Several vector tests are sensitive to the vector size. This patch adds
a VECTOR_BITS macro to tree-vect.h to select the expected vector size
and uses it to influence iteration counts and array sizes. The tests
keep the original values if the vector size is small enough.
For now VECTOR_BITS is always 128, but the SVE patches add other values.
2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
Alan Hayward <alan.hayward@arm.com>
David Sherwood <david.sherwood@arm.com>
gcc/testsuite/
* gcc.dg/vect/tree-vect.h (VECTOR_BITS): Define.
* gcc.dg/vect/bb-slp-pr69907.c: Include tree-vect.h.
(N): New macro.
(foo): Use it instead of hard-coded 320.
* gcc.dg/vect/no-scevccp-outer-7.c (N): Redefine if the default
value is too small for VECTOR_BITS.
* gcc.dg/vect/no-scevccp-vect-iv-3.c (N): Likewise.
* gcc.dg/vect/no-section-anchors-vect-31.c (N): Likewise.
* gcc.dg/vect/no-section-anchors-vect-36.c (N): Likewise.
* gcc.dg/vect/slp-perm-9.c (N): Likewise.
* gcc.dg/vect/vect-32.c (N): Likewise.
* gcc.dg/vect/vect-75.c (N, OFF): Likewise.
* gcc.dg/vect/vect-77-alignchecks.c (N, OFF): Likewise.
* gcc.dg/vect/vect-78-alignchecks.c (N, OFF): Likewise.
* gcc.dg/vect/vect-89.c (N): Likewise.
* gcc.dg/vect/vect-96.c (N): Likewise.
* gcc.dg/vect/vect-multitypes-3.c (N): Likewise.
* gcc.dg/vect/vect-multitypes-6.c (N): Likewise.
* gcc.dg/vect/vect-over-widen-1.c (N): Likewise.
* gcc.dg/vect/vect-over-widen-4.c (N): Likewise.
* gcc.dg/vect/vect-reduc-pattern-1a.c (N): Likewise.
* gcc.dg/vect/vect-reduc-pattern-1b.c (N): Likewise.
* gcc.dg/vect/vect-reduc-pattern-2a.c (N): Likewise.
* gcc.dg/vect/no-section-anchors-vect-64.c (NINTS): New macro.
(N): Redefine in terms of NINTS.
(ia, ib, ic): Use NINTS instead of hard-coded constants in the
array bounds.
* gcc.dg/vect/no-section-anchors-vect-69.c (NINTS): New macro.
(N): Redefine in terms of NINTS.
(test1): Replace a and b fields with NINTS - 2 ints of padding.
(main1): Use NINTS instead of hard-coded constants.
* gcc.dg/vect/section-anchors-vect-69.c (NINTS): New macro.
(N): Redefine in terms of NINTS.
(test1): Replace a and b fields with NINTS - 2 ints of padding.
(test2): Remove incorrect comments about alignment.
(main1): Use NINTS instead of hard-coded constants.
* gcc.dg/vect/pr45752.c (N): Redefine if the default value is
too small for VECTOR_BITS.
(main): Continue to use canned results for the default value of N,
but compute the expected results from scratch for other values.
* gcc.dg/vect/slp-perm-1.c (N, main): As for pr45752.c.
* gcc.dg/vect/slp-perm-4.c (N, main): Likewise.
* gcc.dg/vect/slp-perm-5.c (N, main): Likewise.
* gcc.dg/vect/slp-perm-6.c (N, main): Likewise.
* gcc.dg/vect/slp-perm-7.c (N, main): Likewise.
* gcc.dg/vect/pr65518.c (NINTS, N, RESULT): New macros.
(giga): Use NINTS as the array bound.
(main): Use NINTS, N and RESULT.
* gcc.dg/vect/pr65947-5.c (N): Redefine if the default value is
too small for VECTOR_BITS.
(main): Fill in any remaining elements of A programmatically.
* gcc.dg/vect/pr81136.c: Include tree-vect.h.
(a): Use VECTOR_BITS to set the alignment of the target structure.
* gcc.dg/vect/slp-19c.c (N): Redefine if the default value is
too small for VECTOR_BITS.
(main1): Continue to use the canned input for the default value of N,
but compute the input from scratch for other values.
* gcc.dg/vect/slp-28.c (N): Redefine if the default value is
too small for VECTOR_BITS.
(in1, in2, in3): Remove initialization.
(check1, check2): Delete.
(main1): Initialize in1, in2 and in3 here. Check every element
of the vectors and compute the expected values directly instead
of using an array.
* gcc.dg/vect/slp-perm-8.c (N): Redefine if the default value is
too small for VECTOR_BITS.
(foo, main): Change type of "i" to int.
* gcc.dg/vect/vect-103.c (NINTS): New macro.
(N): Redefine in terms of N.
(c): Delete.
(main1): Use NINTS. Check the result from a and b directly.
* gcc.dg/vect/vect-67.c (NINTS): New macro.
(N): Redefine in terms of N.
(main1): Use NINTS for the inner array bounds.
* gcc.dg/vect/vect-70.c (NINTS, OUTERN): New macros.
(N): Redefine in terms of NINTS.
(s): Keep the outer dimensions as 4 even if N is larger than 24.
(tmp1): New variable.
(main1): Only define a local tmp1 if NINTS is relatively small.
Use OUTERN for the outer loops and NINTS for the inner loops.
* gcc.dg/vect/vect-91.c (OFF): New macro.
(a, main3): Use it.
* gcc.dg/vect/vect-92.c (NITER): New macro.
(main1, main2): Use it.
* gcc.dg/vect/vect-93.c (N): Rename to...
(N1): ...this.
(main): Update accordingly.
(N2): New macro.
(main1): Use N1 instead of 3001 and N2 insteaed of 10.
* gcc.dg/vect/vect-multitypes-1.c (NSHORTS, NINTS): New macros.
(N): Redefine in terms of NSHORTS.
(main1): Use NINTS - 1 instead of 3 and NSHORTS - 1 instead of 7.
(main): Likewise.
* gcc.dg/vect/vect-over-widen-3-big-array.c (N): Define to VECTOR_BITS.
(foo): Truncate the expected value to the type of *d.
* gcc.dg/vect/vect-peel-3.c (NINTS, EXTRA): New macros.
(ia, ib, ic, main): Use EXTRA.
