Add single file version for teaching. Move bench scripts to util. Adapt
README.
This commit is contained in:
Jan Eitzinger
15
README.md
15
README.md
@@ -116,18 +116,3 @@ SDaxpy: 46822.63 23411.32 0.0281 0.0273 0.0325
|
||||
Solution Validates
|
||||
```
|
||||
|
||||
## Benchmarking skript
|
||||
|
||||
A perl wrapper script (bench.pl) is also provided to scan ranges of thread counts and determine the absolute highest sustained main memory bandwidth. In order to use it `likwid-pin` has to be in your path. The script has three required and one optional command line arguments:
|
||||
```
|
||||
$./bench.pl <executable> <thread count range> <repetitions> [<SMT setting>]
|
||||
```
|
||||
Example usage:
|
||||
```
|
||||
$./bench.pl ./bwbench-GCC 2-8 6
|
||||
```
|
||||
The script will always use physical cores only, where two SMT threads is the default. For different SMT thread counts use the 4th command line argument. Example for a processor without SMT:
|
||||
```
|
||||
$./bench.pl ./bwbench-GCC 14-24 10 1
|
||||
```
|
||||
|
||||
|
||||
20
util/README.md
Normal file
20
util/README.md
Normal file
@@ -0,0 +1,20 @@
|
||||
# Single file teaching version
|
||||
|
||||
bwBench.c contains a single file version of The Bandwidth Benchmark that is tailored for usage in Tutorials or Courses.
|
||||
|
||||
It should compile with any C99 compiler.
|
||||
|
||||
# Benchmarking skript
|
||||
|
||||
A wrapper scripts in perl (bench.pl) and python (bench.py) are also provided to scan ranges of thread counts and determine the absolute highest sustained main memory bandwidth. In order to use it `likwid-pin` has to be in your path. The script has three required and one optional command line arguments:
|
||||
```
|
||||
$./bench.pl <executable> <thread count range> <repetitions> [<SMT setting>]
|
||||
```
|
||||
Example usage:
|
||||
```
|
||||
$./bench.pl ./bwbench-GCC 2-8 6
|
||||
```
|
||||
The script will always use physical cores only, where two SMT threads is the default. For different SMT thread counts use the 4th command line argument. Example for a processor without SMT:
|
||||
```
|
||||
$./bench.pl ./bwbench-GCC 14-24 10 1
|
||||
```
|
||||
426
util/bwBench.c
Normal file
426
util/bwBench.c
Normal file
@@ -0,0 +1,426 @@
|
||||
/*
|
||||
* =======================================================================================
|
||||
*
|
||||
* Author: Jan Eitzinger (je), jan.eitzinger@fau.de
|
||||
* Copyright (c) 2019 RRZE, University Erlangen-Nuremberg
|
||||
*
|
||||
* Permission is hereby granted, free of charge, to any person obtaining a copy
|
||||
* of this software and associated documentation files (the "Software"), to deal
|
||||
* in the Software without restriction, including without limitation the rights
|
||||
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
||||
* copies of the Software, and to permit persons to whom the Software is
|
||||
* furnished to do so, subject to the following conditions:
|
||||
*
|
||||
* The above copyright notice and this permission notice shall be included in all
|
||||
* copies or substantial portions of the Software.
|
||||
*
|
||||
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
||||
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
||||
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
||||
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
||||
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
||||
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
||||
* SOFTWARE.
