From: Srinath Parvathaneni Date: Wed, 18 Mar 2020 17:54:30 +0000 (+0000) Subject: [ARM][GCC][1/5x]: MVE store intrinsics. X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=4ff685759913e6b86d027c6007c592460e17974f;p=gcc.git [ARM][GCC][1/5x]: MVE store intrinsics. This patch supports the following MVE ACLE store intrinsics. vstrbq_scatter_offset_s8, vstrbq_scatter_offset_s32, vstrbq_scatter_offset_s16, vstrbq_scatter_offset_u8, vstrbq_scatter_offset_u32, vstrbq_scatter_offset_u16, vstrbq_s8, vstrbq_s32, vstrbq_s16, vstrbq_u8, vstrbq_u32, vstrbq_u16, vstrwq_scatter_base_s32, vstrwq_scatter_base_u32. Please refer to M-profile Vector Extension (MVE) intrinsics [1] for more details. [1] https://developer.arm.com/architectures/instruction-sets/simd-isas/helium/mve-intrinsics 2020-03-18 Andre Vieira Mihail Ionescu Srinath Parvathaneni * config/arm/arm-builtins.c (STRS_QUALIFIERS): Define builtin qualifier. (STRU_QUALIFIERS): Likewise. (STRSS_QUALIFIERS): Likewise. (STRSU_QUALIFIERS): Likewise. (STRSBS_QUALIFIERS): Likewise. (STRSBU_QUALIFIERS): Likewise. * config/arm/arm_mve.h (vstrbq_s8): Define macro. (vstrbq_u8): Likewise. (vstrbq_u16): Likewise. (vstrbq_scatter_offset_s8): Likewise. (vstrbq_scatter_offset_u8): Likewise. (vstrbq_scatter_offset_u16): Likewise. (vstrbq_s16): Likewise. (vstrbq_u32): Likewise. (vstrbq_scatter_offset_s16): Likewise. (vstrbq_scatter_offset_u32): Likewise. (vstrbq_s32): Likewise. (vstrbq_scatter_offset_s32): Likewise. (vstrwq_scatter_base_s32): Likewise. (vstrwq_scatter_base_u32): Likewise. (__arm_vstrbq_scatter_offset_s8): Define intrinsic. (__arm_vstrbq_scatter_offset_s32): Likewise. (__arm_vstrbq_scatter_offset_s16): Likewise. (__arm_vstrbq_scatter_offset_u8): Likewise. (__arm_vstrbq_scatter_offset_u32): Likewise. (__arm_vstrbq_scatter_offset_u16): Likewise. (__arm_vstrbq_s8): Likewise. (__arm_vstrbq_s32): Likewise. (__arm_vstrbq_s16): Likewise. (__arm_vstrbq_u8): Likewise. (__arm_vstrbq_u32): Likewise. (__arm_vstrbq_u16): Likewise. (__arm_vstrwq_scatter_base_s32): Likewise. (__arm_vstrwq_scatter_base_u32): Likewise. (vstrbq): Define polymorphic variant. (vstrbq_scatter_offset): Likewise. (vstrwq_scatter_base): Likewise. * config/arm/arm_mve_builtins.def (STRS_QUALIFIERS): Use builtin qualifier. (STRU_QUALIFIERS): Likewise. (STRSS_QUALIFIERS): Likewise. (STRSU_QUALIFIERS): Likewise. (STRSBS_QUALIFIERS): Likewise. (STRSBU_QUALIFIERS): Likewise. * config/arm/mve.md (MVE_B_ELEM): Define mode attribute iterator. (VSTRWSBQ): Define iterators. (VSTRBSOQ): Likewise. (VSTRBQ): Likewise. (mve_vstrbq_): Define RTL pattern. (mve_vstrbq_scatter_offset_): Likewise. (mve_vstrwq_scatter_base_v4si): Likewise. gcc/testsuite/ChangeLog: 2020-03-18 Andre Vieira Mihail Ionescu Srinath Parvathaneni * gcc.target/arm/mve/intrinsics/vstrbq_s16.c: New test. * gcc.target/arm/mve/intrinsics/vstrbq_s32.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_s8.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s16.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s32.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s8.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u16.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u32.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u8.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_u16.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_u32.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrbq_u8.