(main): Use NINTS.
(RES_A, RES_B, REC_C): New macros.
(RES): Redefine as their sum.
* gcc.dg/vect/vect-reduc-or_1.c (N): New macro.
(in): Change number of elements to N.
(main): Update accordingly. Calculate the expected result.
* gcc.dg/vect/vect-reduc-or_2.c (N, in, main): As for
vect-reduc-or-1.c.
Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r254589
+2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
+ Alan Hayward <alan.hayward@arm.com>
+ David Sherwood <david.sherwood@arm.com>
+
+ * gcc.dg/vect/tree-vect.h (VECTOR_BITS): Define.
+ * gcc.dg/vect/bb-slp-pr69907.c: Include tree-vect.h.
+ (N): New macro.
+ (foo): Use it instead of hard-coded 320.
+ * gcc.dg/vect/no-scevccp-outer-7.c (N): Redefine if the default
+ value is too small for VECTOR_BITS.
+ * gcc.dg/vect/no-scevccp-vect-iv-3.c (N): Likewise.
+ * gcc.dg/vect/no-section-anchors-vect-31.c (N): Likewise.
+ * gcc.dg/vect/no-section-anchors-vect-36.c (N): Likewise.
+ * gcc.dg/vect/slp-perm-9.c (N): Likewise.
+ * gcc.dg/vect/vect-32.c (N): Likewise.
+ * gcc.dg/vect/vect-75.c (N, OFF): Likewise.
+ * gcc.dg/vect/vect-77-alignchecks.c (N, OFF): Likewise.
+ * gcc.dg/vect/vect-78-alignchecks.c (N, OFF): Likewise.
+ * gcc.dg/vect/vect-89.c (N): Likewise.
+ * gcc.dg/vect/vect-96.c (N): Likewise.
+ * gcc.dg/vect/vect-multitypes-3.c (N): Likewise.
+ * gcc.dg/vect/vect-multitypes-6.c (N): Likewise.
+ * gcc.dg/vect/vect-over-widen-1.c (N): Likewise.
+ * gcc.dg/vect/vect-over-widen-4.c (N): Likewise.
+ * gcc.dg/vect/vect-reduc-pattern-1a.c (N): Likewise.
+ * gcc.dg/vect/vect-reduc-pattern-1b.c (N): Likewise.
+ * gcc.dg/vect/vect-reduc-pattern-2a.c (N): Likewise.
+ * gcc.dg/vect/no-section-anchors-vect-64.c (NINTS): New macro.
+ (N): Redefine in terms of NINTS.
+ (ia, ib, ic): Use NINTS instead of hard-coded constants in the
+ array bounds.
+ * gcc.dg/vect/no-section-anchors-vect-69.c (NINTS): New macro.
+ (N): Redefine in terms of NINTS.
+ (test1): Replace a and b fields with NINTS - 2 ints of padding.
+ (main1): Use NINTS instead of hard-coded constants.
+ * gcc.dg/vect/section-anchors-vect-69.c (NINTS): New macro.
+ (N): Redefine in terms of NINTS.
+ (test1): Replace a and b fields with NINTS - 2 ints of padding.
+ (test2): Remove incorrect comments about alignment.
+ (main1): Use NINTS instead of hard-coded constants.
+ * gcc.dg/vect/pr45752.c (N): Redefine if the default value is
+ too small for VECTOR_BITS.
+ (main): Continue to use canned results for the default value of N,
+ but compute the expected results from scratch for other values.
+ * gcc.dg/vect/slp-perm-1.c (N, main): As for pr45752.c.
+ * gcc.dg/vect/slp-perm-4.c (N, main): Likewise.
+ * gcc.dg/vect/slp-perm-5.c (N, main): Likewise.
+ * gcc.dg/vect/slp-perm-6.c (N, main): Likewise.
+ * gcc.dg/vect/slp-perm-7.c (N, main): Likewise.
+ * gcc.dg/vect/pr65518.c (NINTS, N, RESULT): New macros.
+ (giga): Use NINTS as the array bound.
+ (main): Use NINTS, N and RESULT.
+ * gcc.dg/vect/pr65947-5.c (N): Redefine if the default value is
+ too small for VECTOR_BITS.
+ (main): Fill in any remaining elements of A programmatically.
+ * gcc.dg/vect/pr81136.c: Include tree-vect.h.
+ (a): Use VECTOR_BITS to set the alignment of the target structure.
+ * gcc.dg/vect/slp-19c.c (N): Redefine if the default value is
+ too small for VECTOR_BITS.
+ (main1): Continue to use the canned input for the default value of N,
+ but compute the input from scratch for other values.
+ * gcc.dg/vect/slp-28.c (N): Redefine if the default value is
+ too small for VECTOR_BITS.
+ (in1, in2, in3): Remove initialization.
+ (check1, check2): Delete.
+ (main1): Initialize in1, in2 and in3 here. Check every element
+ of the vectors and compute the expected values directly instead
+ of using an array.
+ * gcc.dg/vect/slp-perm-8.c (N): Redefine if the default value is
+ too small for VECTOR_BITS.
+ (foo, main): Change type of "i" to int.
+ * gcc.dg/vect/vect-103.c (NINTS): New macro.
+ (N): Redefine in terms of N.
+ (c): Delete.
+ (main1): Use NINTS. Check the result from a and b directly.
+ * gcc.dg/vect/vect-67.c (NINTS): New macro.
+ (N): Redefine in terms of N.
+ (main1): Use NINTS for the inner array bounds.
+ * gcc.dg/vect/vect-70.c (NINTS, OUTERN): New macros.
+ (N): Redefine in terms of NINTS.
+ (s): Keep the outer dimensions as 4 even if N is larger than 24.
+ (tmp1): New variable.
+ (main1): Only define a local tmp1 if NINTS is relatively small.
+ Use OUTERN for the outer loops and NINTS for the inner loops.
+ * gcc.dg/vect/vect-91.c (OFF): New macro.
+ (a, main3): Use it.
+ * gcc.dg/vect/vect-92.c (NITER): New macro.
+ (main1, main2): Use it.
+ * gcc.dg/vect/vect-93.c (N): Rename to...
+ (N1): ...this.
+ (main): Update accordingly.
+ (N2): New macro.
+ (main1): Use N1 instead of 3001 and N2 insteaed of 10.
