OpenE2K/gcc
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

AArch64: Implement missing vceq*_p* intrinsics

Browse Source 
This patch implements some missing vceq* intrinsics on poly types.
The behaviour is to produce the appropriate CMEQ instruction as for the unsigned types.

Bootstrapped and tested on aarch64-none-linux-gnu.

gcc/
	PR target/71233
	* config/aarch64/arm_neon.h (vceqq_p64, vceqz_p64, vceqzq_p64): Define.

gcc/testsuite/

	PR target/71233
	* gcc.target/aarch64/simd/vceq_poly_1.c: New test.
This commit is contained in:
Kyrylo Tkachov 2020-09-22 12:00:38 +01:00
parent fa9ad35dae
commit d4703be185
2 changed files with 50 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
gcc/config/aarch64/arm_neon.h
View File

@ -12670,6 +12670,13 @@ vceqq_u64 (uint64x2_t __a, uint64x2_t __b)
return (__a == __b);
}
__extension__ extern __inline uint64x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vceqq_p64 (poly64x2_t __a, poly64x2_t __b)
{
return (__a == __b);
}
/* vceq - scalar. */
__extension__ extern __inline uint32_t
@ -12779,6 +12786,13 @@ vceqz_u64 (uint64x1_t __a)
return (__a == __AARCH64_UINT64_C (0));
}
__extension__ extern __inline uint64x1_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vceqz_p64 (poly64x1_t __a)
{
return (__a == __AARCH64_UINT64_C (0));
}
__extension__ extern __inline uint32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vceqzq_f32 (float32x4_t __a)
@ -12856,6 +12870,13 @@ vceqzq_u64 (uint64x2_t __a)
return (__a == __AARCH64_UINT64_C (0));
}
__extension__ extern __inline uint64x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vceqzq_p64 (poly64x2_t __a)
{
return (__a == __AARCH64_UINT64_C (0));
}
/* vceqz - scalar. */
__extension__ extern __inline uint32_t

29
gcc/testsuite/gcc.target/aarch64/simd/vceq_poly_1.c Normal file
View File

@ -0,0 +1,29 @@
/* { dg-do compile } */
/* { dg-options "-O" } */
#include <arm_neon.h>
uint64x2_t
foo (poly64x2_t a, poly64x2_t b)
{
return vceqq_p64 (a, b);
}
/* { dg-final { scan-assembler-times "cmeq\\tv\[0-9\]+\.2d, v\[0-9\]+\.2d, v\[0-9\]+\.2d" 1 } } */
uint64x1_t
fooz (poly64x1_t a)
{
return vceqz_p64 (a);
}
/* { dg-final { scan-assembler-times "cmeq\\td\[0-9\]+, d\[0-9\]+, #0" 1 } } */
uint64x2_t
fooqz (poly64x2_t a)
{
return vceqzq_p64 (a);
}
/* { dg-final { scan-assembler-times "cmeq\\tv\[0-9\]+\.2d, v\[0-9\]+\.2d, #0" 1 } } */