Skip to content

Commit dca3e0c

Browse files
authored
Merge pull request #2491 from chenxuqiang/hbmv_benchmark
benchmark/hpmv&hbmv: add benchmark/hpmv.c and benchmark/hbmv.c
2 parents c9f8db9 + 32c847d commit dca3e0c

3 files changed

Lines changed: 510 additions & 0 deletions

File tree

benchmark/Makefile

Lines changed: 93 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -73,6 +73,8 @@ goto :: slinpack.goto dlinpack.goto clinpack.goto zlinpack.goto \
7373
sasum.goto dasum.goto casum.goto zasum.goto \
7474
ssymv.goto dsymv.goto csymv.goto zsymv.goto \
7575
chemv.goto zhemv.goto \
76+
chbmv.goto zhbmv.goto \
77+
chpmv.goto zhpmv.goto \
7678
chemm.goto zhemm.goto \
7779
cherk.goto zherk.goto \
7880
cher2k.goto zher2k.goto \
@@ -106,6 +108,8 @@ acml :: slinpack.acml dlinpack.acml clinpack.acml zlinpack.acml \
106108
sasum.acml dasum.acml casum.acml zasum.acml \
107109
ssymv.acml dsymv.acml csymv.acml zsymv.acml \
108110
chemv.acml zhemv.acml \
111+
chbmv.acml zhbmv.acml \
112+
chpmv.acml zhpmv.acml \
109113
chemm.acml zhemm.acml \
110114
cherk.acml zherk.acml \
111115
cher2k.acml zher2k.acml \
@@ -139,6 +143,8 @@ atlas :: slinpack.atlas dlinpack.atlas clinpack.atlas zlinpack.atlas \
139143
sasum.atlas dasum.atlas casum.atlas zasum.atlas \
140144
ssymv.atlas dsymv.atlas csymv.atlas zsymv.atlas \
141145
chemv.atlas zhemv.atlas \
146+
chbmv.atlas zhbmv.atlas \
147+
chpmv.atlas zhpmv.atlas \
142148
chemm.acml zhemm.acml \
143149
chemm.atlas zhemm.atlas \
144150
cherk.atlas zherk.atlas \
@@ -173,6 +179,8 @@ mkl :: slinpack.mkl dlinpack.mkl clinpack.mkl zlinpack.mkl \
173179
sasum.mkl dasum.mkl casum.mkl zasum.mkl \
174180
ssymv.mkl dsymv.mkl csymv.mkl zsymv.mkl \
175181
chemv.mkl zhemv.mkl \
182+
chbmv.mkl zhbmv.mkl \
183+
chpmv.mkl zhpmv.mkl \
176184
chemm.mkl zhemm.mkl \
177185
cherk.mkl zherk.mkl \
178186
cher2k.mkl zher2k.mkl \
@@ -207,6 +215,8 @@ goto :: sgemm.goto dgemm.goto cgemm.goto zgemm.goto \
207215
sasum.goto dasum.goto casum.goto zasum.goto \
208216
ssymv.goto dsymv.goto \
209217
chemv.goto zhemv.goto \
218+
chbmv.goto zhbmv.goto \
219+
chpmv.goto zhpmv.goto \
210220
chemm.goto zhemm.goto \
211221
cherk.goto zherk.goto \
212222
cher2k.goto zher2k.goto \
@@ -239,6 +249,8 @@ acml :: slinpack.acml dlinpack.acml clinpack.acml zlinpack.acml \
239249
sasum.acml dasum.acml casum.acml zasum.acml \
240250
ssymv.acml dsymv.acml csymv.acml zsymv.acml \
241251
chemv.acml zhemv.acml \
252+
chbmv.acml zhbmv.acml \
253+
chpmv.acml zhpmv.acml \
242254
chemm.acml zhemm.acml \
243255
cherk.acml zherk.acml \
244256
cher2k.acml zher2k.acml \
@@ -272,6 +284,8 @@ atlas :: slinpack.atlas dlinpack.atlas clinpack.atlas zlinpack.atlas \
272284
sasum.atlas dasum.atlas casum.atlas zasum.atlas \
273285
ssymv.atlas dsymv.atlas csymv.atlas zsymv.atlas \
274286
chemv.atlas zhemv.atlas \
287+
chbmv.atlas zhbmv.atlas \
288+
chpmv.atlas zhpmv.atlas \
275289
chemm.acml zhemm.acml \
276290
chemm.atlas zhemm.atlas \
277291
cherk.atlas zherk.atlas \
@@ -308,6 +322,8 @@ mkl :: slinpack.mkl dlinpack.mkl clinpack.mkl zlinpack.mkl \
308322
sasum.mkl dasum.mkl casum.mkl zasum.mkl \
309323
ssymv.mkl dsymv.mkl csymv.mkl zsymv.mkl \
310324
chemv.mkl zhemv.mkl \
325+
chbmv.mkl zhbmv.mkl \
326+
chpmv.mkl zhpmv.mkl \
311327
chemm.mkl zhemm.mkl \
312328
cherk.mkl zherk.mkl \
313329
cher2k.mkl zher2k.mkl \
@@ -352,6 +368,8 @@ veclib :: slinpack.veclib dlinpack.veclib clinpack.veclib zlinpack.veclib \
352368
sasum.veclib dasum.veclib casum.veclib zasum.veclib \
353369
ssymv.veclib dsymv.veclib csymv.veclib zsymv.veclib \
354370
chemv.veclib zhemv.veclib \
371+
chbmv.veclib zhbmv.veclib \
372+
chpmv.veclib zhpmv.veclib \
355373
chemm.veclib zhemm.veclib \
356374
cherk.veclib zherk.veclib \
357375
cher2k.veclib zher2k.veclib \
@@ -1747,7 +1765,70 @@ zhemv.mkl : zhemv.$(SUFFIX)
17471765

