gcc/libgomp/testsuite/libgomp.c/loop-18.c
Jakub Jelinek aed3ab253d openmp: Non-rectangular loop support for non-composite worksharing loops and distribute
This implements the fallback mentioned in
https://gcc.gnu.org/pipermail/gcc/2020-June/232874.html
Special cases for triangular loops etc. to follow later, also composite
constructs not supported yet (need to check the passing of temporaries around)
and lastprivate might not give the same answers as serial loop if the last
innermost body iteration isn't the last one for some of the outer loops
(that will need to be solved separately together with rectangular loops that have no
innermost body iterations, but some of the outer loops actually iterate).
Also, simd needs work.

2020-06-27  Jakub Jelinek  <jakub@redhat.com>

	* omp-general.h (struct omp_for_data_loop): Add non_rect_referenced
	member, move outer member.
	(struct omp_for_data): Add first_nonrect and last_nonrect members.
	* omp-general.c (omp_extract_for_data): Initialize first_nonrect,
	last_nonrect and non_rect_referenced members.
	* omp-expand.c (expand_omp_for_init_counts): Handle non-rectangular
	loops.
	(expand_omp_for_init_vars): Add nonrect_bounds parameter.  Handle
	non-rectangular loops.
	(extract_omp_for_update_vars): Likewise.
	(expand_omp_for_generic, expand_omp_for_static_nochunk,
	expand_omp_for_static_chunk, expand_omp_simd,
	expand_omp_taskloop_for_outer, expand_omp_taskloop_for_inner): Adjust
	expand_omp_for_init_vars and extract_omp_for_update_vars callers.
	(expand_omp_for): Don't sorry on non-composite worksharing-loop or
	distribute.

	* testsuite/libgomp.c/loop-17.c: New test.
	* testsuite/libgomp.c/loop-18.c: New test.
2020-06-27 12:43:36 +02:00