+ * gcc.dg/vect/vect-multitypes-1.c (NSHORTS, NINTS): New macros.
+ (N): Redefine in terms of NSHORTS.
+ (main1): Use NINTS - 1 instead of 3 and NSHORTS - 1 instead of 7.
+ (main): Likewise.
+ * gcc.dg/vect/vect-over-widen-3-big-array.c (N): Define to VECTOR_BITS.
+ (foo): Truncate the expected value to the type of *d.
+ * gcc.dg/vect/vect-peel-3.c (NINTS, EXTRA): New macros.
+ (ia, ib, ic, main): Use EXTRA.
+ (main): Use NINTS.
+ (RES_A, RES_B, REC_C): New macros.
+ (RES): Redefine as their sum.
+ * gcc.dg/vect/vect-reduc-or_1.c (N): New macro.
+ (in): Change number of elements to N.
+ (main): Update accordingly. Calculate the expected result.
+ * gcc.dg/vect/vect-reduc-or_2.c (N, in, main): As for
+ vect-reduc-or-1.c.
+
2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
Alan Hayward <alan.hayward@arm.com>
David Sherwood <david.sherwood@arm.com>
/* { dg-additional-options "-O3" } */
/* { dg-require-effective-target vect_unpack } */
+#include "tree-vect.h"
+
+#if VECTOR_BITS > 512
+#define N (VECTOR_BITS * 10 / 16)
+#else
+#define N 320
+#endif
+
void foo(unsigned *p1, unsigned short *p2)
{
int n;
- for (n = 0; n < 320; n++)
+ for (n = 0; n < N; n++)
p1[n] = p2[n * 2];
}
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 256
+#define N (VECTOR_BITS / 16)
+#else
#define N 16
+#endif
unsigned short in[N];
unsigned short coef[N];
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 256
+#define N (VECTOR_BITS / 16 + 10)
+#else
#define N 26
+#endif
__attribute__ ((noinline))
unsigned int main1 ()
#include <stdarg.h>
#include "tree-vect.h"
+/* N / 2 bytes has to be worth vectorizing even with peeling. */
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 4 / 8)
+#else
#define N 64
+#endif
struct t{
int k[N];
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 8)
+#else
#define N 32
+#endif
struct {
char ca[N];
#include <stdarg.h>
#include "tree-vect.h"
-#define N 16
+#if VECTOR_BITS > 128
+#define NINTS (VECTOR_BITS / 32)
+#else
+#define NINTS 4
+#endif
+
+#define N (NINTS * 4)
int ib[N] = {0,3,6,9,12,15,18,21,24,27,30,33,36,39,42,45};
-int ia[N][4][N+1];
-int ic[N][N][3][2*N+2];
-int id[N][N][N+4];
+int ia[N][NINTS][N + 1];
+int ic[N][N][NINTS - 1][2 * N + 2];
+int id[N][N][N + NINTS];
__attribute__ ((noinline))
int main1 ()
#include <stdarg.h>
#include "tree-vect.h"
-#define N 24
+#if VECTOR_BITS > 128
+#define NINTS (VECTOR_BITS / 32)
+#else
+#define NINTS 4
+#endif
+
+#define N (NINTS * 6)
struct s{
int m;
struct test1{
struct s a; /* array a.n is unaligned */
- int b;
- int c;
+ int pad[NINTS - 2];
struct s e; /* array e.n is aligned */
};
}
/* 2. aligned */
- for (i = 3; i < N-1; i++)
+ for (i = NINTS - 1; i < N - 1; i++)
{
tmp1[2].a.n[1][2][i] = 6;
}
/* check results: */
- for (i = 3; i < N-1; i++)
+ for (i = NINTS; i < N - 1; i++)
{
if (tmp1[2].a.n[1][2][i] != 6)
abort ();
}
/* 4. unaligned (unknown misalignment) */
- for (i = 0; i < N-4; i++)
+ for (i = 0; i < N - NINTS; i++)
{
- for (j = 0; j < N-4; j++)
+ for (j = 0; j < N - NINTS; j++)
{
tmp2[2].e.n[1][i][j] = 8;
}
}
/* check results: */
- for (i = 0; i < N-4; i++)
+ for (i = 0; i < N - NINTS; i++)
{
- for (j = 0; j < N-4; j++)
+ for (j = 0; j < N - NINTS; j++)
{
if (tmp2[2].e.n[1][i][j] != 8)
abort ();
#define M34 7716
#define M44 16
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 5 / 32)
+#else
#define N 20
+#endif
void foo (unsigned int *__restrict__ pInput,
unsigned int *__restrict__ pOutput,
int main (int argc, const char* argv[])
{
unsigned int input[N], output[N], i, input2[N], output2[N];
- unsigned int check_results[N]
- = {3208, 1334, 28764, 35679, 2789, 13028, 4754, 168364, 91254, 12399,
- 22848, 8174, 307964, 146829, 22009, 32668, 11594, 447564, 202404, 31619 };
- unsigned int check_results2[N]
- = {7136, 2702, 84604, 57909, 6633, 16956, 6122, 224204, 113484, 16243,
- 26776, 9542, 363804, 169059, 25853, 36596, 12962, 503404, 224634, 35463 };
check_vect ();
__asm__ volatile ("");
}
+#if N == 20
+ unsigned int check_results[N]
+ = { 3208, 1334, 28764, 35679, 2789, 13028, 4754, 168364, 91254, 12399,
+ 22848, 8174, 307964, 146829, 22009, 32668, 11594, 447564, 202404,
+ 31619 };
+ unsigned int check_results2[N]
+ = { 7136, 2702, 84604, 57909, 6633, 16956, 6122, 224204, 113484, 16243,
+ 26776, 9542, 363804, 169059, 25853, 36596, 12962, 503404, 224634,
+ 35463 };
+#else
+ volatile unsigned int check_results[N];
+ volatile unsigned int check_results2[N];
+
+ for (i = 0; i < N / 5; i++)
+ {
+ unsigned int a = input[i * 5];
+ unsigned int b = input[i * 5 + 1];
+ unsigned int c = input[i * 5 + 2];
+ unsigned int d = input[i * 5 + 3];
+ unsigned int e = input[i * 5 + 4];
+
+ check_results[i * 5] = M00 * a + M01 * b + M02 * c + M03 * d + M04 * e;
