aboutsummaryrefslogtreecommitdiff
path: root/gcc
diff options
context:
space:
mode:
Diffstat (limited to 'gcc')
-rw-r--r--gcc/config/arm/arm-mve-builtins-base.cc47
-rw-r--r--gcc/config/arm/arm-mve-builtins-base.def6
-rw-r--r--gcc/config/arm/arm-mve-builtins-base.h4
-rw-r--r--gcc/config/arm/arm_mve.h576
-rw-r--r--gcc/config/arm/arm_mve_builtins.def20
-rw-r--r--gcc/config/arm/iterators.md24
-rw-r--r--gcc/config/arm/mve.md349
-rw-r--r--gcc/config/arm/unspecs.md14
8 files changed, 156 insertions, 884 deletions
diff --git a/gcc/config/arm/arm-mve-builtins-base.cc b/gcc/config/arm/arm-mve-builtins-base.cc
index 27e31d6..a4d498d 100644
--- a/gcc/config/arm/arm-mve-builtins-base.cc
+++ b/gcc/config/arm/arm-mve-builtins-base.cc
@@ -411,6 +411,49 @@ public:
}
};
+/* Builds the vldrq_gather*offset intrinsics. */
+class vldrq_gather_impl : public load_extending
+{
+public:
+ using load_extending::load_extending;
+
+ rtx expand (function_expander &e) const override
+ {
+ insn_code icode;
+ switch (e.pred)
+ {
+ case PRED_none:
+ if (e.vector_mode (0) == e.memory_vector_mode ())
+ /* Non-extending load case. */
+ icode = code_for_mve_vldrq_gather_offset (e.vector_mode (0));
+ else
+ /* Extending load case. */
+ icode = code_for_mve_vldrq_gather_offset_extend
+ (e.memory_vector_mode (),
+ e.type_suffix (0).unsigned_p
+ ? ZERO_EXTEND
+ : SIGN_EXTEND);
+ break;
+
+ case PRED_z:
+ if (e.vector_mode (0) == e.memory_vector_mode ())
+ icode = code_for_mve_vldrq_gather_offset_z (e.vector_mode (0));
+ else
+ icode = code_for_mve_vldrq_gather_offset_z_extend
+ (e.memory_vector_mode (),
+ e.type_suffix (0).unsigned_p
+ ? ZERO_EXTEND
+ : SIGN_EXTEND);
+ break;
+
+ default:
+ gcc_unreachable ();
+ }
+
+ return e.use_exact_insn (icode);
+ }
+};
+
/* Implements vctp8q, vctp16q, vctp32q and vctp64q intrinsics. */
class vctpq_impl : public function_base
{
@@ -1208,8 +1251,12 @@ FUNCTION_WITH_M_N_NO_F (vhaddq, VHADDQ)
FUNCTION_WITH_M_N_NO_F (vhsubq, VHSUBQ)
FUNCTION (vld1q, vld1_impl,)
FUNCTION (vldrbq, vldrq_impl, (TYPE_SUFFIX_s8, TYPE_SUFFIX_u8))
+FUNCTION (vldrbq_gather, vldrq_gather_impl, (TYPE_SUFFIX_s8, TYPE_SUFFIX_u8))
+FUNCTION (vldrdq_gather, vldrq_gather_impl, (TYPE_SUFFIX_s64, TYPE_SUFFIX_u64, NUM_TYPE_SUFFIXES))
FUNCTION (vldrhq, vldrq_impl, (TYPE_SUFFIX_s16, TYPE_SUFFIX_u16, TYPE_SUFFIX_f16))
+FUNCTION (vldrhq_gather, vldrq_gather_impl, (TYPE_SUFFIX_s16, TYPE_SUFFIX_u16, TYPE_SUFFIX_f16))
FUNCTION (vldrwq, vldrq_impl, (TYPE_SUFFIX_s32, TYPE_SUFFIX_u32, TYPE_SUFFIX_f32))
+FUNCTION (vldrwq_gather, vldrq_gather_impl, (TYPE_SUFFIX_s32, TYPE_SUFFIX_u32, TYPE_SUFFIX_f32))
FUNCTION_PRED_P_S (vmaxavq, VMAXAVQ)
FUNCTION_WITHOUT_N_NO_U_F (vmaxaq, VMAXAQ)
FUNCTION_ONLY_F (vmaxnmaq, VMAXNMAQ)
diff --git a/gcc/config/arm/arm-mve-builtins-base.def b/gcc/config/arm/arm-mve-builtins-base.def
index a56fae7..5e30a27 100644
--- a/gcc/config/arm/arm-mve-builtins-base.def
+++ b/gcc/config/arm/arm-mve-builtins-base.def
@@ -60,8 +60,12 @@ DEF_MVE_FUNCTION (vidupq, viddup, all_unsigned, mx_or_none)
DEF_MVE_FUNCTION (viwdupq, vidwdup, all_unsigned, mx_or_none)
DEF_MVE_FUNCTION (vld1q, load, all_integer, z_or_none)
DEF_MVE_FUNCTION (vldrbq, load_ext, all_integer, z_or_none)
+DEF_MVE_FUNCTION (vldrbq_gather, load_ext_gather_offset, all_integer, z_or_none)
+DEF_MVE_FUNCTION (vldrdq_gather, load_ext_gather_offset, integer_64, z_or_none)
DEF_MVE_FUNCTION (vldrhq, load_ext, integer_16_32, z_or_none)
+DEF_MVE_FUNCTION (vldrhq_gather, load_ext_gather_offset, integer_16_32, z_or_none)
DEF_MVE_FUNCTION (vldrwq, load_ext, integer_32, z_or_none)
+DEF_MVE_FUNCTION (vldrwq_gather, load_ext_gather_offset, integer_32, z_or_none)
DEF_MVE_FUNCTION (vmaxaq, binary_maxamina, all_signed, m_or_none)
DEF_MVE_FUNCTION (vmaxavq, binary_maxavminav, all_signed, p_or_none)
DEF_MVE_FUNCTION (vmaxq, binary, all_integer, mx_or_none)
@@ -226,7 +230,9 @@ DEF_MVE_FUNCTION (vfmasq, ternary_n, all_float, m_or_none)
DEF_MVE_FUNCTION (vfmsq, ternary, all_float, m_or_none)
DEF_MVE_FUNCTION (vld1q, load, all_float, z_or_none)
DEF_MVE_FUNCTION (vldrhq, load_ext, float_16, z_or_none)
+DEF_MVE_FUNCTION (vldrhq_gather, load_ext_gather_offset, float_16, z_or_none)
DEF_MVE_FUNCTION (vldrwq, load_ext, float_32, z_or_none)
+DEF_MVE_FUNCTION (vldrwq_gather, load_ext_gather_offset, float_32, z_or_none)
DEF_MVE_FUNCTION (vmaxnmaq, binary, all_float, m_or_none)
DEF_MVE_FUNCTION (vmaxnmavq, binary_maxvminv, all_float, p_or_none)
DEF_MVE_FUNCTION (vmaxnmq, binary, all_float, mx_or_none)
diff --git a/gcc/config/arm/arm-mve-builtins-base.h b/gcc/config/arm/arm-mve-builtins-base.h
index 2612480..88fcff3 100644
--- a/gcc/config/arm/arm-mve-builtins-base.h
+++ b/gcc/config/arm/arm-mve-builtins-base.h
@@ -83,8 +83,12 @@ extern const function_base *const vidupq;
extern const function_base *const viwdupq;
extern const function_base *const vld1q;
extern const function_base *const vldrbq;
+extern const function_base *const vldrbq_gather;
+extern const function_base *const vldrdq_gather;
extern const function_base *const vldrhq;
+extern const function_base *const vldrhq_gather;
extern const function_base *const vldrwq;
+extern const function_base *const vldrwq_gather;
extern const function_base *const vmaxaq;
extern const function_base *const vmaxavq;
extern const function_base *const vmaxnmaq;
diff --git a/gcc/config/arm/arm_mve.h b/gcc/config/arm/arm_mve.h
index 26846d1..b9df10d 100644
--- a/gcc/config/arm/arm_mve.h
+++ b/gcc/config/arm/arm_mve.h
@@ -46,18 +46,10 @@
#ifndef __ARM_MVE_PRESERVE_USER_NAMESPACE
#define vst4q(__addr, __value) __arm_vst4q(__addr, __value)
-#define vldrbq_gather_offset(__base, __offset) __arm_vldrbq_gather_offset(__base, __offset)
-#define vldrbq_gather_offset_z(__base, __offset, __p) __arm_vldrbq_gather_offset_z(__base, __offset, __p)
-#define vldrhq_gather_offset(__base, __offset) __arm_vldrhq_gather_offset(__base, __offset)
-#define vldrhq_gather_offset_z(__base, __offset, __p) __arm_vldrhq_gather_offset_z(__base, __offset, __p)
#define vldrhq_gather_shifted_offset(__base, __offset) __arm_vldrhq_gather_shifted_offset(__base, __offset)
#define vldrhq_gather_shifted_offset_z(__base, __offset, __p) __arm_vldrhq_gather_shifted_offset_z(__base, __offset, __p)
-#define vldrdq_gather_offset(__base, __offset) __arm_vldrdq_gather_offset(__base, __offset)
-#define vldrdq_gather_offset_z(__base, __offset, __p) __arm_vldrdq_gather_offset_z(__base, __offset, __p)
#define vldrdq_gather_shifted_offset(__base, __offset) __arm_vldrdq_gather_shifted_offset(__base, __offset)
#define vldrdq_gather_shifted_offset_z(__base, __offset, __p) __arm_vldrdq_gather_shifted_offset_z(__base, __offset, __p)
-#define vldrwq_gather_offset(__base, __offset) __arm_vldrwq_gather_offset(__base, __offset)
-#define vldrwq_gather_offset_z(__base, __offset, __p) __arm_vldrwq_gather_offset_z(__base, __offset, __p)
#define vldrwq_gather_shifted_offset(__base, __offset) __arm_vldrwq_gather_shifted_offset(__base, __offset)
