source: CIVL/examples/omp/dataracebench-1.3.2/micro-benchmarks/DRB056-jacobi2d-tile-no.c

main
Last change on this file was ea777aa, checked in by Alex Wilton <awilton@…>, 3 years ago

Moved examples, include, build_default.properties, common.xml, and README out from dev.civl.com into the root of the repo.

git-svn-id: ​svn://vsl.cis.udel.edu/civl/trunk@5704 fb995dde-84ed-4084-dfe6-e5aef3e2452c
  • Property mode set to 100644
File size: 13.5 KB
RevLine 
[a1acb0c5]1/**
2 * jacobi-2d-imper.c: This file is part of the PolyBench/C 3.2 test suite.
3 * Jacobi with array copying, no reduction. with tiling and nested SIMD.
4 *
5 * Contact: Louis-Noel Pouchet <pouchet@cse.ohio-state.edu>
6 * Web address: http://polybench.sourceforge.net
7 * License: /LICENSE.OSU.txt
8 */
9#include <stdio.h>
10#include <unistd.h>
11#include <string.h>
12#include <math.h>
13/* Include polybench common header. */
[86ee0b6]14#include "polybench/polybench.h"
[a1acb0c5]15/* Include benchmark-specific header. */
16/* Default data type is double, default size is 20x1000. */
[86ee0b6]17#include "polybench/jacobi-2d-imper.h"
[a1acb0c5]18/* Array initialization. */
19
20static void init_array(int n,double A[500 + 0][500 + 0],double B[500 + 0][500 + 0])
21{
22 //int i;
23 //int j;
24{
25 int c1;
26 int c2;
27 int c4;
28 int c3;
29 if (n >= 1) {
30#pragma omp parallel for private(c3, c4, c2)
31 for (c1 = 0; c1 <= (((n + -1) * 16 < 0?((16 < 0?-((-(n + -1) + 16 + 1) / 16) : -((-(n + -1) + 16 - 1) / 16))) : (n + -1) / 16)); c1++) {
32 for (c2 = 0; c2 <= (((n + -1) * 16 < 0?((16 < 0?-((-(n + -1) + 16 + 1) / 16) : -((-(n + -1) + 16 - 1) / 16))) : (n + -1) / 16)); c2++) {
33 for (c3 = 16 * c2; c3 <= ((16 * c2 + 15 < n + -1?16 * c2 + 15 : n + -1)); c3++) {
34#pragma omp simd
35 for (c4 = 16 * c1; c4 <= ((16 * c1 + 15 < n + -1?16 * c1 + 15 : n + -1)); c4++) {
36 A[c4][c3] = (((double )c4) * (c3 + 2) + 2) / n;
37 B[c4][c3] = (((double )c4) * (c3 + 3) + 3) / n;
38 }
39 }
40 }
41 }
42 }
43 }
44}
45/* DCE code. Must scan the entire live-out data.