+ check_results[i * 5 + 1] = (M10 * a + M11 * b + M12 * c
+ + M13 * d + M14 * e);
+ check_results[i * 5 + 2] = (M20 * a + M21 * b + M22 * c
+ + M23 * d + M24 * e);
+ check_results[i * 5 + 3] = (M30 * a + M31 * b + M32 * c
+ + M33 * d + M34 * e);
+ check_results[i * 5 + 4] = (M40 * a + M41 * b + M42 * c
+ + M43 * d + M44 * e);
+
+ a = input2[i * 5];
+ b = input2[i * 5 + 1];
+ c = input2[i * 5 + 2];
+ d = input2[i * 5 + 3];
+ e = input2[i * 5 + 4];
+
+ check_results2[i * 5] = M00 * a + M01 * b + M02 * c + M03 * d + M04 * e;
+ check_results2[i * 5 + 1] = (M10 * a + M11 * b + M12 * c
+ + M13 * d + M14 * e);
+ check_results2[i * 5 + 2] = (M20 * a + M21 * b + M22 * c
+ + M23 * d + M24 * e);
+ check_results2[i * 5 + 3] = (M30 * a + M31 * b + M32 * c
+ + M33 * d + M34 * e);
+ check_results2[i * 5 + 4] = (M40 * a + M41 * b + M42 * c
+ + M43 * d + M44 * e);
+
+ asm volatile ("" ::: "memory");
+ }
+#endif
+
foo (input, output, input2, output2);
for (i = 0; i < N; i++)
#include "tree-vect.h"\r
\r
+#if VECTOR_BITS > 256\r
+#define NINTS (VECTOR_BITS / 32)\r
+#else\r
+#define NINTS 8\r
+#endif\r
+\r
+#define N (NINTS * 2)\r
+#define RESULT (NINTS * (NINTS - 1) / 2 * N + NINTS)\r
+\r
extern void abort (void);\r
\r
typedef struct giga\r
{\r
- unsigned int g[16];\r
+ unsigned int g[N];\r
} giga;\r
\r
unsigned long __attribute__((noinline,noclone))\r
\r
int main ()\r
{\r
- struct giga g[8];\r
+ struct giga g[NINTS];\r
unsigned int n = 1;\r
int i, j;\r
check_vect ();\r
- for (i = 0; i < 8; ++i)\r
- for (j = 0; j < 16; ++j)\r
+ for (i = 0; i < NINTS; ++i)\r
+ for (j = 0; j < N; ++j)\r
{\r
g[i].g[j] = n++;\r
__asm__ volatile ("");\r
}\r
- if (addfst (g, 8) != 456)\r
+ if (addfst (g, NINTS) != RESULT)\r
abort ();\r
return 0;\r
}\r
extern void abort (void) __attribute__ ((noreturn));
+#if VECTOR_BITS > 256
+#define N (VECTOR_BITS / 8)
+#else
#define N 32
+#endif
/* Condition reduction where loop size is not known at compile time. Will fail
to vectorize. Version inlined into main loop will vectorize. */
21, 22, 23, 24, 25, 26, 27, 28, 29, 30,
31, 32
};
+ for (int i = 32; i < N; ++i)
+ {
+ a[i] = 70 + (i & 3);
+ asm volatile ("" ::: "memory");
+ }
check_vect ();
/* { dg-do compile } */
-struct __attribute__((aligned (32)))
+#include "tree-vect.h"
+
+struct __attribute__((aligned (VECTOR_BITS / 8)))
{
char misaligner;
int foo[100];
#include <stdarg.h>
#include "tree-vect.h"
-#define N 32
+#if VECTOR_BITS > 128
+#define NINTS (VECTOR_BITS / 32)
+#else
+#define NINTS 4
+#endif
+
+#define N (NINTS * 8)
struct s{
int m;
struct test1{
struct s a; /* array a.n is unaligned */
- int b;
- int c;
+ int pad[NINTS - 2];
struct s e; /* array e.n is aligned */
};
struct test2{
- struct s2 a; /* array a.n is unaligned */
+ struct s2 a;
int b;
int c;
- struct s2 e; /* array e.n is aligned */
+ struct s2 e;
};
}
/* 2. aligned */
- for (i = 3; i < N-1; i++)
+ for (i = NINTS - 1; i < N - 1; i++)
{
tmp1[2].a.n[1][2][i] = 6;
}
/* check results: */
- for (i = 3; i < N-1; i++)
+ for (i = NINTS - 1; i < N - 1; i++)
{
if (tmp1[2].a.n[1][2][i] != 6)
abort ();
}
/* 4. unaligned */
- for (i = 0; i < N-4; i++)
+ for (i = 0; i < N - NINTS; i++)
{
- for (j = 0; j < N-4; j++)
+ for (j = 0; j < N - NINTS; j++)
{
tmp2[2].e.n[1][i][j] = 8;
}
}
/* check results: */
- for (i = 0; i < N-4; i++)
+ for (i = 0; i < N - NINTS; i++)
{
- for (j = 0; j < N-4; j++)
+ for (j = 0; j < N - NINTS; j++)
{
if (tmp2[2].e.n[1][i][j] != 8)
abort ();
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 32)
+#else
#define N 16
+#endif
int
main1 ()
{
unsigned int i;
unsigned int out[N*8];
+#if N == 16
unsigned int in[N*8] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63};
+#else
+ unsigned int in[N * 8];
+ for (unsigned int i = 0; i < N * 8; ++i)
+ {
+ in[i] = i & 63;
+ asm volatile ("" ::: "memory");
+ }
+#endif
unsigned int ia[N*2], a0, a1, a2, a3;
/* The last stmt requires interleaving of not power of 2 size - not
#include <stdarg.h>
#include "tree-vect.h"
-#define N 32
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 4 / 16)
+#else
+#define N 32
+#endif
-unsigned short in[N] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31};
-unsigned short in2[N] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31};
-unsigned short in3[N] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31};
-unsigned short check[N] = {0,1,2,3,5,6,7,8,10,11,12,13,15,16,17,18,20,21,22,23,25,26,27,28,30,31,32,33,35,36,37,38};
-unsigned short check3[N] = {0,1,2,3,4,5,6,7,8,9,10,11,5,6,7,8,9,10,11,12,13,14,15,16,10,11,12,13,14,15,16,17};
+unsigned short in[N] = {};
+unsigned short in2[N] = {};
+unsigned short in3[N] = {};
int
main1 ()
{
int i;
+ for (i = 0; i < N; i++)
+ {
+ in[i] = in2[i] = in3[i] = i;
+ asm volatile ("" ::: "memory");