17481766
zhemv.veclib : zhemv.$(SUFFIX)
17491767
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBVECLIB) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1768+
##################################### Chbmv ####################################################
17501769

1770+
chbmv.goto : chbmv.$(SUFFIX) ../$(LIBNAME)
1771+
$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB) -lm
1772+
1773+
chbmv.acml : chbmv.$(SUFFIX)
1774+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1775+
1776+
chbmv.atlas : chbmv.$(SUFFIX)
1777+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1778+
1779+
chbmv.mkl : chbmv.$(SUFFIX)
1780+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1781+
1782+
chbmv.veclib : chbmv.$(SUFFIX)
1783+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBVECLIB) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1784+
##################################### Zhbmv ####################################################
1785+
1786+
zhbmv.goto : zhbmv.$(SUFFIX) ../$(LIBNAME)
1787+
$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB) -lm
1788+
1789+
zhbmv.acml : zhbmv.$(SUFFIX)
1790+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1791+
1792+
zhbmv.atlas : zhbmv.$(SUFFIX)
1793+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1794+
1795+
zhbmv.mkl : zhbmv.$(SUFFIX)
1796+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1797+
1798+
zhbmv.veclib : zhbmv.$(SUFFIX)
1799+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBVECLIB) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1800+
##################################### Chpmv ####################################################
1801+
1802+
chpmv.goto : chpmv.$(SUFFIX) ../$(LIBNAME)
1803+
$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB) -lm
1804+
1805+
chpmv.acml : chpmv.$(SUFFIX)
1806+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1807+
1808+
chpmv.atlas : chpmv.$(SUFFIX)
1809+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1810+
1811+
chpmv.mkl : chpmv.$(SUFFIX)
1812+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1813+
1814+
chpmv.veclib : chpmv.$(SUFFIX)
1815+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBVECLIB) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1816+
##################################### Zhpmv ####################################################
1817+
1818+
zhpmv.goto : zhpmv.$(SUFFIX) ../$(LIBNAME)
1819+
$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB) -lm
1820+
1821+
zhpmv.acml : zhpmv.$(SUFFIX)
1822+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1823+
1824+
zhpmv.atlas : zhpmv.$(SUFFIX)
1825+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1826+
1827+
zhpmv.mkl : zhpmv.$(SUFFIX)
1828+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
1829+
1830+
zhpmv.veclib : zhpmv.$(SUFFIX)
1831+
-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBVECLIB) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
17511832
##################################### Sdot ####################################################
17521833
sdot.goto : sdot.$(SUFFIX) ../$(LIBNAME)
17531834
$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB) -lm
@@ -2597,6 +2678,18 @@ chemv.$(SUFFIX) : hemv.c
25972678
zhemv.$(SUFFIX) : hemv.c
25982679
$(CC) $(CFLAGS) -c -DCOMPLEX -DDOUBLE -o $(@F) $^
25992680