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_s32.c: Likewise. * gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_u32.c: Likewise. --- diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 3b1154bca4c..be4eb7e17b8 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,59 @@ +2020-03-18 Andre Vieira + Mihail Ionescu + Srinath Parvathaneni + + * config/arm/arm-builtins.c (STRS_QUALIFIERS): Define builtin qualifier. + (STRU_QUALIFIERS): Likewise. + (STRSS_QUALIFIERS): Likewise. + (STRSU_QUALIFIERS): Likewise. + (STRSBS_QUALIFIERS): Likewise. + (STRSBU_QUALIFIERS): Likewise. + * config/arm/arm_mve.h (vstrbq_s8): Define macro. + (vstrbq_u8): Likewise. + (vstrbq_u16): Likewise. + (vstrbq_scatter_offset_s8): Likewise. + (vstrbq_scatter_offset_u8): Likewise. + (vstrbq_scatter_offset_u16): Likewise. + (vstrbq_s16): Likewise. + (vstrbq_u32): Likewise. + (vstrbq_scatter_offset_s16): Likewise. + (vstrbq_scatter_offset_u32): Likewise. + (vstrbq_s32): Likewise. + (vstrbq_scatter_offset_s32): Likewise. + (vstrwq_scatter_base_s32): Likewise. + (vstrwq_scatter_base_u32): Likewise. + (__arm_vstrbq_scatter_offset_s8): Define intrinsic. + (__arm_vstrbq_scatter_offset_s32): Likewise. + (__arm_vstrbq_scatter_offset_s16): Likewise. + (__arm_vstrbq_scatter_offset_u8): Likewise. + (__arm_vstrbq_scatter_offset_u32): Likewise. + (__arm_vstrbq_scatter_offset_u16): Likewise. + (__arm_vstrbq_s8): Likewise. + (__arm_vstrbq_s32): Likewise. + (__arm_vstrbq_s16): Likewise. + (__arm_vstrbq_u8): Likewise. + (__arm_vstrbq_u32): Likewise. + (__arm_vstrbq_u16): Likewise. + (__arm_vstrwq_scatter_base_s32): Likewise. + (__arm_vstrwq_scatter_base_u32): Likewise. + (vstrbq): Define polymorphic variant. + (vstrbq_scatter_offset): Likewise. + (vstrwq_scatter_base): Likewise. + * config/arm/arm_mve_builtins.def (STRS_QUALIFIERS): Use builtin + qualifier. + (STRU_QUALIFIERS): Likewise. + (STRSS_QUALIFIERS): Likewise. + (STRSU_QUALIFIERS): Likewise. + (STRSBS_QUALIFIERS): Likewise. + (STRSBU_QUALIFIERS): Likewise. + * config/arm/mve.md (MVE_B_ELEM): Define mode attribute iterator. + (VSTRWSBQ): Define iterators. + (VSTRBSOQ): Likewise. + (VSTRBQ): Likewise. + (mve_vstrbq_): Define RTL pattern. + (mve_vstrbq_scatter_offset_): Likewise. + (mve_vstrwq_scatter_base_v4si): Likewise. + 2020-03-18 Andre Vieira Mihail Ionescu Srinath Parvathaneni diff --git a/gcc/config/arm/arm-builtins.c b/gcc/config/arm/arm-builtins.c index 26f0379f62b..b285f074285 100644 --- a/gcc/config/arm/arm-builtins.c +++ b/gcc/config/arm/arm-builtins.c @@ -579,6 +579,39 @@ arm_quadop_unone_unone_unone_none_unone_qualifiers[SIMD_MAX_BUILTIN_ARGS] #define QUADOP_UNONE_UNONE_UNONE_NONE_UNONE_QUALIFIERS \ (arm_quadop_unone_unone_unone_none_unone_qualifiers) +static enum arm_type_qualifiers +arm_strs_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_void, qualifier_pointer, qualifier_none }; +#define STRS_QUALIFIERS (arm_strs_qualifiers) + +static enum arm_type_qualifiers +arm_stru_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_void, qualifier_pointer, qualifier_unsigned }; +#define STRU_QUALIFIERS (arm_stru_qualifiers) + +static enum arm_type_qualifiers +arm_strss_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_void, qualifier_pointer, qualifier_unsigned, + qualifier_none}; +#define STRSS_QUALIFIERS (arm_strss_qualifiers) + +static enum arm_type_qualifiers +arm_strsu_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_void, qualifier_pointer, qualifier_unsigned, + qualifier_unsigned}; +#define STRSU_QUALIFIERS (arm_strsu_qualifiers) + +static enum arm_type_qualifiers +arm_strsbs_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_void, qualifier_unsigned, qualifier_immediate, qualifier_none}; +#define STRSBS_QUALIFIERS (arm_strsbs_qualifiers) + +static