+ }
+
for (i = 0; i < N/4; i++)
{
in[i*4] = in[i*4] + 5;
}
/* check results: */
- for (i = 4; i < N; i++)
+ for (i = 0; i < N; i++)
{
- if (in2[i] != check[i])
+ if (in2[i] != (i % 4) + (i / 4) * 5)
abort ();
}
}
/* check results: */
- for (i = 12; i < N; i++)
+ for (i = 0; i < N; i++)
{
- if (in3[i] != check3[i])
+ if (in3[i] != (i % 12) + (i / 12) * 5)
abort ();
}
#define M12 191
#define M22 500
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 3 / 32 + 4)
+#else
#define N 16
+#endif
void foo (unsigned int *__restrict__ pInput, unsigned int *__restrict__ pOutput)
{
int main (int argc, const char* argv[])
{
unsigned int input[N], output[N], i;
- unsigned int check_results[N] = {1470, 395, 28271, 5958, 1655, 111653, 10446, 2915, 195035, 14934, 4175, 278417, 19422, 5435, 361799, 0};
check_vect ();
__asm__ volatile ("");
}
+#if N == 16
+ unsigned int check_results[N] = {1470, 395, 28271, 5958, 1655, 111653, 10446, 2915, 195035, 14934, 4175, 278417, 19422, 5435, 361799, 0};
+#else
+ volatile unsigned int check_results[N] = {};
+
+ for (unsigned int i = 0; i < N / 3; i++)
+ {
+ unsigned int a = input[i * 3];
+ unsigned int b = input[i * 3 + 1];
+ unsigned int c = input[i * 3 + 2];
+
+ check_results[i * 3] = M00 * a + M01 * b + M02 * c;
+ check_results[i * 3 + 1] = M10 * a + M11 * b + M12 * c;
+ check_results[i * 3 + 2] = M20 * a + M21 * b + M22 * c;
+
+ asm volatile ("" ::: "memory");
+ }
+#endif
+
foo (input, output);
for (i = 0; i < N; i++)
#define M34 7716
#define M44 16
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 5 / 32)
+#else
#define N 20
+#endif
void foo (unsigned int *__restrict__ pInput, unsigned int *__restrict__ pOutput)
{
int main (int argc, const char* argv[])
{
unsigned int input[N], output[N], i;
- unsigned int check_results[N]
- = {3208, 1334, 28764, 35679, 2789, 13028, 4754, 168364, 91254, 12399,
- 22848, 8174, 307964, 146829, 22009, 32668, 11594, 447564, 202404, 31619};
check_vect ();
asm volatile ("" ::: "memory");
}
+#if N == 20
+ unsigned int check_results[N]
+ = {3208, 1334, 28764, 35679, 2789, 13028, 4754, 168364, 91254, 12399,
+ 22848, 8174, 307964, 146829, 22009, 32668, 11594, 447564, 202404, 31619};
+#else
+ volatile unsigned int check_results[N];
+
+ for (i = 0; i < N / 5; i++)
+ {
+ unsigned int a = input[i * 5];
+ unsigned int b = input[i * 5 + 1];
+ unsigned int c = input[i * 5 + 2];
+ unsigned int d = input[i * 5 + 3];
+ unsigned int e = input[i * 5 + 4];
+
+ check_results[i * 5] = M00 * a + M01 * b + M02 * c + M03 * d + M04 * e;
+ check_results[i * 5 + 1] = (M10 * a + M11 * b + M12 * c
+ + M13 * d + M14 * e);
+ check_results[i * 5 + 2] = (M20 * a + M21 * b + M22 * c
+ + M23 * d + M24 * e);
+ check_results[i * 5 + 3] = (M30 * a + M31 * b + M32 * c
+ + M33 * d + M34 * e);
+ check_results[i * 5 + 4] = (M40 * a + M41 * b + M42 * c
+ + M43 * d + M44 * e);
+ asm volatile ("");
+ }
+#endif
+
foo (input, output);
for (i = 0; i < N; i++)
#define K01 4322
#define K11 135
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 3 / 32 + 4)
+#else
#define N 16
+#endif
void foo (int *__restrict__ pInput, int *__restrict__ pOutput,
int *__restrict__ pInput2, int *__restrict__ pOutput2)
int main (int argc, const char* argv[])
{
int input[N], output[N], i;
- int check_results[N] = {1470, 395, 28271, 5958, 1655, 111653, 10446, 2915, 195035, 14934, 4175, 278417, 19422, 5435, 361799, 0};
int input2[N], output2[N];
- int check_results2[N] = {4322, 135, 13776, 629, 23230, 1123, 32684, 1617, 42138, 2111, 0, 0, 0, 0, 0, 0};
check_vect ();
__asm__ volatile ("");
}
+#if N == 16
+ int check_results[N] = { 1470, 395, 28271, 5958, 1655, 111653, 10446, 2915,
+ 195035, 14934, 4175, 278417, 19422, 5435, 361799,
+ 0 };
+ int check_results2[N] = { 4322, 135, 13776, 629, 23230, 1123, 32684, 1617,
+ 42138, 2111, 0, 0, 0, 0, 0, 0 };
+#else
+ volatile int check_results[N] = {};
+ volatile int check_results2[N] = {};
+
+ for (int i = 0; i < N / 3; i++)
+ {
+ int a = input[i * 3];
+ int b = input[i * 3 + 1];
+ int c = input[i * 3 + 2];
+ int d = input2[i * 2];
+ int e = input2[i * 2 + 1];
+
+ check_results[i * 3] = M00 * a + M01 * b + M02 * c;
+ check_results[i * 3 + 1] = M10 * a + M11 * b + M12 * c;
+ check_results[i * 3 + 2] = M20 * a + M21 * b + M22 * c;
+
+ check_results2[i * 2] = K00 * d + K01 * e;
+ check_results2[i * 2 + 1] = K10 * d + K11 * e;
+
+ asm volatile ("" ::: "memory");
+ }
+#endif
+
foo (input, output, input2, output2);
for (i = 0; i < N; i++)
#define K01 4322
#define K11 135
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 3 / 32 + 4)
+#else
#define N 16
+#endif
void foo (int *__restrict__ pInput, int *__restrict__ pOutput,
int *__restrict__ pInput2, int *__restrict__ pOutput2)