#define vldrwq_gather_shifted_offset_z(__base, __offset, __p) __arm_vldrwq_gather_shifted_offset_z(__base, __offset, __p)
#define vuninitializedq(__v) __arm_vuninitializedq(__v)
@@ -77,30 +69,10 @@
#define vst4q_f16( __addr, __value) __arm_vst4q_f16( __addr, __value)
#define vst4q_f32( __addr, __value) __arm_vst4q_f32( __addr, __value)
#define vpnot(__a) __arm_vpnot(__a)
-#define vldrbq_gather_offset_u8(__base, __offset) __arm_vldrbq_gather_offset_u8(__base, __offset)
-#define vldrbq_gather_offset_s8(__base, __offset) __arm_vldrbq_gather_offset_s8(__base, __offset)
-#define vldrbq_gather_offset_u16(__base, __offset) __arm_vldrbq_gather_offset_u16(__base, __offset)
-#define vldrbq_gather_offset_s16(__base, __offset) __arm_vldrbq_gather_offset_s16(__base, __offset)
-#define vldrbq_gather_offset_u32(__base, __offset) __arm_vldrbq_gather_offset_u32(__base, __offset)
-#define vldrbq_gather_offset_s32(__base, __offset) __arm_vldrbq_gather_offset_s32(__base, __offset)
#define vldrwq_gather_base_s32(__addr, __offset) __arm_vldrwq_gather_base_s32(__addr, __offset)
#define vldrwq_gather_base_u32(__addr, __offset) __arm_vldrwq_gather_base_u32(__addr, __offset)
-#define vldrbq_gather_offset_z_s16(__base, __offset, __p) __arm_vldrbq_gather_offset_z_s16(__base, __offset, __p)
-#define vldrbq_gather_offset_z_u8(__base, __offset, __p) __arm_vldrbq_gather_offset_z_u8(__base, __offset, __p)
-#define vldrbq_gather_offset_z_s32(__base, __offset, __p) __arm_vldrbq_gather_offset_z_s32(__base, __offset, __p)
-#define vldrbq_gather_offset_z_u16(__base, __offset, __p) __arm_vldrbq_gather_offset_z_u16(__base, __offset, __p)
-#define vldrbq_gather_offset_z_u32(__base, __offset, __p) __arm_vldrbq_gather_offset_z_u32(__base, __offset, __p)
-#define vldrbq_gather_offset_z_s8(__base, __offset, __p) __arm_vldrbq_gather_offset_z_s8(__base, __offset, __p)
#define vldrwq_gather_base_z_u32(__addr, __offset, __p) __arm_vldrwq_gather_base_z_u32(__addr, __offset, __p)
#define vldrwq_gather_base_z_s32(__addr, __offset, __p) __arm_vldrwq_gather_base_z_s32(__addr, __offset, __p)
-#define vldrhq_gather_offset_s32(__base, __offset) __arm_vldrhq_gather_offset_s32(__base, __offset)
-#define vldrhq_gather_offset_s16(__base, __offset) __arm_vldrhq_gather_offset_s16(__base, __offset)
-#define vldrhq_gather_offset_u32(__base, __offset) __arm_vldrhq_gather_offset_u32(__base, __offset)
-#define vldrhq_gather_offset_u16(__base, __offset) __arm_vldrhq_gather_offset_u16(__base, __offset)
-#define vldrhq_gather_offset_z_s32(__base, __offset, __p) __arm_vldrhq_gather_offset_z_s32(__base, __offset, __p)
-#define vldrhq_gather_offset_z_s16(__base, __offset, __p) __arm_vldrhq_gather_offset_z_s16(__base, __offset, __p)
-#define vldrhq_gather_offset_z_u32(__base, __offset, __p) __arm_vldrhq_gather_offset_z_u32(__base, __offset, __p)
-#define vldrhq_gather_offset_z_u16(__base, __offset, __p) __arm_vldrhq_gather_offset_z_u16(__base, __offset, __p)
#define vldrhq_gather_shifted_offset_s32(__base, __offset) __arm_vldrhq_gather_shifted_offset_s32(__base, __offset)
#define vldrhq_gather_shifted_offset_s16(__base, __offset) __arm_vldrhq_gather_shifted_offset_s16(__base, __offset)
#define vldrhq_gather_shifted_offset_u32(__base, __offset) __arm_vldrhq_gather_shifted_offset_u32(__base, __offset)
@@ -113,26 +85,14 @@
#define vldrdq_gather_base_u64(__addr, __offset) __arm_vldrdq_gather_base_u64(__addr, __offset)
#define vldrdq_gather_base_z_s64(__addr, __offset, __p) __arm_vldrdq_gather_base_z_s64(__addr, __offset, __p)
#define vldrdq_gather_base_z_u64(__addr, __offset, __p) __arm_vldrdq_gather_base_z_u64(__addr, __offset, __p)
-#define vldrdq_gather_offset_s64(__base, __offset) __arm_vldrdq_gather_offset_s64(__base, __offset)
-#define vldrdq_gather_offset_u64(__base, __offset) __arm_vldrdq_gather_offset_u64(__base, __offset)
-#define vldrdq_gather_offset_z_s64(__base, __offset, __p) __arm_vldrdq_gather_offset_z_s64(__base, __offset, __p)
-#define vldrdq_gather_offset_z_u64(__base, __offset, __p) __arm_vldrdq_gather_offset_z_u64(__base, __offset, __p)
#define vldrdq_gather_shifted_offset_s64(__base, __offset) __arm_vldrdq_gather_shifted_offset_s64(__base, __offset)
#define vldrdq_gather_shifted_offset_u64(__base, __offset) __arm_vldrdq_gather_shifted_offset_u64(__base, __offset)
#define vldrdq_gather_shifted_offset_z_s64(__base, __offset, __p) __arm_vldrdq_gather_shifted_offset_z_s64(__base, __offset, __p)
#define vldrdq_gather_shifted_offset_z_u64(__base, __offset, __p) __arm_vldrdq_gather_shifted_offset_z_u64(__base, __offset, __p)
-#define vldrhq_gather_offset_f16(__base, __offset) __arm_vldrhq_gather_offset_f16(__base, __offset)
-#define vldrhq_gather_offset_z_f16(__base, __offset, __p) __arm_vldrhq_gather_offset_z_f16(__base, __offset, __p)
#define vldrhq_gather_shifted_offset_f16(__base, __offset) __arm_vldrhq_gather_shifted_offset_f16(__base, __offset)
#define vldrhq_gather_shifted_offset_z_f16(__base, __offset, __p) __arm_vldrhq_gather_shifted_offset_z_f16(__base, __offset, __p)
#define vldrwq_gather_base_f32(__addr, __offset) __arm_vldrwq_gather_base_f32(__addr, __offset)
#define vldrwq_gather_base_z_f32(__addr, __offset, __p) __arm_vldrwq_gather_base_z_f32(__addr, __offset, __p)
-#define vldrwq_gather_offset_f32(__base, __offset) __arm_vldrwq_gather_offset_f32(__base, __offset)
-#define vldrwq_gather_offset_s32(__base, __offset) __arm_vldrwq_gather_offset_s32(__base, __offset)
-#define vldrwq_gather_offset_u32(__base, __offset) __arm_vldrwq_gather_offset_u32(__base, __offset)
-#define vldrwq_gather_offset_z_f32(__base, __offset, __p) __arm_vldrwq_gather_offset_z_f32(__base, __offset, __p)
-#define vldrwq_gather_offset_z_s32(__base, __offset, __p) __arm_vldrwq_gather_offset_z_s32(__base, __offset, __p)
-#define vldrwq_gather_offset_z_u32(__base, __offset, __p) __arm_vldrwq_gather_offset_z_u32(__base, __offset, __p)
#define vldrwq_gather_shifted_offset_f32(__base, __offset) __arm_vldrwq_gather_shifted_offset_f32(__base, __offset)
#define vldrwq_gather_shifted_offset_s32(__base, __offset) __arm_vldrwq_gather_shifted_offset_s32(__base, __offset)
#define vldrwq_gather_shifted_offset_u32(__base, __offset) __arm_vldrwq_gather_shifted_offset_u32(__base, __offset)
@@ -294,48 +254,6 @@ __arm_vpnot (mve_pred16_t __a)
return __builtin_mve_vpnotv16bi (__a);
}
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_u8 (uint8_t const * __base, uint8x16_t __offset)
-{
- return __builtin_mve_vldrbq_gather_offset_uv16qi ((__builtin_neon_qi *) __base, __offset);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_s8 (int8_t const * __base, uint8x16_t __offset)
-{
- return __builtin_mve_vldrbq_gather_offset_sv16qi ((__builtin_neon_qi *) __base, __offset);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_u16 (uint8_t const * __base, uint16x8_t __offset)
-{
- return __builtin_mve_vldrbq_gather_offset_uv8hi ((__builtin_neon_qi *) __base, __offset);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_s16 (int8_t const * __base, uint16x8_t __offset)