|
||||
*
|
||||
* =======================================================================================
|
||||
*/
|
||||
|
||||
#include <stdlib.h>
|
||||
#include <stdio.h>
|
||||
#include <unistd.h>
|
||||
#include <time.h>
|
||||
#include <limits.h>
|
||||
#include <float.h>
|
||||
|
||||
#ifdef _OPENMP
|
||||
#include <omp.h>
|
||||
#endif
|
||||
|
||||
#define SIZE 40000000ull
|
||||
#define NTIMES 10
|
||||
#define ARRAY_ALIGNMENT 64
|
||||
#define HLINE "----------------------------------------------------------------------------\n"
|
||||
|
||||
#ifndef MIN
|
||||
#define MIN(x,y) ((x)<(y)?(x):(y))
|
||||
#endif
|
||||
#ifndef MAX
|
||||
#define MAX(x,y) ((x)>(y)?(x):(y))
|
||||
#endif
|
||||
#ifndef ABS
|
||||
#define ABS(a) ((a) >= 0 ? (a) : -(a))
|
||||
#endif
|
||||
|
||||
typedef enum benchmark {
|
||||
INIT = 0,
|
||||
SUM,
|
||||
COPY,
|
||||
UPDATE,
|
||||
TRIAD,
|
||||
DAXPY,
|
||||
STRIAD,
|
||||
SDAXPY,
|
||||
NUMBENCH
|
||||
} benchmark;
|
||||
|
||||
typedef struct {
|
||||
char* label;
|
||||
int words;
|
||||
int flops;
|
||||
} benchmarkType;
|
||||
|
||||
extern double init(double*, double, int);
|
||||
extern double sum(double*, int);
|
||||
extern double copy(double*, double*, int);
|
||||
extern double update(double*, double, int);
|
||||
extern double triad(double*, double*, double*, double, int);
|
||||
extern double daxpy(double*, double*, double, int);
|
||||
extern double striad(double*, double*, double*, double*, int);
|
||||
extern double sdaxpy(double*, double*, double*, int);
|
||||
extern void check(double*, double*, double*, double*, int);
|
||||
extern double getTimeStamp();
|
||||
|
||||
int main (int argc, char** argv)
|
||||
{
|
||||
size_t bytesPerWord = sizeof(double);
|
||||
size_t N = SIZE;
|
||||
double *a, *b, *c, *d;
|
||||
double scalar, tmp;
|
||||
double E, S;
|
||||
|
||||
double avgtime[NUMBENCH],
|
||||
maxtime[NUMBENCH],
|
||||
mintime[NUMBENCH];
|
||||
|
||||
double times[NUMBENCH][NTIMES];
|
||||
|
||||
benchmarkType benchmarks[NUMBENCH] = {
|
||||
{"Init: ", 1, 0},
|
||||
{"Sum: ", 1, 1},
|
||||
{"Copy: ", 2, 0},
|
||||
{"Update: ", 2, 1},
|
||||
{"Triad: ", 3, 2},
|
||||
{"Daxpy: ", 3, 2},
|
||||
{"STriad: ", 4, 2},
|
||||
{"SDaxpy: ", 4, 2}
|
||||
};
|
||||
|
||||
posix_memalign((void**) &a, ARRAY_ALIGNMENT, N * bytesPerWord );
|
||||
posix_memalign((void**) &b, ARRAY_ALIGNMENT, N * bytesPerWord );
|
||||
posix_memalign((void**) &c, ARRAY_ALIGNMENT, N * bytesPerWord );
|
||||
posix_memalign((void**) &d, ARRAY_ALIGNMENT, N * bytesPerWord );
|
||||
|
||||
for (int i=0; i<NUMBENCH; i++) {
|
||||
avgtime[i] = 0;
|
||||
maxtime[i] = 0;
|
||||
mintime[i] = FLT_MAX;
|
||||
}
|
||||
|
||||
#ifdef _OPENMP
|
||||
printf(HLINE);
|
||||
#pragma omp parallel
|
||||
{
|
||||
int k = omp_get_num_threads();
|
||||
int i = omp_get_thread_num();
|
||||
|
||||
#pragma omp single
|
||||
printf ("OpenMP enabled, running with %d threads\n", k);
|
||||
}
|
||||
#endif
|
||||
|
||||
#pragma omp parallel for
|
||||
for (int i=0; i<N; i++) {
|
||||
a[i] = 2.0;
|
||||
b[i] = 2.0;
|
||||
c[i] = 0.5;
|
||||
d[i] = 1.0;
|
||||
}
|
||||
|
||||
scalar = 3.0;
|
||||
|
||||
for ( int k=0; k < NTIMES; k++) {
|
||||
times[INIT][k] = init(b, scalar, N);
|
||||
tmp = a[10];
|
||||
times[SUM][k] = sum(a, N);
|