246 lines
6.8 KiB
C

/* { dg-do run } */
extern void abort (void);
int x, i, j;
volatile int a, b, c, d, e, f, g, h;
int k[11][101];
int
main ()
{
int niters;
for (i = 1; i <= 10; i++)
for (j = 1; j <= 10 * i; j++)
k[i][j] = 1;
a = 1; b = 11; c = 1; d = 0; e = 1; f = 10; g = 1; h = 1;
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = 1; i <= 10; i++)
for (j = 1; j <= 10 * i; j++)
{
if (i < 1 || i > 10 || j < 1 || j > 10 * i || k[i][j] != 1)
abort ();
k[i][j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (i != 11 || j != 101 || x != 10340 || niters != 550)
abort ();
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = a; i < b; i += c)
for (j = d * i + e; j < g + i * f; j += h)
{
if (i < 1 || i > 10 || j < 1 || j > 10 * i || k[i][j] != 2)
abort ();
k[i][j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (i != 11 || j != 101 || x != 10340 || niters != 550)
abort ();
for (i = 1; i <= 10; i++)
for (j = 1; j <= 10 * i; j++)
if (k[i][j] == 3)
k[i][j] = 0;
else
abort ();
for (i = 0; i < 11; i++)
for (j = 0; j < 101; j++)
if (k[i][j] != 0)
abort ();
for (i = 0; i < 10; i++)
for (j = 0; j < 10 * i; j++)
k[i][j] = 1;
a = 0; b = 10; c = 1; d = 0; e = 0; f = 10; g = 0; h = 1;
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = 0; i < 10; i++)
for (j = 0; j < 10 * i; j++)
{
if (i < 0 || i >= 10 || j < 0 || j >= 10 * i || k[i][j] != 1)
abort ();
k[i][j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (i != 10 || j != 90 || x != 9305 || niters != 450)
abort ();
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = a; i < b; i += c)
for (j = d * i + e; j < g + i * f; j += h)
{
if (i < 0 || i >= 10 || j < 0 || j >= 10 * i || k[i][j] != 2)
abort ();
k[i][j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (i != 10 || j != 90 || x != 9305 || niters != 450)
abort ();
for (i = 0; i < 10; i++)
for (j = 0; j < 10 * i; j++)
if (k[i][j] == 3)
k[i][j] = 0;
else
abort ();
for (i = 0; i < 11; i++)
for (j = 0; j < 101; j++)
if (k[i][j] != 0)
abort ();
for (i = 4; i < 10; i++)
for (j = -9 + 2 * i; j < i; j++)
k[i][j + 1] = 1;
a = 4; b = 10; c = 1; d = 2; e = -9; f = 1; g = 0; h = 1;
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = 4; i < 10; i++)
for (j = -9 + 2 * i; j < i; j++)
{
if (i < 4 || i >= 10 || j < -9 + 2 * i || j >= i || k[i][j + 1] != 1)
abort ();
k[i][j + 1]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (/*i != 10 || j != 9 || */x != 8199 || niters != 15)
abort ();
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = a; i < b; i += c)
for (j = d * i + e; j < g + i * f; j += h)
{
if (i < 4 || i >= 10 || j < -9 + 2 * i || j >= i || k[i][j + 1] != 2)
abort ();
k[i][j + 1]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (/*i != 10 || j != 9 || */x != 8199 || niters != 15)
abort ();
for (i = 4; i < 10; i++)
for (j = -9 + 2 * i; j < i; j++)
if (k[i][j + 1] == 3)
k[i][j + 1] = 0;
else
abort ();
for (i = 0; i < 11; i++)
for (j = 0; j < 101; j++)
if (k[i][j] != 0)
abort ();
for (i = 1; i < 10; i += 2)
for (j = 1; j < i + 1; j++)
k[i][j] = 1;
a = 1; b = 10; c = 2; d = 0; e = 1; f = 1; g = 1; h = 1;
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = 1; i < 10; i += 2)
for (j = 1; j < i + 1; j++)
{
if (i < 1 || i >= 10 || j < 1 || j >= i + 1 || k[i][j] != 1)
abort ();
k[i][j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (i != 11 || j != 10 || x != 9225 || niters != 25)
abort ();
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = a; i < b; i += c)
for (j = d * i + e; j < g + i * f; j += h)
{
if (i < 1 || i >= 10 || j < 1 || j >= i + 1 || k[i][j] != 2)
abort ();
k[i][j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (i != 11 || j != 10 || x != 9225 || niters != 25)
abort ();
for (i = 1; i < 10; i += 2)
for (j = 1; j < i + 1; j++)
if (k[i][j] == 3)
k[i][j] = 0;
else
abort ();
for (i = 0; i < 11; i++)
for (j = 0; j < 101; j++)
if (k[i][j] != 0)
abort ();
for (j = -11; j >= -41; j -= 15)
k[0][-j] = 1;
a = 4; b = 8; c = 12; d = -8; e = -9; f = -3; g = 6; h = 15;
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = 4; i < 8; i += 12)
for (j = -8 * i - 9; j < i * -3 + 6; j += 15)
{
if (i != 4 || j < -41 || j > -11 || k[0][-j] != 1)
abort ();
k[0][-j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (i != 16 || j != 4 || x != 5109 || niters != 3)
abort ();
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = a; i < b; i += c)
for (j = d * i + e; j < g + i * f; j += h)
{
if (i != 4 || j < -41 || j > -11 || k[0][-j] != 2)
abort ();
k[0][-j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (i != 16 || j != 4 || x != 5109 || niters != 3)
abort ();
for (j = -11; j >= -41; j -= 15)
if (k[0][-j] == 3)
k[0][-j] = 0;
else
abort ();
for (j = -11; j >= -41; j--)
if (k[0][-j] != 0)
abort ();
for (j = -34; j <= -7; j++)
k[0][-j] = 1;
a = -13; b = 7; c = 12; d = 3; e = 5; f = 0; g = -6; h = 1;
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = -13; i < 7; i += 12)
for (j = 3 * i + 5; j < -6; j++)
{
if (i != -13 || j < -34 || j > -7 || k[0][-j] != 1)
abort ();
k[0][-j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (/*i != 11 || j != 2 || */x != -12295 || niters != 28)
abort ();
niters = 0; i = -100; j = -100; x = -100;
#pragma omp parallel for collapse(2) lastprivate (i, j, x) reduction(+:niters)
for (i = a; i < b; i += c)
for (j = d * i + e; j < g + i * f; j += h)
{
if (i != -13 || j < -34 || j > -7 || k[0][-j] != 2)
abort ();
k[0][-j]++;
x = i * 1024 + (j & 1023);
niters++;
}
if (/*i != 11 || j != 2 || */x != -12295 || niters != 28)
abort ();
for (j = -34; j <= -7; j++)
if (k[0][-j] == 3)
k[0][-j] = 0;
else
abort ();
return 0;
}