-{
- return __builtin_mve_vldrbq_gather_offset_sv8hi ((__builtin_neon_qi *) __base, __offset);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_u32 (uint8_t const * __base, uint32x4_t __offset)
-{
- return __builtin_mve_vldrbq_gather_offset_uv4si ((__builtin_neon_qi *) __base, __offset);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_s32 (int8_t const * __base, uint32x4_t __offset)
-{
- return __builtin_mve_vldrbq_gather_offset_sv4si ((__builtin_neon_qi *) __base, __offset);
-}
-
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrwq_gather_base_s32 (uint32x4_t __addr, const int __offset)
@@ -350,48 +268,6 @@ __arm_vldrwq_gather_base_u32 (uint32x4_t __addr, const int __offset)
return __builtin_mve_vldrwq_gather_base_uv4si (__addr, __offset);
}
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z_s8 (int8_t const * __base, uint8x16_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_gather_offset_z_sv16qi ((__builtin_neon_qi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z_s32 (int8_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_gather_offset_z_sv4si ((__builtin_neon_qi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z_s16 (int8_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_gather_offset_z_sv8hi ((__builtin_neon_qi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z_u8 (uint8_t const * __base, uint8x16_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_gather_offset_z_uv16qi ((__builtin_neon_qi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z_u32 (uint8_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_gather_offset_z_uv4si ((__builtin_neon_qi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z_u16 (uint8_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_gather_offset_z_uv8hi ((__builtin_neon_qi *) __base, __offset, __p);
-}
-
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrwq_gather_base_z_s32 (uint32x4_t __addr, const int __offset, mve_pred16_t __p)
@@ -408,62 +284,6 @@ __arm_vldrwq_gather_base_z_u32 (uint32x4_t __addr, const int __offset, mve_pred1
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_s32 (int16_t const * __base, uint32x4_t __offset)
-{
- return __builtin_mve_vldrhq_gather_offset_sv4si ((__builtin_neon_hi *) __base, __offset);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_s16 (int16_t const * __base, uint16x8_t __offset)
-{
- return __builtin_mve_vldrhq_gather_offset_sv8hi ((__builtin_neon_hi *) __base, __offset);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_u32 (uint16_t const * __base, uint32x4_t __offset)
-{
- return __builtin_mve_vldrhq_gather_offset_uv4si ((__builtin_neon_hi *) __base, __offset);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_u16 (uint16_t const * __base, uint16x8_t __offset)
-{
- return __builtin_mve_vldrhq_gather_offset_uv8hi ((__builtin_neon_hi *) __base, __offset);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z_s32 (int16_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_gather_offset_z_sv4si ((__builtin_neon_hi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z_s16 (int16_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_gather_offset_z_sv8hi ((__builtin_neon_hi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z_u32 (uint16_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_gather_offset_z_uv4si ((__builtin_neon_hi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z_u16 (uint16_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_gather_offset_z_uv8hi ((__builtin_neon_hi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrhq_gather_shifted_offset_s32 (int16_t const * __base, uint32x4_t __offset)
{
return __builtin_mve_vldrhq_gather_shifted_offset_sv4si ((__builtin_neon_hi *) __base, __offset);
@@ -548,35 +368,6 @@ __arm_vldrdq_gather_base_z_u64 (uint64x2_t __addr, const int __offset, mve_pred1
__extension__ extern __inline int64x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrdq_gather_offset_s64 (int64_t const * __base, uint64x2_t __offset)
-{
- return __builtin_mve_vldrdq_gather_offset_sv2di ((__builtin_neon_di *) __base, __offset);
-}
-
-__extension__ extern __inline uint64x2_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrdq_gather_offset_u64 (uint64_t const * __base, uint64x2_t __offset)
-{
- return __builtin_mve_vldrdq_gather_offset_uv2di ((__builtin_neon_di *) __base, __offset);
-}
-
-__extension__ extern __inline int64x2_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrdq_gather_offset_z_s64 (int64_t const * __base, uint64x2_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrdq_gather_offset_z_sv2di ((__builtin_neon_di *) __base, __offset, __p);
-}
-
-
-__extension__ extern __inline uint64x2_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrdq_gather_offset_z_u64 (uint64_t const * __base, uint64x2_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrdq_gather_offset_z_uv2di ((__builtin_neon_di *) __base, __offset, __p);
-}
-
-__extension__ extern __inline int64x2_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrdq_gather_shifted_offset_s64 (int64_t const * __base, uint64x2_t __offset)
{
return __builtin_mve_vldrdq_gather_shifted_offset_sv2di ((__builtin_neon_di *) __base, __offset);
@@ -605,34 +396,6 @@ __arm_vldrdq_gather_shifted_offset_z_u64 (uint64_t const * __base, uint64x2_t __
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset_s32 (int32_t const * __base, uint32x4_t __offset)
-{
- return __builtin_mve_vldrwq_gather_offset_sv4si ((__builtin_neon_si *) __base, __offset);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset_u32 (uint32_t const * __base, uint32x4_t __offset)
-{
- return __builtin_mve_vldrwq_gather_offset_uv4si ((__builtin_neon_si *) __base, __offset);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset_z_s32 (int32_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrwq_gather_offset_z_sv4si ((__builtin_neon_si *) __base, __offset, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset_z_u32 (uint32_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrwq_gather_offset_z_uv4si ((__builtin_neon_si *) __base, __offset, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrwq_gather_shifted_offset_s32 (int32_t const * __base, uint32x4_t __offset)
{
return __builtin_mve_vldrwq_gather_shifted_offset_sv4si ((__builtin_neon_si *) __base, __offset);
@@ -1171,20 +934,6 @@ __arm_vst4q_f32 (float32_t * __addr, float32x4x4_t __value)
__extension__ extern __inline float16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_f16 (float16_t const * __base, uint16x8_t __offset)
-{
- return __builtin_mve_vldrhq_gather_offset_fv8hf((__builtin_neon_hi *) __base, __offset);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z_f16 (float16_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_gather_offset_z_fv8hf((__builtin_neon_hi *) __base, __offset, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrhq_gather_shifted_offset_f16 (float16_t const * __base, uint16x8_t __offset)