2681+
chbmv.$(SUFFIX) : hbmv.c
2682+
$(CC) $(CFLAGS) -c -DCOMPLEX -UDOUBLE -o $(@F) $^
2683+
2684+
zhbmv.$(SUFFIX) : hbmv.c
2685+
$(CC) $(CFLAGS) -c -DCOMPLEX -DDOUBLE -o $(@F) $^
2686+
2687+
chpmv.$(SUFFIX) : hpmv.c
2688+
$(CC) $(CFLAGS) -c -DCOMPLEX -UDOUBLE -o $(@F) $^
2689+
2690+
zhpmv.$(SUFFIX) : hpmv.c
2691+
$(CC) $(CFLAGS) -c -DCOMPLEX -DDOUBLE -o $(@F) $^
2692+
26002693
sdot.$(SUFFIX) : dot.c
26012694
$(CC) $(CFLAGS) -c -UCOMPLEX -UDOUBLE -o $(@F) $^
26022695

benchmark/hbmv.c

Lines changed: 210 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,210 @@
1+
/***************************************************************************
2+
Copyright (c) 2014, The OpenBLAS Project
3+
All rights reserved.
4+
Redistribution and use in source and binary forms, with or without
5+
modification, are permitted provided that the following conditions are
6+
met:
7+
1. Redistributions of source code must retain the above copyright
8+
notice, this list of conditions and the following disclaimer.
9+
2. Redistributions in binary form must reproduce the above copyright
10+
notice, this list of conditions and the following disclaimer in
11+
the documentation and/or other materials provided with the
12+
distribution.
13+
3. Neither the name of the OpenBLAS project nor the names of
14+
its contributors may be used to endorse or promote products
15+
derived from this software without specific prior written permission.
16+
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
17+
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18+
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19+
ARE DISCLAIMED. IN NO EVENT SHALL THE OPENBLAS PROJECT OR CONTRIBUTORS BE
20+
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21+
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
22+
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
23+
CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
24+
OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
25+
USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26+
*****************************************************************************/
27+
28+
#include <stdio.h>
29+
#include <stdlib.h>
30+
#ifdef __CYGWIN32__
31+
#include <sys/time.h>
32+
#endif
33+
#include "common.h"
34+
35+
36+
#undef HBMV
37+
38+
39+
#ifdef DOUBLE
40+
#define HBMV BLASFUNC(zhbmv)
41+
#else
42+
#define HBMV BLASFUNC(chbmv)
43+
#endif
44+
45+
46+
#if defined(__WIN32__) || defined(__WIN64__)
47+
48+
#ifndef DELTA_EPOCH_IN_MICROSECS
49+
#define DELTA_EPOCH_IN_MICROSECS 11644473600000000ULL
50+
#endif
51+
52+
int gettimeofday(struct timeval *tv, void *tz) {
53+
54+
FILETIME ft;
55+
unsigned __int64 tmpres = 0;
56+
static int tzflag;
57+
58+
if (NULL != tv)
59+
{
60+
GetSystemTimeAsFileTime(&ft);
61+
62+
tmpres |= ft.dwHighDateTime;
63+
tmpres <<= 32;
64+
tmpres |= ft.dwLowDateTime;
65+
66+
/*converting file time to unix epoch*/
67+
tmpres /= 10; /*convert into microseconds*/
68+
tmpres -= DELTA_EPOCH_IN_MICROSECS;
69+
tv->tv_sec = (long)(tmpres / 1000000UL);
70+
tv->tv_usec = (long)(tmpres % 1000000UL);
71+
}
72+
73+
return 0;
74+
}
75+
76+
#endif
77+
78+
#if !defined(__WIN32__) && !defined(__WIN64__) && !defined(__CYGWIN32__) && 0
79+
80+
static void *huge_malloc(BLASLONG size) {
81+
int shmid;
82+
void *address;
83+
84+
#ifndef SHM_HUGETLB
85+
#define SHM_HUGETLB 04000
