOUBLE) || !defined(QUAD_PRECISION)
KERNEL_OPERATION(min_i, min_jj, min_l, alpha,
sa, sb + min_l * (jjs - js) * COMPSIZE * l1stride, c, ldc, m_from, jjs);
#else
KERNEL_OPERATION(min_i, min_jj, min_l, (void *)&xalpha,
sa, sb + min_l * (jjs - js) * COMPSIZE * l1stride, c, ldc, m_from, jjs);
#endif
STOP_RPCC(kernelcost);
}
#endif
for(is = m_from + min_i; is < m_to; is += min_i){
min_i = m_to - is;
if (min_i >= GEMM_P * 2) {
min_i = GEMM_P;
} else
if (min_i > GEMM_P) {
min_i = ((min_i / 2 + GEMM_UNROLL_M - 1)/GEMM_UNROLL_M) * GEMM_UNROLL_M;
}
START_RPCC();
ICOPY_OPERATION(min_l, min_i, a, lda, ls, is, sa);
STOP_RPCC(innercost);
START_RPCC();
#if !defined(XDOUBLE) || !defined(QUAD_PRECISION)
KERNEL_OPERATION(min_i, min_j, min_l, alpha, sa, sb, c, ldc, is, js);
#else
KERNEL_OPERATION(min_i, min_j, min_l, (void *)&xalpha, sa, sb, c, ldc, is, js);
#endif
STOP_RPCC(kernelcost);
} /* end of is */
} /* end of js */
} /* end of ls */
#ifdef TIMING
total = (double)outercost + (double)innercost + (double)kernelcost;
printf( "Copy A : %5.2f Copy B: %5.2f Kernel : %5.2f kernel Effi. : %5.2f Total Effi. : %5.2f\n",
innercost / total * 100., outercost / total * 100.,
kernelcost / total * 100.,
(double)(m_to - m_from) * (double)(n_to - n_from) * (double)k / (double)kernelcost * 100. * (double)COMPSIZE / 2.,
(double)(m_to - m_from) * (double)(n_to - n_from) * (double)k / total * 100. * (double)COMPSIZE / 2.);
#endif
return 0;
}
Reference in New Issue View Git Blame Copy Permalink