{
return __builtin_mve_vldrhq_gather_shifted_offset_fv8hf ((__builtin_neon_hi *) __base, __offset);
@@ -1213,20 +962,6 @@ __arm_vldrwq_gather_base_z_f32 (uint32x4_t __addr, const int __offset, mve_pred1
__extension__ extern __inline float32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset_f32 (float32_t const * __base, uint32x4_t __offset)
-{
- return __builtin_mve_vldrwq_gather_offset_fv4sf((__builtin_neon_si *) __base, __offset);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset_z_f32 (float32_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __builtin_mve_vldrwq_gather_offset_z_fv4sf((__builtin_neon_si *) __base, __offset, __p);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrwq_gather_shifted_offset_f32 (float32_t const * __base, uint32x4_t __offset)
{
return __builtin_mve_vldrwq_gather_shifted_offset_fv4sf ((__builtin_neon_si *) __base, __offset);
@@ -1391,146 +1126,6 @@ __arm_vst4q (uint32_t * __addr, uint32x4x4_t __value)
__arm_vst4q_u32 (__addr, __value);
}
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset (uint8_t const * __base, uint8x16_t __offset)
-{
- return __arm_vldrbq_gather_offset_u8 (__base, __offset);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset (int8_t const * __base, uint8x16_t __offset)
-{
- return __arm_vldrbq_gather_offset_s8 (__base, __offset);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset (uint8_t const * __base, uint16x8_t __offset)
-{
- return __arm_vldrbq_gather_offset_u16 (__base, __offset);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset (int8_t const * __base, uint16x8_t __offset)
-{
- return __arm_vldrbq_gather_offset_s16 (__base, __offset);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset (uint8_t const * __base, uint32x4_t __offset)
-{
- return __arm_vldrbq_gather_offset_u32 (__base, __offset);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset (int8_t const * __base, uint32x4_t __offset)
-{
- return __arm_vldrbq_gather_offset_s32 (__base, __offset);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z (int8_t const * __base, uint8x16_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrbq_gather_offset_z_s8 (__base, __offset, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z (int8_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrbq_gather_offset_z_s32 (__base, __offset, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z (int8_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrbq_gather_offset_z_s16 (__base, __offset, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z (uint8_t const * __base, uint8x16_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrbq_gather_offset_z_u8 (__base, __offset, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z (uint8_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrbq_gather_offset_z_u32 (__base, __offset, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_gather_offset_z (uint8_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrbq_gather_offset_z_u16 (__base, __offset, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset (int16_t const * __base, uint32x4_t __offset)
-{
- return __arm_vldrhq_gather_offset_s32 (__base, __offset);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset (int16_t const * __base, uint16x8_t __offset)
-{
- return __arm_vldrhq_gather_offset_s16 (__base, __offset);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset (uint16_t const * __base, uint32x4_t __offset)
-{
- return __arm_vldrhq_gather_offset_u32 (__base, __offset);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset (uint16_t const * __base, uint16x8_t __offset)
-{
- return __arm_vldrhq_gather_offset_u16 (__base, __offset);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z (int16_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrhq_gather_offset_z_s32 (__base, __offset, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z (int16_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrhq_gather_offset_z_s16 (__base, __offset, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z (uint16_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrhq_gather_offset_z_u32 (__base, __offset, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z (uint16_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrhq_gather_offset_z_u16 (__base, __offset, __p);
-}
-
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrhq_gather_shifted_offset (int16_t const * __base, uint32x4_t __offset)
@@ -1589,34 +1184,6 @@ __arm_vldrhq_gather_shifted_offset_z (uint16_t const * __base, uint16x8_t __offs
__extension__ extern __inline int64x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrdq_gather_offset (int64_t const * __base, uint64x2_t __offset)
-{
- return __arm_vldrdq_gather_offset_s64 (__base, __offset);
-}
-
-__extension__ extern __inline uint64x2_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrdq_gather_offset (uint64_t const * __base, uint64x2_t __offset)
-{
- return __arm_vldrdq_gather_offset_u64 (__base, __offset);
-}
-
-__extension__ extern __inline int64x2_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrdq_gather_offset_z (int64_t const * __base, uint64x2_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrdq_gather_offset_z_s64 (__base, __offset, __p);
-}
-
-__extension__ extern __inline uint64x2_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrdq_gather_offset_z (uint64_t const * __base, uint64x2_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrdq_gather_offset_z_u64 (__base, __offset, __p);
-}
-
-__extension__ extern __inline int64x2_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrdq_gather_shifted_offset (int64_t const * __base, uint64x2_t __offset)
{
return __arm_vldrdq_gather_shifted_offset_s64 (__base, __offset);
@@ -1645,34 +1212,6 @@ __arm_vldrdq_gather_shifted_offset_z (uint64_t const * __base, uint64x2_t __offs
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset (int32_t const * __base, uint32x4_t __offset)
-{
- return __arm_vldrwq_gather_offset_s32 (__base, __offset);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset (uint32_t const * __base, uint32x4_t __offset)
-{
- return __arm_vldrwq_gather_offset_u32 (__base, __offset);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset_z (int32_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrwq_gather_offset_z_s32 (__base, __offset, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset_z (uint32_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrwq_gather_offset_z_u32 (__base, __offset, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrwq_gather_shifted_offset (int32_t const * __base, uint32x4_t __offset)
{
return __arm_vldrwq_gather_shifted_offset_s32 (__base, __offset);
@@ -1955,20 +1494,6 @@ __arm_vst4q (float32_t * __addr, float32x4x4_t __value)
__extension__ extern __inline float16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset (float16_t const * __base, uint16x8_t __offset)
-{