86+
#endif
87+
88+
if ((shmid =shmget(IPC_PRIVATE,
89+
(size + HUGE_PAGESIZE) & ~(HUGE_PAGESIZE - 1),
90+
SHM_HUGETLB | IPC_CREAT |0600)) < 0) {
91+
printf( "Memory allocation failed(shmget).\n");
92+
exit(1);
93+
}
94+
95+
address = shmat(shmid, NULL, SHM_RND);
96+
97+
if ((BLASLONG)address == -1){
98+
printf( "Memory allocation failed(shmat).\n");
99+
exit(1);
100+
}
101+
102+
shmctl(shmid, IPC_RMID, 0);
103+
104+
return address;
105+
}
106+
107+
#define malloc huge_malloc
108+
109+
#endif
110+
111+
int main(int argc, char *argv[]){
112+
113+
FLOAT *a, *x, *y;
114+
FLOAT alpha[] = {1.0, 1.0};
115+
FLOAT beta [] = {0.0, 0.0};
116+
blasint k = 1;
117+
char uplo='L';
118+
blasint m, i, j;
119+
blasint inc_x=1, inc_y=1;
120+
int loops = 1;
121+
int l;
122+
char *p;
123+
124+
int from = 1;
125+
int to = 200;
126+
int step = 1;
127+
128+
struct timeval start, stop;
129+
double time1,timeg;
130+
131+
argc--;argv++;
132+
133+
if (argc > 0) { from = atol(*argv); argc--; argv++;}
134+
if (argc > 0) { to = MAX(atol(*argv), from); argc--; argv++;}
135+
if (argc > 0) { step = atol(*argv); argc--; argv++;}
136+
137+
if ((p = getenv("OPENBLAS_LOOPS"))) loops = atoi(p);
138+
if ((p = getenv("OPENBLAS_INCX"))) inc_x = atoi(p);
139+
if ((p = getenv("OPENBLAS_INCY"))) inc_y = atoi(p);
140+
if ((p = getenv("OPENBLAS_UPLO"))) uplo=*p;
141+
if ((p = getenv("OPENBLAS_K"))) k = atoi(p);
142+
143+
fprintf(stderr, "From : %3d To : %3d Step = %3d Uplo = '%c' k = %d Inc_x = %d Inc_y = %d Loops = %d\n",
144+
from, to, step, uplo, k, inc_x, inc_y, loops);
145+
146+
if (( a = (FLOAT *)malloc(sizeof(FLOAT) * to * to * COMPSIZE)) == NULL) {
147+
fprintf(stderr,"Out of Memory!!\n");
148+
exit(1);
149+
}
150+
151+
if (( x = (FLOAT *)malloc(sizeof(FLOAT) * to * abs(inc_x) * COMPSIZE)) == NULL) {
152+
fprintf(stderr,"Out of Memory!!\n");
153+
exit(1);
154+
}
155+
156+
if (( y = (FLOAT *)malloc(sizeof(FLOAT) * to * abs(inc_y) * COMPSIZE)) == NULL) {
157+
fprintf(stderr,"Out of Memory!!\n");
158+
exit(1);
159+
}
160+
161+
#ifdef linux
162+
srandom(getpid());
163+
#endif
164+
165+
fprintf(stderr, " SIZE Flops\n");
166+
167+
for(m = from; m <= to; m += step) {
168+
169+
timeg=0;
170+
171+
fprintf(stderr, " %6dx%d : ", (int)m, (int)m);
172+
173+
for(j = 0; j < m; j++) {
174+
for(i = 0; i < m * COMPSIZE; i++) {
175+
a[i + j * m * COMPSIZE] = ((FLOAT) rand() / (FLOAT) RAND_MAX) - 0.5;
176+
}
177+
}
178+
179+
for (l = 0; l < loops; l++) {
180+
181+
for (i = 0; i < m * COMPSIZE * abs(inc_x); i++) {
182+
x[i] = ((FLOAT) rand() / (FLOAT) RAND_MAX) - 0.5;
183+
}
184+
185+
for (i = 0; i < m * COMPSIZE * abs(inc_y); i++) {
186+
y[i] = ((FLOAT) rand() / (FLOAT) RAND_MAX) - 0.5;
187+
}
188+
189+
gettimeofday( &start, (struct timezone *)0);
190+
191+
HBMV (&uplo, &m, &k, alpha, a, &m, x, &inc_x, beta, y, &inc_y );
192+
193+
gettimeofday( &stop, (struct timezone *)0);
194+
195+
time1 = (double)(stop.tv_sec - start.tv_sec) + (double)((stop.tv_usec - start.tv_usec)) * 1.e-6;
196+
197+
timeg += time1;
198+
199+
}
200+
201+
timeg /= loops;
202+
203+
fprintf(stderr, " %10.2f MFlops\n",
204+
COMPSIZE * COMPSIZE * 2. * (double)(2 * k + 1) * (double)m / timeg * 1.e-6);
205+
}
206+
207+
return 0;
208+
}
209+
210+
// void main(int argc, char *argv[]) __attribute__((weak, alias("MAIN__")));

0 commit comments

Comments
 (0)