int main (int argc, const char* argv[])
{
int input[N], output[N], i;
- int check_results[N] = {1470, 395, 28271, 5958, 1655, 111653, 10446, 2915, 195035, 14934, 4175, 278417, 19422, 5435, 361799, 0};
int input2[N], output2[N];
- int check_results2[N] = {0, 112, 810, 336, 1620, 560, 2430, 784, 3240, 1008, 0, 0, 0, 0, 0, 0};
check_vect ();
__asm__ volatile ("");
}
+#if N == 16
+ int check_results[N] = { 1470, 395, 28271, 5958, 1655, 111653, 10446, 2915,
+ 195035, 14934, 4175, 278417, 19422, 5435, 361799,
+ 0 };
+ int check_results2[N] = { 0, 112, 810, 336, 1620, 560, 2430, 784, 3240, 1008,
+ 0, 0, 0, 0, 0, 0 };
+#else
+ volatile int check_results[N] = {};
+ volatile int check_results2[N] = {};
+
+ for (int i = 0; i < N / 3; i++)
+ {
+ int a = input[i * 3];
+ int b = input[i * 3 + 1];
+ int c = input[i * 3 + 2];
+ int d = input2[i * 2];
+ int e = input2[i * 2 + 1];
+
+ check_results[i * 3] = M00 * a + M01 * b + M02 * c;
+ check_results[i * 3 + 1] = M10 * a + M11 * b + M12 * c;
+ check_results[i * 3 + 2] = M20 * a + M21 * b + M22 * c;
+
+ check_results2[i * 2] = K00 * d;
+ check_results2[i * 2 + 1] = K10 * e;
+
+ asm volatile ("" ::: "memory");
+ }
+#endif
+
foo (input, output, input2, output2);
for (i = 0; i < N; i++)
#define K01 4322
#define K11 135
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 3 / 32 + 4)
+#else
#define N 16
+#endif
/* SLP with load permutation and loop-based vectorization. */
void foo (int *__restrict__ pInput, int *__restrict__ pOutput,
int main (int argc, const char* argv[])
{
int input[N], output[N], i;
- int check_results[N] = {1470, 395, 28271, 5958, 1655, 111653, 10446, 2915, 195035, 14934, 4175, 278417, 19422, 5435, 361799, 0};
int input2[N], output2[N];
- int check_results2[N] = {0, 405, 810, 1215, 1620, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
check_vect ();
asm volatile ("" ::: "memory");
}
+#if N == 16
+ int check_results[N] = {1470, 395, 28271, 5958, 1655, 111653, 10446, 2915, 195035, 14934, 4175, 278417, 19422, 5435, 361799, 0};
+ int check_results2[N] = {0, 405, 810, 1215, 1620, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
+#else
+ volatile int check_results[N] = {};
+ volatile int check_results2[N] = {};
+
+ for (int i = 0; i < N / 3; i++)
+ {
+ int a = input[i * 3];
+ int b = input[i * 3 + 1];
+ int c = input[i * 3 + 2];
+ int d = input2[i];
+
+ check_results[i * 3] = M00 * a + M01 * b + M02 * c;
+ check_results[i * 3 + 1] = M10 * a + M11 * b + M12 * c;
+ check_results[i * 3 + 2] = M20 * a + M21 * b + M22 * c;
+
+ check_results2[i] = K00 * d;
+
+ asm volatile ("" ::: "memory");
+ }
+#endif
+
foo (input, output, input2, output2);
for (i = 0; i < N; i++)
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 512
+#define N (VECTOR_BITS * 6 / 16)
+#else
#define N 200
+#endif
void __attribute__((noinline))
foo (unsigned char *__restrict__ pInput, unsigned char *__restrict__ pOutput)
{
- unsigned char i, a, b, c;
+ unsigned char a, b, c;
+ unsigned int i;
for (i = 0; i < N / 3; i++)
{
int main (int argc, const char* argv[])
{
- unsigned char input[N], output[N], i;
+ unsigned char input[N], output[N];
unsigned char check_results[N];
+ unsigned int i;
check_vect ();
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 512
+#define N (VECTOR_BITS * 6 / 16)
+#else
#define N 200
+#endif
void __attribute__((noinline))
foo (unsigned short *__restrict__ pInput, unsigned short *__restrict__ pOutput)
#endif
signal (SIGILL, SIG_DFL);
}
+
+#define VECTOR_BITS 128
#include <stdarg.h>
#include "tree-vect.h"
-#define N 9
+#if VECTOR_BITS > 256
+#define NINTS (VECTOR_BITS / 32)
+#else
+#define NINTS 8
+#endif
+
+#define N (NINTS + 1)
struct extraction
{
static int a[N] = {1,2,3,4,5,6,7,8,9};
static int b[N] = {17,24,7,0,2,3,4,31,82};
-static int c[N] = {9,17,24,7,0,2,3,4,31};
__attribute__ ((noinline))
int main1 (int x, int y) {
/* Vectorizable: distance > VF. */
for (i = 0; i < N; i++)
- {
- *((int *)p + x + i) = *((int *)p + x + i + 8);
- }
+ *((int *)p + x + i) = *((int *)p + x + i + NINTS);
/* check results: */
- for (i = 0; i < N; i++)
- {
- if (p->a[i] != c[i])
- abort();
- }
+ if (p->a[0] != a[N - 1])
+ abort ();
+ for (i = 1; i < N; i++)
+ if (p->a[i] != b[i - 1])
+ abort ();
+
return 0;
}
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS / 8)
+#else
#define N 16
-
+#endif
+
__attribute__ ((noinline))
int main1 ()
{
#include <stdarg.h>
#include "tree-vect.h"
-#define N 16
+#if VECTOR_BITS > 256
+#define NINTS (VECTOR_BITS / 32)
+#else
+#define NINTS 8
+#endif
+
+#define N (NINTS * 2)
__attribute__ ((noinline))
int main1 (int a, int b)
{
int i, j;
- int ia[N][4][N+8];
+ int ia[N][4][N + NINTS];
/* Multidimensional array. Aligned. The "inner" dimensions
are invariant in the inner loop. Store.