- return __arm_vldrhq_gather_offset_f16 (__base, __offset);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_gather_offset_z (float16_t const * __base, uint16x8_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrhq_gather_offset_z_f16 (__base, __offset, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrhq_gather_shifted_offset (float16_t const * __base, uint16x8_t __offset)
{
return __arm_vldrhq_gather_shifted_offset_f16 (__base, __offset);
@@ -1983,20 +1508,6 @@ __arm_vldrhq_gather_shifted_offset_z (float16_t const * __base, uint16x8_t __off
__extension__ extern __inline float32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset (float32_t const * __base, uint32x4_t __offset)
-{
- return __arm_vldrwq_gather_offset_f32 (__base, __offset);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_gather_offset_z (float32_t const * __base, uint32x4_t __offset, mve_pred16_t __p)
-{
- return __arm_vldrwq_gather_offset_z_f32 (__base, __offset, __p);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrwq_gather_shifted_offset (float32_t const * __base, uint32x4_t __offset)
{
return __arm_vldrwq_gather_shifted_offset_f32 (__base, __offset);
@@ -2424,22 +1935,6 @@ extern void *__ARM_undef;
int (*)[__ARM_mve_type_float16_t_ptr]: __arm_vld4q_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *)), \
int (*)[__ARM_mve_type_float32_t_ptr]: __arm_vld4q_f32 (__ARM_mve_coerce_f32_ptr(p0, float32_t *))))
-#define __arm_vldrhq_gather_offset(p0,p1) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_offset_s32 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint32x4_t)), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_offset_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t)), \
- int (*)[__ARM_mve_type_float16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *), __ARM_mve_coerce(__p1, uint16x8_t)));})
-
-#define __arm_vldrhq_gather_offset_z(p0,p1,p2) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_z_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_offset_z_s32 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_z_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_offset_z_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2), \
- int (*)[__ARM_mve_type_float16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_z_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2));})
-
#define __arm_vldrhq_gather_shifted_offset(p0,p1) ({ __typeof(p1) __p1 = (p1); \
_Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_shifted_offset_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
@@ -2456,18 +1951,6 @@ extern void *__ARM_undef;
int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_shifted_offset_z_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2), \
int (*)[__ARM_mve_type_float16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_shifted_offset_z_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2));})
-#define __arm_vldrwq_gather_offset(p0,p1) ( \
- _Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
- int (*)[__ARM_mve_type_int32_t_ptr]: __arm_vldrwq_gather_offset_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), p1), \
- int (*)[__ARM_mve_type_uint32_t_ptr]: __arm_vldrwq_gather_offset_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), p1), \
- int (*)[__ARM_mve_type_float32_t_ptr]: __arm_vldrwq_gather_offset_f32 (__ARM_mve_coerce_f32_ptr(p0, float32_t *), p1)))
-
-#define __arm_vldrwq_gather_offset_z(p0,p1,p2) ( \
- _Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
- int (*)[__ARM_mve_type_int32_t_ptr]: __arm_vldrwq_gather_offset_z_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), p1, p2), \
- int (*)[__ARM_mve_type_uint32_t_ptr]: __arm_vldrwq_gather_offset_z_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), p1, p2), \
- int (*)[__ARM_mve_type_float32_t_ptr]: __arm_vldrwq_gather_offset_z_f32 (__ARM_mve_coerce_f32_ptr(p0, float32_t *), p1, p2)))
-
#define __arm_vldrwq_gather_shifted_offset(p0,p1) ( \
_Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
int (*)[__ARM_mve_type_int32_t_ptr]: __arm_vldrwq_gather_shifted_offset_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), p1), \
@@ -2542,29 +2025,6 @@ extern void *__ARM_undef;
int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8x4_t]: __arm_vst4q_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8x4_t)), \
int (*)[__ARM_mve_type_uint32_t_ptr][__ARM_mve_type_uint32x4x4_t]: __arm_vst4q_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), __ARM_mve_coerce(__p1, uint32x4x4_t)));})
-#define __arm_vldrbq_gather_offset(p0,p1) ({ __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_uint8x16_t]: __arm_vldrbq_gather_offset_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, uint8x16_t)), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_s16 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_s32 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, uint32x4_t)), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint8x16_t]: __arm_vldrbq_gather_offset_u8 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint8x16_t)), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_u16 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_u32 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint32x4_t)));})
-
-#define __arm_vldrhq_gather_offset(p0,p1) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_offset_s32 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint32x4_t)), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_offset_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t)));})
-
-#define __arm_vldrhq_gather_offset_z(p0,p1,p2) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_z_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_offset_z_s32 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_offset_z_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_offset_z_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2));})
-
#define __arm_vldrhq_gather_shifted_offset(p0,p1) ({ __typeof(p1) __p1 = (p1); \
_Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_shifted_offset_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
@@ -2579,16 +2039,6 @@ extern void *__ARM_undef;
int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrhq_gather_shifted_offset_z_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrhq_gather_shifted_offset_z_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2));})
-#define __arm_vldrwq_gather_offset(p0,p1) ({ __typeof(p0) __p0 = (p0); \
- _Generic( (int (*)[__ARM_mve_typeid(__p0)])0, \
- int (*)[__ARM_mve_type_int32_t_ptr]: __arm_vldrwq_gather_offset_s32 (__ARM_mve_coerce_s32_ptr(__p0, int32_t *), p1), \
- int (*)[__ARM_mve_type_uint32_t_ptr]: __arm_vldrwq_gather_offset_u32 (__ARM_mve_coerce_u32_ptr(__p0, uint32_t *), p1));})
-
-#define __arm_vldrwq_gather_offset_z(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \
- _Generic( (int (*)[__ARM_mve_typeid(__p0)])0, \
- int (*)[__ARM_mve_type_int32_t_ptr]: __arm_vldrwq_gather_offset_z_s32 (__ARM_mve_coerce_s32_ptr(__p0, int32_t *), p1, p2), \
- int (*)[__ARM_mve_type_uint32_t_ptr]: __arm_vldrwq_gather_offset_z_u32 (__ARM_mve_coerce_u32_ptr(__p0, uint32_t *), p1, p2));})
-
#define __arm_vldrwq_gather_shifted_offset(p0,p1) ({ __typeof(p0) __p0 = (p0); \
_Generic( (int (*)[__ARM_mve_typeid(__p0)])0, \
int (*)[__ARM_mve_type_int32_t_ptr]: __arm_vldrwq_gather_shifted_offset_s32 (__ARM_mve_coerce_s32_ptr(__p0, int32_t *), p1), \
@@ -2662,14 +2112,6 @@ extern void *__ARM_undef;
#endif /* MVE Integer. */
-#define __arm_vldrdq_gather_offset(p0,p1) ( _Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
- int (*)[__ARM_mve_type_int64_t_ptr]: __arm_vldrdq_gather_offset_s64 (__ARM_mve_coerce_s64_ptr(p0, int64_t *), p1), \
- int (*)[__ARM_mve_type_uint64_t_ptr]: __arm_vldrdq_gather_offset_u64 (__ARM_mve_coerce_u64_ptr(p0, uint64_t *), p1)))
-
-#define __arm_vldrdq_gather_offset_z(p0,p1,p2) ( _Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
- int (*)[__ARM_mve_type_int64_t_ptr]: __arm_vldrdq_gather_offset_z_s64 (__ARM_mve_coerce_s64_ptr(p0, int64_t *), p1, p2), \
- int (*)[__ARM_mve_type_uint64_t_ptr]: __arm_vldrdq_gather_offset_z_u64 (__ARM_mve_coerce_u64_ptr(p0, uint64_t *), p1, p2)))
-
#define __arm_vldrdq_gather_shifted_offset(p0,p1) ( _Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
int (*)[__ARM_mve_type_int64_t_ptr]: __arm_vldrdq_gather_shifted_offset_s64 (__ARM_mve_coerce_s64_ptr(p0, int64_t *), p1), \
int (*)[__ARM_mve_type_uint64_t_ptr]: __arm_vldrdq_gather_shifted_offset_u64 (__ARM_mve_coerce_u64_ptr(p0, uint64_t *), p1)))
@@ -2678,24 +2120,6 @@ extern void *__ARM_undef;
int (*)[__ARM_mve_type_int64_t_ptr]: __arm_vldrdq_gather_shifted_offset_z_s64 (__ARM_mve_coerce_s64_ptr(p0, int64_t *), p1, p2), \
int (*)[__ARM_mve_type_uint64_t_ptr]: __arm_vldrdq_gather_shifted_offset_z_u64 (__ARM_mve_coerce_u64_ptr(p0, uint64_t *), p1, p2)))
-#define __arm_vldrbq_gather_offset_z(p0,p1,p2) ({ __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_uint8x16_t]: __arm_vldrbq_gather_offset_z_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, uint8x16_t), p2), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_z_s16 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_z_s32 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint8x16_t]: __arm_vldrbq_gather_offset_z_u8 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint8x16_t), p2), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_z_u16 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_z_u32 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2));})
-
-#define __arm_vldrbq_gather_offset(p0,p1) ({ __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_uint8x16_t]: __arm_vldrbq_gather_offset_s8(__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, uint8x16_t)), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_s16(__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_s32(__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, uint32x4_t)), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint8x16_t]: __arm_vldrbq_gather_offset_u8(__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint8x16_t)), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_u16(__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_u32(__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint32x4_t)));})
-
#endif /* __cplusplus */
#endif /* __ARM_FEATURE_MVE */
#endif /* _GCC_ARM_MVE_H. */
diff --git a/gcc/config/arm/arm_mve_builtins.def b/gcc/config/arm/arm_mve_builtins.def
index 07f5a59..5f328f5 100644
--- a/gcc/config/arm/arm_mve_builtins.def
+++ b/gcc/config/arm/arm_mve_builtins.def
@@ -663,51 +663,31 @@ VAR2 (QUADOP_NONE_NONE_NONE_NONE_PRED, vandq_m_f, v8hf, v4sf)
VAR2 (QUADOP_NONE_NONE_NONE_NONE_PRED, vaddq_m_n_f, v8hf, v4sf)
VAR2 (QUADOP_NONE_NONE_NONE_NONE_PRED, vaddq_m_f, v8hf, v4sf)
VAR2 (QUADOP_NONE_NONE_NONE_NONE_PRED, vabdq_m_f, v8hf, v4sf)
-VAR3 (LDRGU, vldrbq_gather_offset_u, v16qi, v8hi, v4si)
-VAR3 (LDRGS, vldrbq_gather_offset_s, v16qi, v8hi, v4si)
VAR1 (LDRGBS, vldrwq_gather_base_s, v4si)
VAR1 (LDRGBU, vldrwq_gather_base_u, v4si)
VAR1 (LDRGBS_Z, vldrwq_gather_base_z_s, v4si)
VAR1 (LDRGBU_Z, vldrwq_gather_base_z_u, v4si)
-VAR3 (LDRGS_Z, vldrbq_gather_offset_z_s, v16qi, v8hi, v4si)
-VAR3 (LDRGU_Z, vldrbq_gather_offset_z_u, v16qi, v8hi, v4si)
VAR2 (LDRGU_Z, vldrhq_gather_shifted_offset_z_u, v8hi, v4si)
-VAR2 (LDRGU_Z, vldrhq_gather_offset_z_u, v8hi, v4si)
VAR2 (LDRGU, vldrhq_gather_shifted_offset_u, v8hi, v4si)
-VAR2 (LDRGU, vldrhq_gather_offset_u, v8hi, v4si)
VAR2 (LDRGS_Z, vldrhq_gather_shifted_offset_z_s, v8hi, v4si)
-VAR2 (LDRGS_Z, vldrhq_gather_offset_z_s, v8hi, v4si)
VAR2 (LDRGS, vldrhq_gather_shifted_offset_s, v8hi, v4si)
-VAR2 (LDRGS, vldrhq_gather_offset_s, v8hi, v4si)
VAR1 (LDRGBS, vldrdq_gather_base_s, v2di)
VAR1 (LDRGBS, vldrwq_gather_base_f, v4sf)
VAR1 (LDRGBS_Z, vldrdq_gather_base_z_s, v2di)
VAR1 (LDRGBS_Z, vldrwq_gather_base_z_f, v4sf)
VAR1 (LDRGBU, vldrdq_gather_base_u, v2di)
VAR1 (LDRGBU_Z, vldrdq_gather_base_z_u, v2di)
-VAR1 (LDRGS, vldrdq_gather_offset_s, v2di)
VAR1 (LDRGS, vldrdq_gather_shifted_offset_s, v2di)
-VAR1 (LDRGS, vldrhq_gather_offset_f, v8hf)
VAR1 (LDRGS, vldrhq_gather_shifted_offset_f, v8hf)
-VAR1 (LDRGS, vldrwq_gather_offset_f, v4sf)
-VAR1 (LDRGS, vldrwq_gather_offset_s, v4si)
VAR1 (LDRGS, vldrwq_gather_shifted_offset_f, v4sf)
VAR1 (LDRGS, vldrwq_gather_shifted_offset_s, v4si)
-VAR1 (LDRGS_Z, vldrdq_gather_offset_z_s, v2di)
VAR1 (LDRGS_Z, vldrdq_gather_shifted_offset_z_s, v2di)
-VAR1 (LDRGS_Z, vldrhq_gather_offset_z_f, v8hf)
VAR1 (LDRGS_Z, vldrhq_gather_shifted_offset_z_f, v8hf)
-VAR1 (LDRGS_Z, vldrwq_gather_offset_z_f, v4sf)
-VAR1 (LDRGS_Z, vldrwq_gather_offset_z_s, v4si)
VAR1 (LDRGS_Z, vldrwq_gather_shifted_offset_z_f, v4sf)
VAR1 (LDRGS_Z, vldrwq_gather_shifted_offset_z_s, v4si)
-VAR1 (LDRGU, vldrdq_gather_offset_u, v2di)
VAR1 (LDRGU, vldrdq_gather_shifted_offset_u, v2di)
-VAR1 (LDRGU, vldrwq_gather_offset_u, v4si)
VAR1 (LDRGU, vldrwq_gather_shifted_offset_u, v4si)
-VAR1 (LDRGU_Z, vldrdq_gather_offset_z_u, v2di)
VAR1 (LDRGU_Z, vldrdq_gather_shifted_offset_z_u, v2di)
-VAR1 (LDRGU_Z, vldrwq_gather_offset_z_u, v4si)
VAR1 (LDRGU_Z, vldrwq_gather_shifted_offset_z_u, v4si)
VAR1 (LDRGBWBU_Z, vldrwq_gather_base_nowb_z_u, v4si)
VAR1 (LDRGBWBU_Z, vldrdq_gather_base_nowb_z_u, v2di)
diff --git a/gcc/config/arm/iterators.md b/gcc/config/arm/iterators.md
index 155e9ef..4c0aac5 100644
--- a/gcc/config/arm/iterators.md
+++ b/gcc/config/arm/iterators.md
@@ -1797,6 +1797,16 @@
(DI "u64") (V2DI "u64")
(V2SF "f32") (V4SF "f32")])
+;; Same, but for MVE gather loads.
+;; Note that using "uNN" or "NN" everywhere would work too.
+;; We use this to match the expected output described in ACLE.
+(define_mode_attr MVE_u_elem [(V16QI "u8")
+ (V8HI "u16")
+ (V4SI "u32")
+ (V2DI "u64")
+ (V8HF "f16")