enum arm_type_qualifiers +arm_strsbu_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_void, qualifier_unsigned, qualifier_immediate, + qualifier_unsigned}; +#define STRSBU_QUALIFIERS (arm_strsbu_qualifiers) + /* End of Qualifier for MVE builtins. */ /* void ([T element type] *, T, immediate). */ diff --git a/gcc/config/arm/arm_mve.h b/gcc/config/arm/arm_mve.h index 4f8135d83f9..019e907c6dc 100644 --- a/gcc/config/arm/arm_mve.h +++ b/gcc/config/arm/arm_mve.h @@ -1702,6 +1702,20 @@ typedef struct { uint8x16_t val[4]; } uint8x16x4_t; #define vsubq_m_f16(__inactive, __a, __b, __p) __arm_vsubq_m_f16(__inactive, __a, __b, __p) #define vsubq_m_n_f32(__inactive, __a, __b, __p) __arm_vsubq_m_n_f32(__inactive, __a, __b, __p) #define vsubq_m_n_f16(__inactive, __a, __b, __p) __arm_vsubq_m_n_f16(__inactive, __a, __b, __p) +#define vstrbq_s8( __addr, __value) __arm_vstrbq_s8( __addr, __value) +#define vstrbq_u8( __addr, __value) __arm_vstrbq_u8( __addr, __value) +#define vstrbq_u16( __addr, __value) __arm_vstrbq_u16( __addr, __value) +#define vstrbq_scatter_offset_s8( __base, __offset, __value) __arm_vstrbq_scatter_offset_s8( __base, __offset, __value) +#define vstrbq_scatter_offset_u8( __base, __offset, __value) __arm_vstrbq_scatter_offset_u8( __base, __offset, __value) +#define vstrbq_scatter_offset_u16( __base, __offset, __value) __arm_vstrbq_scatter_offset_u16( __base, __offset, __value) +#define vstrbq_s16( __addr, __value) __arm_vstrbq_s16( __addr, __value) +#define vstrbq_u32( __addr, __value) __arm_vstrbq_u32( __addr, __value) +#define vstrbq_scatter_offset_s16( __base, __offset, __value) __arm_vstrbq_scatter_offset_s16( __base, __offset, __value) +#define vstrbq_scatter_offset_u32( __base, __offset, __value) __arm_vstrbq_scatter_offset_u32( __base, __offset, __value) +#define vstrbq_s32( __addr, __value) __arm_vstrbq_s32( __addr, __value) +#define vstrbq_scatter_offset_s32( __base, __offset, __value) __arm_vstrbq_scatter_offset_s32( __base, __offset, __value) +#define vstrwq_scatter_base_s32(__addr, __offset, __value) __arm_vstrwq_scatter_base_s32(__addr, __offset, __value) +#define vstrwq_scatter_base_u32(__addr, __offset, __value) __arm_vstrwq_scatter_base_u32(__addr, __offset, __value) #endif __extension__ extern __inline void @@ -10995,6 +11009,103 @@ __arm_vshrntq_m_n_u16 (uint8x16_t __a, uint16x8_t __b, const int __imm, mve_pred return __builtin_mve_vshrntq_m_n_uv8hi (__a, __b, __imm, __p); } +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_scatter_offset_s8 (int8_t * __base, uint8x16_t __offset, int8x16_t __value) +{ + __builtin_mve_vstrbq_scatter_offset_sv16qi ((__builtin_neon_qi *) __base, __offset, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_scatter_offset_s32 (int8_t * __base, uint32x4_t __offset, int32x4_t __value) +{ + __builtin_mve_vstrbq_scatter_offset_sv4si ((__builtin_neon_qi *) __base, __offset, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_scatter_offset_s16 (int8_t * __base, uint16x8_t __offset, int16x8_t __value) +{ + __builtin_mve_vstrbq_scatter_offset_sv8hi ((__builtin_neon_qi *) __base, __offset, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_scatter_offset_u8 (uint8_t * __base, uint8x16_t __offset, uint8x16_t __value) +{ + __builtin_mve_vstrbq_scatter_offset_uv16qi ((__builtin_neon_qi *) __base, __offset, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_scatter_offset_u32 (uint8_t * __base, uint32x4_t __offset, uint32x4_t __value) +{ + __builtin_mve_vstrbq_scatter_offset_uv4si ((__builtin_neon_qi *) __base, __offset, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_scatter_offset_u16 (uint8_t * __base, uint16x8_t __offset, uint16x8_t __value) +{ + __builtin_mve_vstrbq_scatter_offset_uv8hi ((__builtin_neon_qi *) __base, __offset, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_s8 (int8_t * __addr, int8x16_t __value) +{ + __builtin_mve_vstrbq_sv16qi ((__builtin_neon_qi *) __addr, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_s32 (int8_t * __addr, int32x4_t __value) +{ + __builtin_mve_vstrbq_sv4si ((__builtin_neon_qi *) __addr, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_s16 (int8_t * __addr, int16x8_t __value) +{ + __builtin_mve_vstrbq_sv8hi ((__builtin_neon_qi *) __addr, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_u8 (uint8_t * __addr, uint8x16_t __value) +{ + __builtin_mve_vstrbq_uv16qi ((__builtin_neon_qi *) __addr, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_u32 (uint8_t * __addr, uint32x4_t __value) +{ + __builtin_mve_vstrbq_uv4si ((__builtin_neon_qi *) __addr, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrbq_u16 (uint8_t * __addr, uint16x8_t __value) +{ + __builtin_mve_vstrbq_uv8hi ((__builtin_neon_qi *) __addr, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrwq_scatter_base_s32 (uint32x4_t __addr, const int __offset, int32x4_t __value) +{ + __builtin_mve_vstrwq_scatter_base_sv4si (__addr, __offset, __value); +} + +__extension__ extern __inline void +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vstrwq_scatter_base_u32 (uint32x4_t __addr, const int __offset, uint32x4_t __value) +{ + __builtin_mve_vstrwq_scatter_base_uv4si (__addr, __offset, __value); +} #if (__ARM_FEATURE_MVE & 2) /* MVE Floating point. */ __extension__ extern __inline void @@ -17630,6 +17741,35 @@ extern void *__ARM_undef; int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vmulq_m_u16 (__ARM_mve_coerce(__p0, uint16x8_t), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \ int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vmulq_m_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3));}) +#define vstrbq(p0,p1) __arm_vstrbq(p0,p1) +#define __arm_vstrbq(p0,p1) ({ __typeof(p0) __p0 = (p0); \ + __typeof(p1) __p1 = (p1); \ + _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0, \ + int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int8x16_t]: __arm_vstrbq_s8 (__ARM_mve_coerce(__p0, int8_t *), __ARM_mve_coerce(__p1, int8x16_t)), \ + int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int16x8_t]: __arm_vstrbq_s16 (__ARM_mve_coerce(__p0, int8_t *), __ARM_mve_coerce(__p1, int16x8_t)), \ + int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrbq_s32 (__ARM_mve_coerce(__p0, int8_t *), __ARM_mve_coerce(__p1, int32x4_t)), \ + int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint8x16_t]: __arm_vstrbq_u8 (__ARM_mve_coerce(__p0, uint8_t *), __ARM_mve_coerce(__p1, uint8x16_t)), \ + int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vstrbq_u16 (__ARM_mve_coerce(__p0, uint8_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \ + int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrbq_u32 (__ARM_mve_coerce(__p0, uint8_t *), __ARM_mve_coerce(__p1, uint32x4_t)));}) + +#define vstrbq_scatter_offset(p0,p1,p2) __arm_vstrbq_scatter_offset(p0,p1,p2) +#define __arm_vstrbq_scatter_offset(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \ + __typeof(p1) __p1 = (p1); \ + __typeof(p2) __p2 = (p2); \ + _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \ + int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_uint8x16_t][__ARM_mve_type_int8x16_t]: __arm_vstrbq_scatter_offset_s8 (__ARM_mve_coerce(__p0, int8_t *), __ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, int8x16_t)), \ + int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_uint16x8_t][__ARM_mve_type_int16x8_t]: __arm_vstrbq_scatter_offset_s16 (__ARM_mve_coerce(__p0, int8_t *), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, int16x8_t)), \ + int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_uint32x4_t][__ARM_mve_type_int32x4_t]: __arm_vstrbq_scatter_offset_s32 (__ARM_mve_coerce(__p0, int8_t *), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, int32x4_t)), \ + int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vstrbq_scatter_offset_u8 (__ARM_mve_coerce(__p0, uint8_t *), __ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, uint8x16_t)), \ + int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vstrbq_scatter_offset_u16 (__ARM_mve_coerce(__p0, uint8_t *), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t)), \ + int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vstrbq_scatter_offset_u32 (__ARM_mve_coerce(__p0, uint8_t *), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t)));}) + +#define vstrwq_scatter_base(p0,p1,p2) __arm_vstrwq_scatter_base(p0,p1,p2) +#define __arm_vstrwq_scatter_base(p0,p1,p2) ({ __typeof(p2) __p2 = (p2); \ + _Generic( (int (*)[__ARM_mve_typeid(__p2)])0, \ + int (*)[__ARM_mve_type_int32x4_t]: __arm_vstrwq_scatter_base_s32(p0, p1, __ARM_mve_coerce(__p2, int32x4_t)), \ + int (*)[__ARM_mve_type_uint32x4_t]: __arm_vstrwq_scatter_base_u32(p0, p1, __ARM_mve_coerce(__p2, uint32x4_t)));}) + #endif /* MVE Integer. */ #define vqrdmlahq_m(p0,p1,p2,p3) __arm_vqrdmlahq_m(p0,p1,p2,p3) diff --git a/gcc/config/arm/arm_mve_builtins.def b/gcc/config/arm/arm_mve_builtins.def index b448889fc5f..2c675240984 100644 --- a/gcc/config/arm/arm_mve_builtins.def +++ b/gcc/config/arm/arm_mve_builtins.def @@ -685,3 +685,9 @@ VAR2 (QUADOP_NONE_NONE_NONE_NONE_UNONE, vandq_m_f, v8hf, v4sf) VAR2 (QUADOP_NONE_NONE_NONE_NONE_UNONE, vaddq_m_n_f, v8hf, v4sf) VAR2 (QUADOP_NONE_NONE_NONE_NONE_UNONE, vaddq_m_f, v8hf, v4sf) VAR2 (QUADOP_NONE_NONE_NONE_NONE_UNONE, vabdq_m_f, v8hf, v4sf) +VAR3 (STRS, vstrbq_s, v16qi, v8hi, v4si) +VAR3 (STRU, vstrbq_u, v16qi, v8hi, v4si) +VAR3 (STRSS, vstrbq_scatter_offset_s, v16qi, v8hi, v4si) +VAR3 (STRSU, vstrbq_scatter_offset_u, v16qi, v8hi, v4si) +VAR1 (STRSBS, vstrwq_scatter_base_s, v4si) +VAR1 (STRSBU, vstrwq_scatter_base_u, v4si) diff --git a/gcc/config/arm/mve.md b/gcc/config/arm/mve.md index bf4eb5da11c..3d0a1725d41 100644 --- a/gcc/config/arm/mve.md +++ b/gcc/config/arm/mve.md @@ -191,7 +191,8 @@ VCMLAQ_ROT90_M_F VCMULQ_M_F VCMULQ_ROT180_M_F VCMULQ_ROT270_M_F VCMULQ_ROT90_M_F VFMAQ_M_F VFMAQ_M_N_F VFMASQ_M_N_F VFMSQ_M_F VMAXNMQ_M_F - VMINNMQ_M_F VSUBQ_M_F]) + VMINNMQ_M_F VSUBQ_M_F VSTRWQSB_S VSTRWQSB_U + VSTRBQSO_S VSTRBQSO_U VSTRBQ_S VSTRBQ_U]) (define_mode_attr MVE_CNVT [(V8HI "V8HF") (V4SI "V4SF") (V8HF "V8HI") (V4SF "V4SI")]) @@ -342,7 +343,9 @@ (VQRSHRNTQ_M_N_S "s") (VQRSHRNTQ_M_N_U "u") (VQRSHRNBQ_M_N_S "s") (VQRSHRNBQ_M_N_U "u") (VMLALDAVAXQ_P_S "s") (VMLALDAVAXQ_P_U "u") - (VMLALDAVAQ_P_S "s") (VMLALDAVAQ_P_U "u")]) + (VMLALDAVAQ_P_S "s") (VMLALDAVAQ_P_U "u") + (VSTRWQSB_S "s") (VSTRWQSB_U "u") (VSTRBQSO_S "s") + (VSTRBQSO_U "u") (VSTRBQ_S "s") (VSTRBQ_U "u")]) (define_int_attr mode1 [(VCTP8Q "8") (VCTP16Q "16") (VCTP32Q "32") (VCTP64Q "64") (VCTP8Q_M "8") (VCTP16Q_M "16") @@ -359,6 +362,7 @@ (define_mode_attr MVE_constraint1 [ (V8HI "Ra") (V4SI "Rc")]) (define_mode_attr MVE_pred1 [ (V8HI "mve_imm_7") (V4SI "mve_imm_15")]) +(define_mode_attr MVE_B_ELEM [ (V16QI "V16QI") (V8HI "V8QI") (V4SI "V4QI")]) (define_int_iterator