{
for (j = 0; j < N; j++)
{
- ia[i][1][j+8] = (a == b);
+ ia[i][1][j + NINTS] = (a == b);
}
}
{
for (j = 0; j < N; j++)
{
- if (ia[i][1][j+8] != (a == b))
+ if (ia[i][1][j + NINTS] != (a == b))
abort();
}
}
#include <stdarg.h>
#include "tree-vect.h"
-#define N 24
+#if VECTOR_BITS > 128
+#define NINTS (VECTOR_BITS / 32)
+#else
+#define NINTS 4
+#endif
+
+#define N (NINTS * 6)
+
+/* Keep execution time down. */
+#if N <= 24
+#define OUTERN N
+#else
+#define OUTERN NINTS
+#endif
struct s{
int m;
- int n[N/6][N/6][N];
+ int n[4][4][N];
};
struct test1{
struct s e[N]; /* array e.n is aligned */
};
+/* Avoid big local temporaries. */
+#if NINTS > 8
+struct test1 tmp1;
+#endif
+
__attribute__ ((noinline))
int main1 ()
{
int i,j;
+#if NINTS <= 8
struct test1 tmp1;
+#endif
- for (i = 0; i < N; i++)
- for (j = 3; j < N-3; j++)
+ for (i = 0; i < OUTERN; i++)
+ for (j = NINTS - 1; j < N - NINTS + 1; j++)
{
tmp1.e[i].n[1][2][j] = 8;
}
/* check results: */
- for (i = 0; i < N; i++)
- for (j = 3; j < N-3; j++)
+ for (i = 0; i < OUTERN; i++)
+ for (j = NINTS - 1; j < N - NINTS + 1; j++)
{
if (tmp1.e[i].n[1][2][j] != 8)
abort ();
}
/* not consecutive, will use strided stores */
- for (i = 0; i < N; i++)
- for (j = 3; j < N-3; j++)
+ for (i = 0; i < OUTERN; i++)
+ for (j = NINTS - 1; j < N - NINTS + 1; j++)
{
tmp1.e[j].n[1][2][j] = 8;
}
/* check results: */
- for (i = 0; i < N; i++)
- for (j = 3; j < N-3; j++)
+ for (i = 0; i < OUTERN; i++)
+ for (j = NINTS - 1; j < N - NINTS + 1; j++)
{
if (tmp1.e[j].n[1][2][j] != 8)
abort ();
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 32)
+#define OFF (VECTOR_BITS / 32)
+#else
#define N 8
#define OFF 8
+#endif
/* Check handling of accesses for which the "initial condition" -
the expression that represents the first location accessed - is
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 32)
+#define OFF (VECTOR_BITS / 32)
+#else
#define N 8
#define OFF 8
+#endif
/* Check handling of accesses for which the "initial condition" -
the expression that represents the first location accessed - is
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 32)
+#define OFF (VECTOR_BITS / 32)
+#else
#define N 8
#define OFF 8
+#endif
/* Check handling of accesses for which the "initial condition" -
the expression that represents the first location accessed - is
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 256
+#define N (VECTOR_BITS * 2 / 32)
+#else
#define N 16
+#endif
struct tmp_struct
{
#define N 256
-extern int a[N+20];
+/* Pick a value greater than the vector length. */
+#if VECTOR_BITS > 128
+#define OFF (VECTOR_BITS * 5 / 32)
+#else
+#define OFF 20
+#endif
+
+extern int a[N + OFF];
/* The alignment of 'pa' is unknown.
Yet we do know that both the read access and write access have
for (i = 0; i < N; i++)
{
- a[i] = a[i+20];
+ a[i] = a[i + OFF];
}
return 0;
can use this information (generate prolog and epilog loops
with known number of iterations, and only if needed). */
+#if VECTOR_BITS > 128
+#define NITER (VECTOR_BITS * 3 / 32)
+#else
+#define NITER 12
+#endif
+
__attribute__ ((noinline)) int
main1 ()
{
int i;
- for (i = 0; i < 10; i++)
+ for (i = 0; i < NITER - 2; i++)
{
pa[i+1] = pb[i+1] * pc[i+1];
}
{
int i;
- for (i = 0; i < 12; i++)
+ for (i = 0; i < NITER; i++)
{
pa[i+1] = pb[i+1] * pc[i+1];
}
#include <stdarg.h>
#include "tree-vect.h"
-#define N 3001
+#define N1 3001
+#if VECTOR_BITS > 256
+#define N2 (VECTOR_BITS / 32 + 2)
+#else
+#define N2 10
+#endif
__attribute__ ((noinline)) int
main1 (float *pa)
{
int i;
- for (i = 0; i < 3001; i++)
+ for (i = 0; i < N1; i++)
{
pa[i] = 2.0;
}
/* check results: */
- for (i = 0; i < 3001; i++)
+ for (i = 0; i < N1; i++)
{
if (pa[i] != 2.0)
abort ();
}
- for (i = 1; i <= 10; i++)
+ for (i = 1; i <= N2; i++)
{
pa[i] = 3.0;
}
/* check results: */
- for (i = 1; i <= 10; i++)
+ for (i = 1; i <= N2; i++)
{
if (pa[i] != 3.0)
abort ();
int main (void)
{
int i;
- float a[N] __attribute__ ((__aligned__(__BIGGEST_ALIGNMENT__)));
- float b[N] __attribute__ ((__aligned__(__BIGGEST_ALIGNMENT__)));
+ float a[N1] __attribute__ ((__aligned__(__BIGGEST_ALIGNMENT__)));
+ float b[N1] __attribute__ ((__aligned__(__BIGGEST_ALIGNMENT__)));
check_vect ();
/* from bzip2: */
- for (i=0; i<N; i++) b[i] = i;
+ for (i = 0; i < N1; i++)
+ b[i] = i;
a[0] = 0;
for (i = 1; i <= 256; i++) a[i] = b[i-1];
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 256
+#define N (VECTOR_BITS * 2 / 32)
+#else
#define N 16
+#endif
struct tmp
{
#include <stdarg.h>
#include "tree-vect.h"
-#define N 32
+#if VECTOR_BITS > 128
+#define NSHORTS (VECTOR_BITS / 16)
+#else
+#define NSHORTS 8
+#endif
+
+#define NINTS (NSHORTS / 2)
+#define N (NSHORTS * 4)
short sa[N];
short sb[N] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,
copmutations. Vectorizable. */
for (i = 0; i < n; i++)
{
- sa[i+7] = sb[i];