+ (V4SF "u32")])
+
;; Element types for extraction of unsigned scalars.
(define_mode_attr V_uf_sclr [(V8QI "u8") (V16QI "u8")
(V4HI "u16") (V8HI "u16")
@@ -2527,12 +2537,12 @@
(VQRSHRNBQ_M_N_S "s") (VQRSHRNBQ_M_N_U "u")
(VMLALDAVAXQ_P_S "s")
(VMLALDAVAQ_P_S "s") (VMLALDAVAQ_P_U "u")
- (VLDRBQGO_S "s") (VLDRBQGO_U "u") (VLDRWQGB_S "s")
- (VLDRWQGB_U "u") (VLDRHQGO_S "s")
- (VLDRHQGO_U "u") (VLDRHQGSO_S "s") (VLDRHQGSO_U "u")
+ (VLDRWQGB_S "s")
+ (VLDRWQGB_U "u")
+ (VLDRHQGSO_S "s") (VLDRHQGSO_U "u")
(VLDRDQGB_S "s") (VLDRDQGB_U "u")
- (VLDRDQGO_S "s") (VLDRDQGO_U "u") (VLDRDQGSO_S "s")
- (VLDRDQGSO_U "u") (VLDRWQGO_S "s") (VLDRWQGO_U "u")
+ (VLDRDQGSO_S "s")
+ (VLDRDQGSO_U "u")
(VLDRWQGSO_S "s") (VLDRWQGSO_U "u")
(VSTRDQSB_S "s") (VSTRDQSB_U "u")
(VLDRWQGBWB_S "s") (VLDRWQGBWB_U "u") (VLDRDQGBWB_S "s")
@@ -2935,14 +2945,10 @@
(define_int_iterator VSHLLxQ_M_N [VSHLLBQ_M_N_U VSHLLBQ_M_N_S 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 VLDRBGOQ [VLDRBQGO_S VLDRBQGO_U])
(define_int_iterator VLDRWGBQ [VLDRWQGB_S VLDRWQGB_U])
-(define_int_iterator VLDRHGOQ [VLDRHQGO_S VLDRHQGO_U])
(define_int_iterator VLDRHGSOQ [VLDRHQGSO_S VLDRHQGSO_U])
(define_int_iterator VLDRDGBQ [VLDRDQGB_S VLDRDQGB_U])
-(define_int_iterator VLDRDGOQ [VLDRDQGO_S VLDRDQGO_U])
(define_int_iterator VLDRDGSOQ [VLDRDQGSO_S VLDRDQGSO_U])
-(define_int_iterator VLDRWGOQ [VLDRWQGO_S VLDRWQGO_U])
(define_int_iterator VLDRWGSOQ [VLDRWQGSO_S VLDRWQGSO_U])
(define_int_iterator VLDRWGBWBQ [VLDRWQGBWB_S VLDRWQGBWB_U])
(define_int_iterator VLDRDGBWBQ [VLDRDQGBWB_S VLDRDQGBWB_U])
diff --git a/gcc/config/arm/mve.md b/gcc/config/arm/mve.md
index 1963a1e..b437fc9 100644
--- a/gcc/config/arm/mve.md
+++ b/gcc/config/arm/mve.md
@@ -3416,30 +3416,92 @@
[(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrq_scatter_base_<mode>"))
(set_attr "length" "4")])
+;; Vector gather loads with offset
;;
;; [vldrbq_gather_offset_s vldrbq_gather_offset_u]
+;; [vldrhq_gather_offset_s vldrhq_gather_offset_u]
+;; [vldrhq_gather_offset_f]
+;; [vldrwq_gather_offset_s vldrwq_gather_offset_u]
+;; [vldrwq_gather_offset_f]
+;; [vldrdq_gather_offset_s vldrdq_gather_offset_u]
;;
-(define_insn "mve_vldrbq_gather_offset_<supf><mode>"
- [(set (match_operand:MVE_2 0 "s_register_operand" "=&w")
- (unspec:MVE_2 [(match_operand:<MVE_B_ELEM> 1 "memory_operand" "Us")
- (match_operand:MVE_2 2 "s_register_operand" "w")]
- VLDRBGOQ))
+(define_insn "@mve_vldrq_gather_offset_<mode>"
+ [(set (match_operand:MVE_VLD_ST_scatter 0 "s_register_operand" "=&w")
+ (unspec:MVE_VLD_ST_scatter
+ [(match_operand:SI 1 "register_operand" "r")
+ (match_operand:<MVE_scatter_offset> 2 "s_register_operand" "w")
+ (mem:BLK (scratch))]
+ VLDRGOQ))
]
- "TARGET_HAVE_MVE"
-{
- rtx ops[3];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- if (!strcmp ("<supf>","s") && <V_sz_elem> == 8)
- output_asm_insn ("vldrb.u8\t%q0, [%m1, %q2]",ops);
- else
- output_asm_insn ("vldrb.<supf><V_sz_elem>\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrbq_gather_offset_<supf><mode>"))
+ "(TARGET_HAVE_MVE && VALID_MVE_SI_MODE (<MODE>mode))
+ || (TARGET_HAVE_MVE_FLOAT && VALID_MVE_SF_MODE (<MODE>mode))"
+ "vldr<MVE_elem_ch>.<MVE_u_elem>\t%q0, [%1, %q2]"
+ [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrq_gather_offset_<mode>"))
+ (set_attr "length" "4")])
+
+;; Extending vector gather loads with offset
+;;
+;; [vldrbq_gather_offset_s vldrbq_gather_offset_u]
+;; [vldrhq_gather_offset_s vldrhq_gather_offset_u]
+;;
+(define_insn "@mve_vldrq_gather_offset_extend_<mode><US>"
+ [(set (match_operand:<MVE_wide_n_TYPE> 0 "s_register_operand" "=&w")
+ (SE:<MVE_wide_n_TYPE>
+ (unspec:MVE_w_narrow_TYPE
+ [(match_operand:SI 1 "register_operand" "r")
+ (match_operand:<MVE_wide_n_TYPE> 2 "s_register_operand" "w")
+ (mem:BLK (scratch))]
+ VLDRGOQ_EXT)))
+ ]
+ "(TARGET_HAVE_MVE && VALID_MVE_SI_MODE (<MVE_wide_n_TYPE>mode))"
+ "vldr<MVE_elem_ch>.<US><MVE_wide_n_sz_elem>\t%q0, [%1, %q2]"
+ [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrq_gather_offset_extend_<mode><US>"))
(set_attr "length" "4")])
+;; Predicated gather loads with offset
+;;
+;; [vldrbq_gather_offset_z_s vldrbq_gather_offset_z_u]
+;; [vldrhq_gather_offset_z_s vldrhq_gather_offset_z_u]
+;; [vldrhq_gather_offset_z_f]
+;; [vldrwq_gather_offset_z_s vldrwq_gather_offset_z_u]
+;; [vldrwq_gather_offset_z_f]
+;; [vldrdq_gather_offset_z_s vldrdq_gather_offset_z_u]
+;;
+(define_insn "@mve_vldrq_gather_offset_z_<mode>"
+ [(set (match_operand:MVE_VLD_ST_scatter 0 "s_register_operand" "=&w")
+ (unspec:MVE_VLD_ST_scatter
+ [(match_operand:SI 1 "register_operand" "r")
+ (match_operand:<MVE_scatter_offset> 2 "s_register_operand" "w")
+ (match_operand:<MVE_VPRED> 3 "vpr_register_operand" "Up")
+ (mem:BLK (scratch))]
+ VLDRGOQ_Z))
+ ]
+ "(TARGET_HAVE_MVE && VALID_MVE_SI_MODE (<MODE>mode))
+ || (TARGET_HAVE_MVE_FLOAT && VALID_MVE_SF_MODE (<MODE>mode))"
+ "vpst\n\tvldr<MVE_elem_ch>t.<MVE_u_elem>\t%q0, [%1, %q2]"
+ [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrq_gather_offset_<mode>"))
+ (set_attr "length" "8")])
+
+;; Predicated extending gather loads with offset
+;;
+;; [vldrbq_gather_offset_z_s vldrbq_gather_offset_z_u]
+;; [vldrhq_gather_offset_z_s vldrhq_gather_offset_z_u]
+;;
+(define_insn "@mve_vldrq_gather_offset_z_extend_<mode><US>"
+ [(set (match_operand:<MVE_wide_n_TYPE> 0 "s_register_operand" "=&w")
+ (SE:<MVE_wide_n_TYPE>
+ (unspec:MVE_w_narrow_TYPE
+ [(match_operand:SI 1 "register_operand" "r")
+ (match_operand:<MVE_wide_n_TYPE> 2 "s_register_operand" "w")
+ (match_operand:<MVE_wide_n_VPRED> 3 "vpr_register_operand" "Up")
+ (mem:BLK (scratch))]
+ VLDRGOQ_EXT_Z)))
+ ]
+ "(TARGET_HAVE_MVE && VALID_MVE_SI_MODE (<MVE_wide_n_TYPE>mode))"
+ "vpst\n\tvldr<MVE_elem_ch>t.<US><MVE_wide_n_sz_elem>\t%q0, [%1, %q2]"
+ [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrq_gather_offset_extend_<mode><US>"))
+ (set_attr "length" "8")])
+
;;
;; [vldrwq_gather_base_s vldrwq_gather_base_u]
;;
@@ -3483,32 +3545,6 @@
(set_attr "length" "8")])
;;
-;; [vldrbq_gather_offset_z_s vldrbq_gather_offset_z_u]
-;;
-(define_insn "mve_vldrbq_gather_offset_z_<supf><mode>"
- [(set (match_operand:MVE_2 0 "s_register_operand" "=&w")
- (unspec:MVE_2 [(match_operand:<MVE_B_ELEM> 1 "memory_operand" "Us")
- (match_operand:MVE_2 2 "s_register_operand" "w")
- (match_operand:<MVE_VPRED> 3 "vpr_register_operand" "Up")]
- VLDRBGOQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[4];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- ops[3] = operands[3];
- if (!strcmp ("<supf>","s") && <V_sz_elem> == 8)
- output_asm_insn ("vpst\n\tvldrbt.u8\t%q0, [%m1, %q2]",ops);
- else
- output_asm_insn ("vpst\n\tvldrbt.<supf><V_sz_elem>\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrbq_gather_offset_<supf><mode>"))
- (set_attr "length" "8")])
-
-;;
;; [vldrwq_gather_base_z_s vldrwq_gather_base_z_u]
;;
(define_insn "mve_vldrwq_gather_base_z_<supf>v4si"
@@ -3531,56 +3567,6 @@
(set_attr "length" "8")])
