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
|
/* Library libcerf:
* Compute complex error functions, based on a new implementation of
* Faddeeva's w_of_z. Also provide Dawson and Voigt functions.
*
* File bigloop.c:
* Compute many function values, to measure timing.
*
* Copyright:
* (C) 2022 Forschungszentrum Jülich GmbH
*
* Licence:
* Public domain.
*
* Author:
* Joachim Wuttke, Forschungszentrum Jülich, 2022
*
* Website:
* http://apps.jcns.fz-juelich.de/libcerf
*/
/*
Prevent temperature dependent variation of processor frequency.
Under Windows:
set SpeedStep and Turbo(boost).
Under Linux:
echo "1" > /sys/devices/system/cpu/intel_pstate/no_turbo
*/
#include "cerf.h"
#include <stdio.h>
#define size_t long int
int main()
{
const size_t N = 1<<29; // number of function calls
const size_t n = N>>7; // number of sweeps through the x range
double sum = 0;
/* We run n times through the range x=0..1, each time with slightly different x values.
This is intended to prevent the function under test from keeping code for nearby x values
in the L1 cache.
To scan one single time in tiny steps through the range x=0..1, just exchange the order
of the following two for loops.
*/
for (size_t j=0; j<n; ++j) {
for (size_t i=0; i<N; i += n) {
const double x = (i+j) * (1.0/N);
sum += im_w_of_x(x*2.);
sum += im_w_of_x(x*10.);
sum += im_w_of_x(x*50.);
}
}
return (int)sum;
}
|