46 Can be used also to check the correctness of the output. */
47
48static void print_array(int n,double A[500 + 0][500 + 0])
49{
50 int i;
51 int j;
52 for (i = 0; i < n; i++)
53 for (j = 0; j < n; j++) {
54 fprintf(stderr,"%0.2lf ",A[i][j]);
55 if ((i * n + j) % 20 == 0)
56 fprintf(stderr,"\n");
57 }
58 fprintf(stderr,"\n");
59}
60/* Main computational kernel. The whole function will be timed,
61 including the call and return. */
62
63static void kernel_jacobi_2d_imper(int tsteps,int n,double A[500 + 0][500 + 0],double B[500 + 0][500 + 0])
64{
65 //int t;
66 //int i;
67 //int j;
68
69 //#pragma scop
70{
71 int c0;
72 int c1;
73 int c3;
74 int c2;
75 int c4;
76 int c5;
77 if (n >= 3 && tsteps >= 1) {
78 for (c0 = 0; c0 <= (((n + 3 * tsteps + -4) * 16 < 0?((16 < 0?-((-(n + 3 * tsteps + -4) + 16 + 1) / 16) : -((-(n + 3 * tsteps + -4) + 16 - 1) / 16))) : (n + 3 * tsteps + -4) / 16)); c0++) {
79#pragma omp parallel for private(c5, c4, c2, c3)
80 for (c1 = (((2 * c0 * 3 < 0?-(-(2 * c0) / 3) : ((3 < 0?(-(2 * c0) + - 3 - 1) / - 3 : (2 * c0 + 3 - 1) / 3)))) > (((16 * c0 + -1 * tsteps + 1) * 16 < 0?-(-(16 * c0 + -1 * tsteps + 1) / 16) : ((16 < 0?(-(16 * c0 + -1 * tsteps + 1) + - 16 - 1) / - 16 : (16 * c0 + -1 * tsteps + 1 + 16 - 1) / 16))))?((2 * c0 * 3 < 0?-(-(2 * c0) / 3) : ((3 < 0?(-(2 * c0) + - 3 - 1) / - 3 : (2 * c0 + 3 - 1) / 3)))) : (((16 * c0 + -1 * tsteps + 1) * 16 < 0?-(-(16 * c0 + -1 * tsteps + 1) / 16) : ((16 < 0?(-(16 * c0 + -1 * tsteps + 1) + - 16 - 1) / - 16 : (16 * c0 + -1 * tsteps + 1 + 16 - 1) / 16))))); c1 <= (((((((n + 2 * tsteps + -3) * 16 < 0?((16 < 0?-((-(n + 2 * tsteps + -3) + 16 + 1) / 16) : -((-(n + 2 * tsteps + -3) + 16 - 1) / 16))) : (n + 2 * tsteps + -3) / 16)) < (((32 * c0 + n + 29) * 48 < 0?((48 < 0?-((-(32 * c0 + n + 29) + 48 + 1) / 48) : -((-(32 * c0 + n + 29) + 48 - 1) / 48))) : (32 * c0 + n + 29) / 48))?(((n + 2 * tsteps + -3) * 16 < 0?((16 < 0?-((-(n + 2 * tsteps + -3) + 16 + 1) / 16) : -((-(n + 2 * tsteps + -3) + 16 - 1) / 16))) : (n + 2 * tsteps + -3) / 16)) : (((32 * c0 + n + 29) * 48 < 0?((48 < 0?-((-(32 * c0 + n + 29) + 48 + 1) / 48) : -((-(32 * c0 + n + 29) + 48 - 1) / 48))) : (32 * c0 + n + 29) / 48)))) < c0?(((((n + 2 * tsteps + -3) * 16 < 0?((16 < 0?-((-(n + 2 * tsteps + -3) + 16 + 1) / 16) : -((-(n + 2 * tsteps + -3) + 16 - 1) / 16))) : (n + 2 * tsteps + -3) / 16)) < (((32 * c0 + n + 29) * 48 < 0?