VCVTQ_TO_F [VCVTQ_TO_F_S VCVTQ_TO_F_U]) (define_int_iterator VMVNQ_N [VMVNQ_N_U VMVNQ_N_S]) @@ -562,6 +566,9 @@ (define_int_iterator VSHLLTQ_M_N [VSHLLTQ_M_N_U VSHLLTQ_M_N_S]) (define_int_iterator VSHRNBQ_M_N [VSHRNBQ_M_N_S VSHRNBQ_M_N_U]) (define_int_iterator VSHRNTQ_M_N [VSHRNTQ_M_N_S VSHRNTQ_M_N_U]) +(define_int_iterator VSTRWSBQ [VSTRWQSB_S VSTRWQSB_U]) +(define_int_iterator VSTRBSOQ [VSTRBQSO_S VSTRBQSO_U]) +(define_int_iterator VSTRBQ [VSTRBQ_S VSTRBQ_U]) (define_insn "*mve_mov" [(set (match_operand:MVE_types 0 "nonimmediate_operand" "=w,w,r,w,w,r,w,Us") @@ -7940,3 +7947,65 @@ "vpst\;vsubt.f%#\t%q0, %q2, %3" [(set_attr "type" "mve_move") (set_attr "length""8")]) + +;; +;; [vstrbq_s vstrbq_u] +;; +(define_insn "mve_vstrbq_" + [(set (match_operand: 0 "memory_operand" "=Us") + (unspec: [(match_operand:MVE_2 1 "s_register_operand" "w")] + VSTRBQ)) + ] + "TARGET_HAVE_MVE" +{ + rtx ops[2]; + int regno = REGNO (operands[1]); + ops[1] = gen_rtx_REG (TImode, regno); + ops[0] = operands[0]; + output_asm_insn("vstrb.\t%q1, %E0",ops); + return ""; +} + [(set_attr "length" "4")]) + +;; +;; [vstrbq_scatter_offset_s vstrbq_scatter_offset_u] +;; +(define_insn "mve_vstrbq_scatter_offset_" + [(set (match_operand: 0 "memory_operand" "=Us") + (unspec: + [(match_operand:MVE_2 1 "s_register_operand" "w") + (match_operand:MVE_2 2 "s_register_operand" "w")] + VSTRBSOQ)) + ] + "TARGET_HAVE_MVE" +{ + rtx ops[3]; + ops[0] = operands[0]; + ops[1] = operands[1]; + ops[2] = operands[2]; + output_asm_insn("vstrb.\t%q2, [%m0, %q1]",ops); + return ""; +} + [(set_attr "length" "4")]) + +;; +;; [vstrwq_scatter_base_s vstrwq_scatter_base_u] +;; +(define_insn "mve_vstrwq_scatter_base_v4si" + [(set (mem:BLK (scratch)) + (unspec:BLK + [(match_operand:V4SI 0 "s_register_operand" "w") + (match_operand:SI 1 "immediate_operand" "i") + (match_operand:V4SI 2 "s_register_operand" "w")] + VSTRWSBQ)) + ] + "TARGET_HAVE_MVE" +{ + rtx ops[3]; + ops[0] = operands[0]; + ops[1] = operands[1]; + ops[2] = operands[2]; + output_asm_insn("vstrw.u32\t%q2, [%q0, %1]",ops); + return ""; +} + [(set_attr "length" "4")]) diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog index 4ed6dc7fafb..9b79f8d3b9f 100644 --- a/gcc/testsuite/ChangeLog +++ b/gcc/testsuite/ChangeLog @@ -1,3 +1,22 @@ +2020-03-18 Andre Vieira + Mihail Ionescu + Srinath Parvathaneni + + * gcc.target/arm/mve/intrinsics/vstrbq_s16.c: New test. + * gcc.target/arm/mve/intrinsics/vstrbq_s32.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_s8.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s16.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s32.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s8.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u16.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u32.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u8.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_u16.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_u32.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrbq_u8.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_s32.c: Likewise. + * gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_u32.c: Likewise. + 2020-03-18 Andre Vieira Mihail Ionescu Srinath Parvathaneni diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_s16.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_s16.c new file mode 100644 index 00000000000..80fce9caebe --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_s16.