Debian Codegraph @

gcc/testsuite/gcc.target/arm/simd/mve-vabs.c

raw
1 /* { dg-do assemble } */
2 /* { dg-require-effective-target arm_v8_1m_mve_fp_ok } */
3 /* { dg-add-options arm_v8_1m_mve_fp } */
4 /* { dg-additional-options "-O3 -funsafe-math-optimizations -fdump-tree-optimized" } */
5
6 #include <stdint.h>
7 #include <arm_mve.h>
8
9 #define ABS(a) ((a < 0) ? -a : a)
10
11 #define FUNC(SIGN, TYPE, BITS, NB, NAME) \
12 void test_ ## NAME ##_ ## SIGN ## BITS ## x ## NB (TYPE##BITS##_t * __restrict__ dest, TYPE##BITS##_t *a) { \
13 int i; \
14 for (i=0; i<NB; i++) { \
15 dest[i] = ABS(a[i]); \
16 } \
17 }
18
19 #define FUNC_FLOAT(SIGN, TYPE, BITS, NB, NAME) \
20 void test_ ## NAME ##_ ## SIGN ## BITS ## x ## NB (TYPE * __restrict__ dest, TYPE *a) { \
21 int i; \
22 for (i=0; i<NB; i++) { \
23 dest[i] = ABS(a[i]); \
24 } \
25 }
26
27 /* 128-bit vectors. */
28 FUNC(s, int, 32, 4, vabs)
29 FUNC(u, uint, 32, 4, vabs)
30 FUNC(s, int, 16, 8, vabs)
31 FUNC(u, uint, 16, 8, vabs)
32 FUNC(s, int, 8, 16, vabs)
33 FUNC(u, uint, 8, 16, vabs)
34 FUNC_FLOAT(f, float, 32, 4, vabs)
35 FUNC(f, float, 16, 8, vabs)
36
37 /* Taking the absolute value of an unsigned value is a no-op, so half of the
38 integer optimizations actually generate a call to memcpy, the other ones a
39 'vabs'. */
40 /* { dg-final { scan-assembler-times {vabs.s[0-9]+\tq[0-9]+, q[0-9]+} 3 } } */
41 /* { dg-final { scan-assembler-times {vabs.f[0-9]+\tq[0-9]+, q[0-9]+} 2 } } */
42 /* { dg-final { scan-assembler-times {vldr[bhw].[0-9]+\tq[0-9]+} 5 } } */
43 /* { dg-final { scan-assembler-times {vstr[bhw].[0-9]+\tq[0-9]+} 5 } } */
44 /* { dg-final { scan-tree-dump-times "memcpy" 3 "optimized" } } */