((48 < 0?-((-(32 * c0 + n + 29) + 48 + 1) / 48) : -((-(32 * c0 + n + 29) + 48 - 1) / 48))) : (32 * c0 + n + 29) / 48))?(((n + 2 * tsteps + -3) * 16 < 0?((16 < 0?-((-(n + 2 * tsteps + -3) + 16 + 1) / 16) : -((-(n + 2 * tsteps + -3) + 16 - 1) / 16))) : (n + 2 * tsteps + -3) / 16)) : (((32 * c0 + n + 29) * 48 < 0?((48 < 0?-((-(32 * c0 + n + 29) + 48 + 1) / 48) : -((-(32 * c0 + n + 29) + 48 - 1) / 48))) : (32 * c0 + n + 29) / 48)))) : c0)); c1++) {
81 for (c2 = ((((16 * c1 + -1 * n + -12) * 16 < 0?-(-(16 * c1 + -1 * n + -12) / 16) : ((16 < 0?(-(16 * c1 + -1 * n + -12) + - 16 - 1) / - 16 : (16 * c1 + -1 * n + -12 + 16 - 1) / 16)))) > 2 * c0 + -2 * c1?(((16 * c1 + -1 * n + -12) * 16 < 0?-(-(16 * c1 + -1 * n + -12) / 16) : ((16 < 0?(-(16 * c1 + -1 * n + -12) + - 16 - 1) / - 16 : (16 * c1 + -1 * n + -12 + 16 - 1) / 16)))) : 2 * c0 + -2 * c1); c2 <= (((((((16 * c1 + n + 12) * 16 < 0?((16 < 0?-((-(16 * c1 + n + 12) + 16 + 1) / 16) : -((-(16 * c1 + n + 12) + 16 - 1) / 16))) : (16 * c1 + n + 12) / 16)) < (((n + 2 * tsteps + -3) * 16 < 0?((16 < 0?-((-(n + 2 * tsteps + -3) + 16 + 1) / 16) : -((-(n + 2 * tsteps + -3) + 16 - 1) / 16))) : (n + 2 * tsteps + -3) / 16))?(((16 * c1 + n + 12) * 16 < 0?((16 < 0?-((-(16 * c1 + n + 12) + 16 + 1) / 16) : -((-(16 * c1 + n + 12) + 16 - 1) / 16))) : (16 * c1 + n + 12) / 16)) : (((n + 2 * tsteps + -3) * 16 < 0?((16 < 0?-((-(n + 2 * tsteps + -3) + 16 + 1) / 16) : -((-(n + 2 * tsteps + -3) + 16 - 1) / 16))) : (n + 2 * tsteps + -3) / 16)))) < (((32 * c0 + -32 * c1 + n + 29) * 16 < 0?((16 < 0?-((-(32 * c0 + -32 * c1 + n + 29) + 16 + 1) / 16) : -((-(32 * c0 + -32 * c1 + n + 29) + 16 - 1) / 16))) : (32 * c0 + -32 * c1 + n + 29) / 16))?(((((16 * c1 + n + 12) * 16 < 0?((16 < 0?-((-(16 * c1 + n + 12) + 16 + 1) / 16) : -((-(16 * c1 + n + 12) + 16 - 1) / 16))) : (16 * c1 + n + 12) / 16)) < (((n + 2 * tsteps + -3) * 16 < 0?((16 < 0?-((-(n + 2 * tsteps + -3) + 16 + 1) / 16) : -((-(n + 2 * tsteps + -3) + 16 - 1) / 16))) : (n + 2 * tsteps + -3) / 16))?(((16 * c1 + n + 12) * 16 < 0?((16 < 0?-((-(16 * c1 + n + 12) + 16 + 1) / 16) : -((-(16 * c1 + n + 12) + 16 - 1) / 16))) : (16 * c1 + n + 12) / 16)) : (((n + 2 * tsteps + -3) * 16 < 0?((16 < 0?-((-(n + 2 * tsteps + -3) + 16 + 1) / 16) : -((-(n + 2 * tsteps + -3) + 16 - 1) / 16))) : (n + 2 * tsteps + -3) / 16)))) : (((32 * c0 + -32 * c1 + n + 29) * 16 < 0?((16 < 0?-((-(32 * c0 + -32 * c1 + n + 29) + 16 + 1) / 16) : -((-(32 * c0 + -32 * c1 + n + 29) + 16 - 1) / 16))) : (32 * c0 + -32 * c1 + n + 29) / 16)))); c2++) {