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (int8_t * addr, int16x8_t value) +{ + vstrbq_s16 (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.16" } } */ + +void +foo1 (int8_t * addr, int16x8_t value) +{ + vstrbq (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.16" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_s32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_s32.c new file mode 100644 index 00000000000..ae3c89e0d4b --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_s32.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (int8_t * addr, int32x4_t value) +{ + vstrbq_s32 (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.32" } } */ + +void +foo1 (int8_t * addr, int32x4_t value) +{ + vstrbq (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_s8.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_s8.c new file mode 100644 index 00000000000..72a3673dd5f --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_s8.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (int8_t * addr, int8x16_t value) +{ + vstrbq_s8 (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.8" } } */ + +void +foo1 (int8_t * addr, int8x16_t value) +{ + vstrbq (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.8" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s16.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s16.c new file mode 100644 index 00000000000..b4ad4f7e3a5 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s16.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (int8_t * base, uint16x8_t offset, int16x8_t value) +{ + vstrbq_scatter_offset_s16 (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.16" } } */ + +void +foo1 (int8_t * base, uint16x8_t offset, int16x8_t value) +{ + vstrbq_scatter_offset (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.16" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s32.c new file mode 100644 index 00000000000..507eef81991 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s32.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (int8_t * base, uint32x4_t offset, int32x4_t value) +{ + vstrbq_scatter_offset_s32 (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.32" } } */ + +void +foo1 (int8_t * base, uint32x4_t offset, int32x4_t value) +{ + vstrbq_scatter_offset (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s8.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s8.c new file mode 100644 index 00000000000..f5467c3bc3d --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_s8.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (int8_t * base, uint8x16_t offset, int8x16_t value) +{ + vstrbq_scatter_offset_s8 (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.8" } } */ + +void +foo1 (int8_t * base, uint8x16_t offset, int8x16_t value) +{ + vstrbq_scatter_offset (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.8" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u16.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u16.c new file mode 100644 index 00000000000..cea66992543 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u16.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (uint8_t * base, uint16x8_t offset, uint16x8_t value) +{ + vstrbq_scatter_offset_u16 (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.16" } } */ + +void +foo1 (uint8_t * base, uint16x8_t offset, uint16x8_t value) +{ + vstrbq_scatter_offset (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.16" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u32.c new file mode 100644 index 00000000000..6ad7ec6f738 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u32.