- ia[i+3] = ib[i+1];
+ sa[i + NSHORTS - 1] = sb[i];
+ ia[i + NINTS - 1] = ib[i + 1];
}
/* check results: */
for (i = 0; i < n; i++)
{
- if (sa[i+7] != sb[i] || ia[i+3] != ib[i+1])
+ if (sa[i + NSHORTS - 1] != sb[i] || ia[i + NINTS - 1] != ib[i + 1])
abort ();
}
copmutations. */
for (i = 0; i < n; i++)
{
- ia[i+3] = ib[i];
- sa[i+3] = sb[i+1];
+ ia[i + NINTS - 1] = ib[i];
+ sa[i + NINTS - 1] = sb[i + 1];
}
/* check results: */
for (i = 0; i < n; i++)
{
- if (sa[i+3] != sb[i+1] || ia[i+3] != ib[i])
+ if (sa[i + NINTS - 1] != sb[i + 1] || ia[i + NINTS - 1] != ib[i])
abort ();
}
{
check_vect ();
- main1 (N-7);
- main2 (N-3);
+ main1 (N - NSHORTS + 1);
+ main2 (N - NINTS + 1);
return 0;
}
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 8)
+#else
#define N 32
+#endif
int ib[N] __attribute__ ((__aligned__(__BIGGEST_ALIGNMENT__))) =
{0,3,6,9,12,15,18,21,24,27,30,33,36,39,42,45,0,3,6,9,12,15,18,21,24,27,30,33,36,39,42,45};
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 8)
+#else
#define N 32
+#endif
unsigned int ic[N] __attribute__ ((__aligned__(__BIGGEST_ALIGNMENT__))) =
{0,3,6,9,12,15,18,21,24,27,30,33,36,39,42,45,0,3,6,9,12,15,18,21,24,27,30,33,36,39,42,45};
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 8 / 16)
+#else
#define N 64
+#endif
/* Modified rgb to rgb conversion from FFmpeg. */
__attribute__ ((noinline)) void
#include <stdarg.h>
#include "tree-vect.h"
-#define N 128
+#define N VECTOR_BITS
/* Modified rgb to rgb conversion from FFmpeg. */
__attribute__ ((noinline)) void
const int g = *s++;
const int r = *s++;
const int a = *s++;
- if (*d != ((b>>3) | ((g&0xFFC)<<3) | ((r+0xF8)>>8) | (a<<9)))
+ unsigned short expected
+ = ((b>>3) | ((g&0xFFC)<<3) | ((r+0xF8)>>8) | (a<<9));
+ if (*d != expected)
abort ();
d++;
}
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 8 / 16)
+#else
#define N 64
+#endif
/* Modified rgb to rgb conversion from FFmpeg. */
__attribute__ ((noinline)) int
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define NINTS (VECTOR_BITS / 32)
+#define EXTRA (NINTS * 2)
+#else
+#define NINTS 4
+#define EXTRA 10
+#endif
+
#define N 128
-#define RES 21640
-int ib[N+10];
-int ia[N+10];
-int ic[N+10];
+#define RES_A (N * N / 4)
+#define RES_B (N * (N + 1) / 2 + (NINTS + 3) * (N + 1))
+#define RES_C (N * (N + 1) / 2 + (N + 1))
+#define RES (RES_A + RES_B + RES_C)
+
+int ib[N + EXTRA];
+int ia[N + EXTRA];
+int ic[N + EXTRA];
__attribute__ ((noinline))
int main1 ()
for (i = 0; i <= N; i++)
{
suma += ia[i];
- sumb += ib[i+5];
- sumc += ic[i+1];
+ sumb += ib[i + NINTS + 1];
+ sumc += ic[i + 1];
}
/* check results: */
check_vect ();
- for (i = 0; i < N+10; i++)
+ for (i = 0; i < N + EXTRA; i++)
{
asm volatile ("" : "+r" (i));
ib[i] = i;
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS / 8)
+#else
+#define N 16
+#endif
+
extern void abort(void);
-unsigned char in[16] __attribute__((__aligned__(16)));
+unsigned char in[N] __attribute__((__aligned__(16)));
int
main (unsigned char argc, char **argv)
{
unsigned char i = 0;
unsigned char sum = 1;
+ unsigned char expected = 1;
check_vect ();
- for (i = 0; i < 16; i++)
+ for (i = 0; i < N; i++)
in[i] = (i + i + 1) & 0xfd;
+ for (i = 0; i < N; i++)
+ {
+ expected |= in[i];
+ asm volatile ("");
+ }
+
/* Prevent constant propagation of the entire loop below. */
asm volatile ("" : : : "memory");
- for (i = 0; i < 16; i++)
+ for (i = 0; i < N; i++)
sum |= in[i];
- if (sum != 29)
+ if (sum != expected)
{
__builtin_printf("Failed %d\n", sum);
abort();
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS / 8)
+#else
+#define N 16
+#endif
+
extern void abort(void);
int
main (unsigned char argc, char **argv)
{
- unsigned char in[16] __attribute__((aligned(16)));
+ unsigned char in[N] __attribute__((aligned(16)));
unsigned char i = 0;
unsigned char sum = 1;
+ unsigned char expected = 1;
check_vect ();
- for (i = 0; i < 16; i++)
+ for (i = 0; i < N; i++)
in[i] = (i + i + 1) & 0xfd;
- for (i = 0; i < 16; i++)
+ for (i = 0; i < N; i++)
+ {
+ expected |= in[i];
+ asm volatile ("");
+ }
+
+ for (i = 0; i < N; i++)
sum |= in[i];
- if (sum != 29)
+ if (sum != expected)
{
__builtin_printf("Failed %d\n", sum);
abort();
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 16)
+#else
#define N 16
+#endif
+
unsigned short udata_sh[N] =
{ 0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28 };
#define SUM 210
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 16)
+#else
#define N 16
+#endif
+
unsigned char udata_ch[N] =
{ 0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28 };
#define SUM 210
#include <stdarg.h>
#include "tree-vect.h"
+#if VECTOR_BITS > 128
+#define N (VECTOR_BITS * 2 / 16)
+#else
#define N 16
+#endif
+
signed short data_sh[N] =
{ 0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28 };
#define SUM 210
projects / gcc.git / commitdiff