82 if (c0 <= (((32 * c1 + 16 * c2 + -1 * n + 1) * 32 < 0?((32 < 0?-((-(32 * c1 + 16 * c2 + -1 * n + 1) + 32 + 1) / 32) : -((-(32 * c1 + 16 * c2 + -1 * n + 1) + 32 - 1) / 32))) : (32 * c1 + 16 * c2 + -1 * n + 1) / 32)) && c1 <= c2 + -1) {
83 if ((n + 1) % 2 == 0) {
84 for (c4 = (16 * c1 > 16 * c2 + -1 * n + 3?16 * c1 : 16 * c2 + -1 * n + 3); c4 <= 16 * c1 + 15; c4++) {
85 A[-16 * c2 + c4 + n + -2][n + -2] = B[-16 * c2 + c4 + n + -2][n + -2];
86 }
87 }
88 }
89 if (c0 <= (((48 * c1 + -1 * n + 1) * 32 < 0?((32 < 0?-((-(48 * c1 + -1 * n + 1) + 32 + 1) / 32) : -((-(48 * c1 + -1 * n + 1) + 32 - 1) / 32))) : (48 * c1 + -1 * n + 1) / 32)) && c1 >= c2) {
90 if ((n + 1) % 2 == 0) {
91 for (c5 = (16 * c2 > 16 * c1 + -1 * n + 3?16 * c2 : 16 * c1 + -1 * n + 3); c5 <= ((16 * c1 < 16 * c2 + 15?16 * c1 : 16 * c2 + 15)); c5++) {
92 A[n + -2][-16 * c1 + c5 + n + -2] = B[n + -2][-16 * c1 + c5 + n + -2];
93 }
94 }
95 }
96 for (c3 = ((((((16 * c1 + -1 * n + 2) * 2 < 0?-(-(16 * c1 + -1 * n + 2) / 2) : ((2 < 0?(-(16 * c1 + -1 * n + 2) + - 2 - 1) / - 2 : (16 * c1 + -1 * n + 2 + 2 - 1) / 2)))) > (((16 * c2 + -1 * n + 2) * 2 < 0?-(-(16 * c2 + -1 * n + 2) / 2) : ((2 < 0?(-(16 * c2 + -1 * n + 2) + - 2 - 1) / - 2 : (16 * c2 + -1 * n + 2 + 2 - 1) / 2))))?(((16 * c1 + -1 * n + 2) * 2 < 0?-(-(16 * c1 + -1 * n + 2) / 2) : ((2 < 0?(-(16 * c1 + -1 * n + 2) + - 2 - 1) / - 2 : (16 * c1 + -1 * n + 2 + 2 - 1) / 2)))) : (((16 * c2 + -1 * n + 2) * 2 < 0?-(-(16 * c2 + -1 * n + 2) / 2) : ((2 < 0?(-(16 * c2 + -1 * n + 2) + - 2 - 1) / - 2 : (16 * c2 + -1 * n + 2 + 2 - 1) / 2)))))) > 16 * c0 + -16 * c1?(((((16 * c1 + -1 * n + 2) * 2 < 0?-(-(16 * c1 + -1 * n + 2) / 2) : ((2 < 0?(-(16 * c1 + -1 * n + 2) + - 2 - 1) / - 2 : (16 * c1 + -1 * n + 2 + 2 - 1) / 2)))) > (((16 * c2 + -1 * n + 2) * 2 < 0?-(-(16 * c2 + -1 * n + 2) / 2) : ((2 < 0?(-(16 * c2 + -1 * n + 2) + - 2 - 1) / - 2 : (16 * c2 + -1 * n + 2 + 2 - 1) / 2))))?(((16 * c1 + -1 * n + 2) * 2 < 0?-(-(16 * c1 + -1 * n + 2) / 2) : ((2 < 0?(-(16 * c1 + -1 * n + 2) + - 2 - 1) / - 2 : (16 * c1 + -1 * n + 2 + 2 - 1) / 2)))) : (((16 * c2 + -1 * n + 2) * 2 < 0?-(-(16 * c2 + -1 * n + 2) / 2) : ((2 < 0?(-(16 * c2 + -1 * n + 2) + - 2 - 1) / - 2 : (16 * c2 + -1 * n + 2 + 2 - 1) / 2)))))) : 16 * c0 + -16 * c1); c3 <= ((((((8 * c1 + 6 < 8 * c2 + 6?8 * c1 + 6 : 8 * c2 + 6)) < tsteps + -1?((8 * c1 + 6 < 8 * c2 + 6?8 * c1 + 6 : 8 * c2 + 6)) : tsteps + -1)) < 16 * c0 + -16 * c1 + 15?((((8 * c1 + 6 < 8 * c2 + 6?8 * c1 + 6 : 8 * c2 + 6)) < tsteps + -1?((8 * c1 + 6 < 8 * c2 + 6?8 * c1 + 6 : 8 * c2 + 6)) : tsteps + -1)) : 16 * c0 + -16 * c1 + 15)); c3++) {
