Debian Codegraph @

gcc/testsuite/gcc.target/aarch64/sve/slp_14.c

raw
1 /* { dg-do compile } */
2 /* { dg-options "-O2 -ftree-vectorize" } */
3
4 #include <stdint.h>
5
6 #define VEC_PERM(TYPE) \
7 void __attribute__ ((weak)) \
8 vec_slp_##TYPE (TYPE *restrict a, TYPE *restrict b, int n) \
9 { \
10 for (int i = 0; i < n; ++i) \
11 { \
12 TYPE a1 = a[i * 2]; \
13 TYPE a2 = a[i * 2 + 1]; \
14 TYPE b1 = b[i * 2]; \
15 TYPE b2 = b[i * 2 + 1]; \
16 a[i * 2] = b1 > 1 ? a1 / b1 : a1; \
17 a[i * 2 + 1] = b2 > 2 ? a2 / b2 : a2; \
18 } \
19 }
20
21 #define TEST_ALL(T) \
22 T (int32_t) \
23 T (uint32_t) \
24 T (int64_t) \
25 T (uint64_t) \
26 T (float) \
27 T (double)
28
29 TEST_ALL (VEC_PERM)
30
31 /* The loop should be fully-masked. The load XFAILs for fixed-length
32 SVE account for extra loads from the constant pool. */
33 /* { dg-final { scan-assembler-times {\tld1w\t} 6 { xfail { aarch64_sve && { ! vect_variable_length } } } } } */
34 /* { dg-final { scan-assembler-times {\tst1w\t} 3 } } */
35 /* { dg-final { scan-assembler-times {\tld1d\t} 6 { xfail { aarch64_sve && { ! vect_variable_length } } } } } */
36 /* { dg-final { scan-assembler-times {\tst1d\t} 3 } } */
37 /* { dg-final { scan-assembler-not {\tldr} } } */
38 /* { dg-final { scan-assembler-not {\tstr} } } */
39
40 /* { dg-final { scan-assembler-times {\twhilelo\tp[0-7]\.s} 6 } } */
41 /* { dg-final { scan-assembler-times {\twhilelo\tp[0-7]\.d} 6 } } */
42
43 /* { dg-final { scan-assembler-times {\tsdiv\tz[0-9]+\.s} 1 } } */
44 /* { dg-final { scan-assembler-times {\tudiv\tz[0-9]+\.s} 1 } } */
45 /* { dg-final { scan-assembler-times {\tfdiv\tz[0-9]+\.s} 1 } } */
46 /* { dg-final { scan-assembler-times {\tsdiv\tz[0-9]+\.d} 1 } } */
47 /* { dg-final { scan-assembler-times {\tudiv\tz[0-9]+\.d} 1 } } */
48 /* { dg-final { scan-assembler-times {\tfdiv\tz[0-9]+\.d} 1 } } */