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (uint8_t * base, uint32x4_t offset, uint32x4_t value) +{ + vstrbq_scatter_offset_u32 (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.32" } } */ + +void +foo1 (uint8_t * base, uint32x4_t offset, uint32x4_t value) +{ + vstrbq_scatter_offset (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u8.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u8.c new file mode 100644 index 00000000000..4122622142d --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_scatter_offset_u8.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (uint8_t * base, uint8x16_t offset, uint8x16_t value) +{ + vstrbq_scatter_offset_u8 (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.8" } } */ + +void +foo1 (uint8_t * base, uint8x16_t offset, uint8x16_t value) +{ + vstrbq_scatter_offset (base, offset, value); +} + +/* { dg-final { scan-assembler "vstrb.8" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_u16.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_u16.c new file mode 100644 index 00000000000..5871fbdbfa0 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_u16.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (uint8_t * addr, uint16x8_t value) +{ + vstrbq_u16 (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.16" } } */ + +void +foo1 (uint8_t * addr, uint16x8_t value) +{ + vstrbq (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.16" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_u32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_u32.c new file mode 100644 index 00000000000..31d6d86f9fd --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_u32.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (uint8_t * addr, uint32x4_t value) +{ + vstrbq_u32 (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.32" } } */ + +void +foo1 (uint8_t * addr, uint32x4_t value) +{ + vstrbq (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_u8.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_u8.c new file mode 100644 index 00000000000..7415b98e35e --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrbq_u8.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (uint8_t * addr, uint8x16_t value) +{ + vstrbq_u8 (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.8" } } */ + +void +foo1 (uint8_t * addr, uint8x16_t value) +{ + vstrbq (addr, value); +} + +/* { dg-final { scan-assembler "vstrb.8" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_s32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_s32.c new file mode 100644 index 00000000000..0fb6af6b7c9 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_s32.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (uint32x4_t addr, int32x4_t value) +{ + vstrwq_scatter_base_s32 (addr, 8, value); +} + +/* { dg-final { scan-assembler "vstrw.u32" } } */ + +void +foo1 (uint32x4_t addr, int32x4_t value) +{ + vstrwq_scatter_base (addr, 8, value); +} + +/* { dg-final { scan-assembler "vstrw.u32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_u32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_u32.c new file mode 100644 index 00000000000..795c0b44e69 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vstrwq_scatter_base_u32.c @@ -0,0 +1,22 @@ +/* { dg-do compile } */ +/* { dg-require-effective-target arm_v8_1m_mve_ok } */ +/* { dg-add-options arm_v8_1m_mve } */ +/* { dg-additional-options "-O2" } */ + +#include "arm_mve.h" + +void +foo (uint32x4_t addr, uint32x4_t value) +{ + vstrwq_scatter_base_u32 (addr, 8, value); +} + +/* { dg-final { scan-assembler "vstrw.u32" } } */ + +void +foo1 (uint32x4_t addr, uint32x4_t value) +{ + vstrwq_scatter_base (addr, 8, value); +} + +/* { dg-final { scan-assembler "vstrw.u32" } } */