97 if (c1 <= ((c3 * 8 < 0?((8 < 0?-((-c3 + 8 + 1) / 8) : -((-c3 + 8 - 1) / 8))) : c3 / 8))) {
98 for (c5 = (16 * c2 > 2 * c3 + 1?16 * c2 : 2 * c3 + 1); c5 <= ((16 * c2 + 15 < 2 * c3 + n + -2?16 * c2 + 15 : 2 * c3 + n + -2)); c5++) {
99 B[1][-2 * c3 + c5] = 0.2 * (A[1][-2 * c3 + c5] + A[1][-2 * c3 + c5 - 1] + A[1][1 + (-2 * c3 + c5)] + A[1 + 1][-2 * c3 + c5] + A[1 - 1][-2 * c3 + c5]);
100 }
101 }
102 for (c4 = (16 * c1 > 2 * c3 + 2?16 * c1 : 2 * c3 + 2); c4 <= ((16 * c1 + 15 < 2 * c3 + n + -2?16 * c1 + 15 : 2 * c3 + n + -2)); c4++) {
103 if (c2 <= ((c3 * 8 < 0?((8 < 0?-((-c3 + 8 + 1) / 8) : -((-c3 + 8 - 1) / 8))) : c3 / 8))) {
104 B[-2 * c3 + c4][1] = 0.2 * (A[-2 * c3 + c4][1] + A[-2 * c3 + c4][1 - 1] + A[-2 * c3 + c4][1 + 1] + A[1 + (-2 * c3 + c4)][1] + A[-2 * c3 + c4 - 1][1]);
105 }
106 for (c5 = (16 * c2 > 2 * c3 + 2?16 * c2 : 2 * c3 + 2); c5 <= ((16 * c2 + 15 < 2 * c3 + n + -2?16 * c2 + 15 : 2 * c3 + n + -2)); c5++) {
107 B[-2 * c3 + c4][-2 * c3 + c5] = 0.2 * (A[-2 * c3 + c4][-2 * c3 + c5] + A[-2 * c3 + c4][-2 * c3 + c5 - 1] + A[-2 * c3 + c4][1 + (-2 * c3 + c5)] + A[1 + (-2 * c3 + c4)][-2 * c3 + c5] + A[-2 * c3 + c4 - 1][-2 * c3 + c5]);
108 A[-2 * c3 + c4 + -1][-2 * c3 + c5 + -1] = B[-2 * c3 + c4 + -1][-2 * c3 + c5 + -1];
109 }
110 if (c2 >= (((2 * c3 + n + -16) * 16 < 0?-(-(2 * c3 + n + -16) / 16) : ((16 < 0?(-(2 * c3 + n + -16) + - 16 - 1) / - 16 : (2 * c3 + n + -16 + 16 - 1) / 16))))) {
111 A[-2 * c3 + c4 + -1][n + -2] = B[-2 * c3 + c4 + -1][n + -2];
112 }
113 }
114 if (c1 >= (((2 * c3 + n + -16) * 16 < 0?-(-(2 * c3 + n + -16) / 16) : ((16 < 0?(-(2 * c3 + n + -16) + - 16 - 1) / - 16 : (2 * c3 + n + -16 + 16 - 1) / 16))))) {
115 for (c5 = (16 * c2 > 2 * c3 + 2?16 * c2 : 2 * c3 + 2); c5 <= ((16 * c2 + 15 < 2 * c3 + n + -1?16 * c2 + 15 : 2 * c3 + n + -1)); c5++) {
116 A[n + -2][-2 * c3 + c5 + -1] = B[n + -2][-2 * c3 + c5 + -1];
117 }
118 }
119 }