||||
a[10] = tmp;
|
||||
times[COPY][k] = copy(c, a, N);
|
||||
times[UPDATE][k] = update(a, scalar, N);
|
||||
times[TRIAD][k] = triad(a, b, c, scalar, N);
|
||||
times[DAXPY][k] = daxpy(a, b, scalar, N);
|
||||
times[STRIAD][k] = striad(a, b, c, d, N);
|
||||
times[SDAXPY][k] = sdaxpy(a, b, c, N);
|
||||
}
|
||||
|
||||
for (int j=0; j<NUMBENCH; j++) {
|
||||
for (int k=1; k<NTIMES; k++) {
|
||||
avgtime[j] = avgtime[j] + times[j][k];
|
||||
mintime[j] = MIN(mintime[j], times[j][k]);
|
||||
maxtime[j] = MAX(maxtime[j], times[j][k]);
|
||||
}
|
||||
}
|
||||
|
||||
printf(HLINE);
|
||||
printf("Function Rate(MB/s) Rate(MFlop/s) Avg time Min time Max time\n");
|
||||
for (int j=0; j<NUMBENCH; j++) {
|
||||
avgtime[j] = avgtime[j]/(double)(NTIMES-1);
|
||||
double bytes = (double) benchmarks[j].words * sizeof(double) * N;
|
||||
double flops = (double) benchmarks[j].flops * N;
|
||||
|
||||
if (flops > 0){
|
||||
printf("%s%11.2f %11.2f %11.4f %11.4f %11.4f\n", benchmarks[j].label,
|
||||
1.0E-06 * bytes/mintime[j],
|
||||
1.0E-06 * flops/mintime[j],
|
||||
avgtime[j],
|
||||
mintime[j],
|
||||
maxtime[j]);
|
||||
} else {
|
||||
printf("%s%11.2f - %11.4f %11.4f %11.4f\n", benchmarks[j].label,
|
||||
1.0E-06 * bytes/mintime[j],
|
||||
avgtime[j],
|
||||
mintime[j],
|
||||
maxtime[j]);
|
||||
}
|
||||
}
|
||||
printf(HLINE);
|
||||
check(a, b, c, d, N);
|
||||
|
||||
return EXIT_SUCCESS;
|
||||
}
|
||||
|
||||
void check(
|
||||
double * a,
|
||||
double * b,
|
||||
double * c,
|
||||
double * d,
|
||||
int N
|
||||
)
|
||||
{
|
||||
double aj, bj, cj, dj, scalar;
|
||||
double asum, bsum, csum, dsum;
|
||||
double epsilon;
|
||||
|
||||
/* reproduce initialization */
|
||||
aj = 2.0;
|
||||
bj = 2.0;
|
||||
cj = 0.5;
|
||||
dj = 1.0;
|
||||
|
||||
/* now execute timing loop */
|
||||
scalar = 3.0;
|
||||
|
||||
for (int k=0; k<NTIMES; k++) {
|
||||
bj = scalar;
|
||||
cj = aj;
|
||||
aj = aj * scalar;
|
||||
aj = bj + scalar * cj;
|
||||
aj = aj + scalar * bj;
|
||||
aj = bj + cj * dj;
|
||||
aj = aj + bj * cj;
|
||||
}
|
||||
|
||||
aj = aj * (double) (N);
|
||||
bj = bj * (double) (N);
|
||||
cj = cj * (double) (N);
|
||||
dj = dj * (double) (N);
|
||||
|
||||
asum = 0.0; bsum = 0.0; csum = 0.0; dsum = 0.0;
|
||||
|
||||
for (int i=0; i<N; i++) {
|
||||
asum += a[i];
|
||||
bsum += b[i];
|
||||
csum += c[i];
|
||||
dsum += d[i];
|
||||
}
|
||||
|
||||
#ifdef VERBOSE
|
||||
printf ("Results Comparison: \n");
|
||||
printf (" Expected : %f %f %f \n",aj,bj,cj);
|
||||
printf (" Observed : %f %f %f \n",asum,bsum,csum);
|
||||
#endif
|
||||
|
||||
epsilon = 1.e-8;
|
||||
|
||||
if (ABS(aj-asum)/asum > epsilon) {
|
||||
printf ("Failed Validation on array a[]\n");
|
||||
printf (" Expected : %f \n",aj);
|
||||
printf (" Observed : %f \n",asum);
|
||||
}
|
||||
else if (ABS(bj-bsum)/bsum > epsilon) {
|
||||
printf ("Failed Validation on array b[]\n");
|
||||
printf (" Expected : %f \n",bj);
|
||||
printf (" Observed : %f \n",bsum);
|
||||
}
|
||||
else if (ABS(cj-csum)/csum > epsilon) {
|
||||
printf ("Failed Validation on array c[]\n");
|
||||
printf (" Expected : %f \n",cj);
|
||||
printf (" Observed : %f \n",csum);
|
||||
}
|
||||
else if (ABS(dj-dsum)/dsum > epsilon) {
|
||||
printf ("Failed Validation on array d[]\n");
|
||||