120 if (c0 >= (((2 * c1 + c2 + -1) * 2 < 0?-(-(2 * c1 + c2 + -1) / 2) : ((2 < 0?(-(2 * c1 + c2 + -1) + - 2 - 1) / - 2 : (2 * c1 + c2 + -1 + 2 - 1) / 2)))) && c1 >= c2 + 1 && c2 <= (((tsteps + -8) * 8 < 0?((8 < 0?-((-(tsteps + -8) + 8 + 1) / 8) : -((-(tsteps + -8) + 8 - 1) / 8))) : (tsteps + -8) / 8))) {
121 for (c4 = 16 * c1; c4 <= ((16 * c1 + 15 < 16 * c2 + n + 12?16 * c1 + 15 : 16 * c2 + n + 12)); c4++) {
122 B[-16 * c2 + c4 + -14][1] = 0.2 * (A[-16 * c2 + c4 + -14][1] + A[-16 * c2 + c4 + -14][1 - 1] + A[-16 * c2 + c4 + -14][1 + 1] + A[1 + (-16 * c2 + c4 + -14)][1] + A[-16 * c2 + c4 + -14 - 1][1]);
123 }
124 }
125 if (c0 >= (((3 * c1 + -1) * 2 < 0?-(-(3 * c1 + -1) / 2) : ((2 < 0?(-(3 * c1 + -1) + - 2 - 1) / - 2 : (3 * c1 + -1 + 2 - 1) / 2)))) && c1 <= (((((tsteps + -8) * 8 < 0?((8 < 0?-((-(tsteps + -8) + 8 + 1) / 8) : -((-(tsteps + -8) + 8 - 1) / 8))) : (tsteps + -8) / 8)) < c2?(((tsteps + -8) * 8 < 0?((8 < 0?-((-(tsteps + -8) + 8 + 1) / 8) : -((-(tsteps + -8) + 8 - 1) / 8))) : (tsteps + -8) / 8)) : c2))) {
126 for (c5 = (16 * c2 > 16 * c1 + 15?16 * c2 : 16 * c1 + 15); c5 <= ((16 * c2 + 15 < 16 * c1 + n + 12?16 * c2 + 15 : 16 * c1 + n + 12)); c5++) {
127 B[1][-16 * c1 + c5 + -14] = 0.2 * (A[1][-16 * c1 + c5 + -14] + A[1][-16 * c1 + c5 + -14 - 1] + A[1][1 + (-16 * c1 + c5 + -14)] + A[1 + 1][-16 * c1 + c5 + -14] + A[1 - 1][-16 * c1 + c5 + -14]);
128 }
129 }
130 }
131 }
132 }
133 }
134 }
135
136//#pragma endscop
137}
138
139int main(int argc,char **argv)
140{
141/* Retrieve problem size. */
142 int n = 500;
143 int tsteps = 10;
144/* Variable declaration/allocation. */
145 double (*A)[500 + 0][500 + 0];
146 A = ((double (*)[500 + 0][500 + 0])(polybench_alloc_data(((500 + 0) * (500 + 0)),(sizeof(double )))));
147 ;
148 double (*B)[500 + 0][500 + 0];
149 B = ((double (*)[500 + 0][500 + 0])(polybench_alloc_data(((500 + 0) * (500 + 0)),(sizeof(double )))));
150 ;
151/* Initialize array(s). */
152 init_array(n, *A, *B);
153/* Start timer. */
154 polybench_timer_start();
155 ;
156/* Run kernel. */
157 kernel_jacobi_2d_imper(tsteps,n, *A, *B);
158/* Stop and print timer. */
159 polybench_timer_stop();
160 ;
161 polybench_timer_print();
162 ;
163/* Prevent dead-code elimination. All live-out data must be printed
164 by the function call in argument. */
165 if (argc > 42 && !strcmp(argv[0],""))
166 print_array(n, *A);
167/* Be clean. */
168 free(((void *)A));
169 ;
170 free(((void *)B));
171 ;
172 return 0;
173}
Note: See TracBrowser for help on using the repository browser.