printf (" Expected : %f \n",dj);
|
||||
printf (" Observed : %f \n",dsum);
|
||||
}
|
||||
else {
|
||||
printf ("Solution Validates\n");
|
||||
}
|
||||
}
|
||||
|
||||
double getTimeStamp()
|
||||
{
|
||||
struct timespec ts;
|
||||
clock_gettime(CLOCK_MONOTONIC, &ts);
|
||||
return (double)ts.tv_sec + (double)ts.tv_nsec * 1.e-9;
|
||||
}
|
||||
|
||||
double init(
|
||||
double * restrict a,
|
||||
double scalar,
|
||||
int N
|
||||
)
|
||||
{
|
||||
double S, E;
|
||||
|
||||
S = getTimeStamp();
|
||||
#pragma omp parallel for
|
||||
for (int i=0; i<N; i++) {
|
||||
a[i] = scalar;
|
||||
}
|
||||
E = getTimeStamp();
|
||||
|
||||
return E-S;
|
||||
}
|
||||
|
||||
double sum(
|
||||
double * restrict a,
|
||||
int N
|
||||
)
|
||||
{
|
||||
double S, E;
|
||||
double sum = 0.0;
|
||||
|
||||
S = getTimeStamp();
|
||||
#pragma omp parallel for reduction(+:sum)
|
||||
for (int i=0; i<N; i++) {
|
||||
sum += a[i];
|
||||
}
|
||||
E = getTimeStamp();
|
||||
|
||||
/* make the compiler think this makes actually sense */
|
||||
a[10] = sum;
|
||||
|
||||
return E-S;
|
||||
}
|
||||
|
||||
double copy(
|
||||
double * restrict a,
|
||||
double * restrict b,
|
||||
int N
|
||||
)
|
||||
{
|
||||
double S, E;
|
||||
|
||||
S = getTimeStamp();
|
||||
#pragma omp parallel for
|
||||
for (int i=0; i<N; i++) {
|
||||
a[i] = b[i];
|
||||
}
|
||||
E = getTimeStamp();
|
||||
|
||||
return E-S;
|
||||
}
|
||||
|
||||
double update(
|
||||
double * restrict a,
|
||||
double scalar,
|
||||
int N
|
||||
)
|
||||
{
|
||||
double S, E;
|
||||
|
||||
S = getTimeStamp();
|
||||
#pragma omp parallel for
|
||||
for (int i=0; i<N; i++) {
|
||||
a[i] = a[i] * scalar;
|
||||
}
|
||||
E = getTimeStamp();
|
||||
|
||||
return E-S;
|
||||
}
|
||||
|
||||
double triad(
|
||||
double * restrict a,
|
||||
double * restrict b,
|
||||
double * restrict c,
|
||||
double scalar,
|
||||
int N
|
||||
)
|
||||
{
|
||||
double S, E;
|
||||
|
||||
S = getTimeStamp();
|
||||
#pragma omp parallel for
|
||||
for (int i=0; i<N; i++) {
|
||||
a[i] = b[i] + scalar * c[i];
|
||||
}
|
||||
E = getTimeStamp();
|
||||
|
||||
return E-S;
|
||||
}
|
||||
|
||||
double daxpy(
|
||||
double * restrict a,
|
||||
double * restrict b,
|
||||
double scalar,
|
||||
int N
|
||||
)
|
||||
{
|
||||
double S, E;
|
||||
|
||||
S = getTimeStamp();
|
||||
#pragma omp parallel for
|
||||
for (int i=0; i<N; i++) {
|
||||
a[i] = a[i] + scalar * b[i];
|
||||
}
|
||||
E = getTimeStamp();
|
||||
|
||||
return E-S;
|
||||
}
|
||||
|
||||
double striad(
|
||||
double * restrict a,
|
||||
double * restrict b,
|
||||
double * restrict c,
|
||||
double * restrict d,
|
||||
int N
|
||||
)
|
||||
{
|
||||
double S, E;
|
||||
|
||||
S = getTimeStamp();
|
||||
#pragma omp parallel for
|
||||
for (int i=0; i<N; i++) {
|
||||
a[i] = b[i] + d[i] * c[i];
|
||||
}
|
||||
E = getTimeStamp();
|
||||
|
||||
return E-S;
|
||||
}
|
||||
|
||||
double sdaxpy(
|
||||
double * restrict a,
|
||||
double * restrict b,
|
||||
double * restrict c,
|
||||
int N
|
||||
)
|
||||
{
|
||||
double S, E;
|
||||
|
||||
S = getTimeStamp();
|
||||
#pragma omp parallel for
|
||||
for (int i=0; i<N; i++) {
|
||||
a[i] = a[i] + b[i] * c[i];
|
||||
}
|
||||
E = getTimeStamp();
|
||||
|
||||
return E-S;
|
||||
}
|
||||
Reference in New Issue
Block a user