;;
-;; [vldrhq_gather_offset_s vldrhq_gather_offset_u]
-;;
-(define_insn "mve_vldrhq_gather_offset_<supf><mode>"
- [(set (match_operand:MVE_5 0 "s_register_operand" "=&w")
- (unspec:MVE_5 [(match_operand:<MVE_H_ELEM> 1 "memory_operand" "Us")
- (match_operand:MVE_5 2 "s_register_operand" "w")]
- VLDRHGOQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[3];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- if (!strcmp ("<supf>","s") && <V_sz_elem> == 16)
- output_asm_insn ("vldrh.u16\t%q0, [%m1, %q2]",ops);
- else
- output_asm_insn ("vldrh.<supf><V_sz_elem>\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrhq_gather_offset_<supf><mode>"))
- (set_attr "length" "4")])
-
-;;
-;; [vldrhq_gather_offset_z_s vldrhq_gather_offset_z_u]
-;;
-(define_insn "mve_vldrhq_gather_offset_z_<supf><mode>"
- [(set (match_operand:MVE_5 0 "s_register_operand" "=&w")
- (unspec:MVE_5 [(match_operand:<MVE_H_ELEM> 1 "memory_operand" "Us")
- (match_operand:MVE_5 2 "s_register_operand" "w")
- (match_operand:<MVE_VPRED> 3 "vpr_register_operand" "Up")
- ]VLDRHGOQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[4];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- ops[3] = operands[3];
- if (!strcmp ("<supf>","s") && <V_sz_elem> == 16)
- output_asm_insn ("vpst\n\tvldrht.u16\t%q0, [%m1, %q2]",ops);
- else
- output_asm_insn ("vpst\n\tvldrht.<supf><V_sz_elem>\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrhq_gather_offset_<supf><mode>"))
- (set_attr "length" "8")])
-
-;;
;; [vldrhq_gather_shifted_offset_s vldrhq_gather_shifted_offset_u]
;;
(define_insn "mve_vldrhq_gather_shifted_offset_<supf><mode>"
@@ -3674,49 +3660,6 @@
(set_attr "length" "8")])
;;
-;; [vldrdq_gather_offset_s vldrdq_gather_offset_u]
-;;
-(define_insn "mve_vldrdq_gather_offset_<supf>v2di"
- [(set (match_operand:V2DI 0 "s_register_operand" "=&w")
- (unspec:V2DI [(match_operand:V2DI 1 "memory_operand" "Us")
- (match_operand:V2DI 2 "s_register_operand" "w")]
- VLDRDGOQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[3];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- output_asm_insn ("vldrd.u64\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrdq_gather_offset_<supf>v2di"))
- (set_attr "length" "4")])
-
-;;
-;; [vldrdq_gather_offset_z_s vldrdq_gather_offset_z_u]
-;;
-(define_insn "mve_vldrdq_gather_offset_z_<supf>v2di"
- [(set (match_operand:V2DI 0 "s_register_operand" "=&w")
- (unspec:V2DI [(match_operand:V2DI 1 "memory_operand" "Us")
- (match_operand:V2DI 2 "s_register_operand" "w")
- (match_operand:V2QI 3 "vpr_register_operand" "Up")]
- VLDRDGOQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[3];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- output_asm_insn ("vpst\n\tvldrdt.u64\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrdq_gather_offset_<supf>v2di"))
- (set_attr "length" "8")])
-
-;;
;; [vldrdq_gather_shifted_offset_s vldrdq_gather_shifted_offset_u]
;;
(define_insn "mve_vldrdq_gather_shifted_offset_<supf>v2di"
@@ -3760,50 +3703,6 @@
(set_attr "length" "8")])
;;
-;; [vldrhq_gather_offset_f]
-;;
-(define_insn "mve_vldrhq_gather_offset_fv8hf"
- [(set (match_operand:V8HF 0 "s_register_operand" "=&w")
- (unspec:V8HF [(match_operand:V8HI 1 "memory_operand" "Us")
- (match_operand:V8HI 2 "s_register_operand" "w")]
- VLDRHQGO_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[3];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- output_asm_insn ("vldrh.f16\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrhq_gather_offset_fv8hf"))
- (set_attr "length" "4")])
-
-;;
-;; [vldrhq_gather_offset_z_f]
-;;
-(define_insn "mve_vldrhq_gather_offset_z_fv8hf"
- [(set (match_operand:V8HF 0 "s_register_operand" "=&w")
- (unspec:V8HF [(match_operand:V8HI 1 "memory_operand" "Us")
- (match_operand:V8HI 2 "s_register_operand" "w")
- (match_operand:V8BI 3 "vpr_register_operand" "Up")]
- VLDRHQGO_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[4];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- ops[3] = operands[3];
- output_asm_insn ("vpst\n\tvldrht.f16\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrhq_gather_offset_fv8hf"))
- (set_attr "length" "8")])
-
-;;
;; [vldrhq_gather_shifted_offset_f]
;;
(define_insn "mve_vldrhq_gather_shifted_offset_fv8hf"
@@ -3891,94 +3790,6 @@
(set_attr "length" "8")])
;;
-;; [vldrwq_gather_offset_f]
-;;
-(define_insn "mve_vldrwq_gather_offset_fv4sf"
- [(set (match_operand:V4SF 0 "s_register_operand" "=&w")
- (unspec:V4SF [(match_operand:V4SI 1 "memory_operand" "Us")
- (match_operand:V4SI 2 "s_register_operand" "w")]
- VLDRWQGO_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[3];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- output_asm_insn ("vldrw.u32\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_gather_offset_fv4sf"))
- (set_attr "length" "4")])
-
-;;
-;; [vldrwq_gather_offset_s vldrwq_gather_offset_u]
-;;
-(define_insn "mve_vldrwq_gather_offset_<supf>v4si"
- [(set (match_operand:V4SI 0 "s_register_operand" "=&w")
- (unspec:V4SI [(match_operand:V4SI 1 "memory_operand" "Us")
- (match_operand:V4SI 2 "s_register_operand" "w")]
- VLDRWGOQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[3];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- output_asm_insn ("vldrw.u32\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_gather_offset_<supf>v4si"))
- (set_attr "length" "4")])
-
-;;
-;; [vldrwq_gather_offset_z_f]
-;;
-(define_insn "mve_vldrwq_gather_offset_z_fv4sf"
- [(set (match_operand:V4SF 0 "s_register_operand" "=&w")
- (unspec:V4SF [(match_operand:V4SI 1 "memory_operand" "Us")
- (match_operand:V4SI 2 "s_register_operand" "w")
- (match_operand:V4BI 3 "vpr_register_operand" "Up")]
- VLDRWQGO_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[4];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- ops[3] = operands[3];
- output_asm_insn ("vpst\n\tvldrwt.u32\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_gather_offset_fv4sf"))
- (set_attr "length" "8")])
-
-;;
-;; [vldrwq_gather_offset_z_s vldrwq_gather_offset_z_u]
-;;
-(define_insn "mve_vldrwq_gather_offset_z_<supf>v4si"
- [(set (match_operand:V4SI 0 "s_register_operand" "=&w")
- (unspec:V4SI [(match_operand:V4SI 1 "memory_operand" "Us")
- (match_operand:V4SI 2 "s_register_operand" "w")
- (match_operand:V4BI 3 "vpr_register_operand" "Up")]
- VLDRWGOQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[4];
- ops[0] = operands[0];
- ops[1] = operands[1];
- ops[2] = operands[2];
- ops[3] = operands[3];
- output_asm_insn ("vpst\n\tvldrwt.u32\t%q0, [%m1, %q2]",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_gather_offset_<supf>v4si"))
- (set_attr "length" "8")])
-
-;;
;; [vldrwq_gather_shifted_offset_f]
;;
(define_insn "mve_vldrwq_gather_shifted_offset_fv4sf"
diff --git a/gcc/config/arm/unspecs.md b/gcc/config/arm/unspecs.md
index 1829089..cdad4ed 100644
--- a/gcc/config/arm/unspecs.md
+++ b/gcc/config/arm/unspecs.md
@@ -1156,26 +1156,20 @@
VLDRQ_Z
VLDRQ_EXT
VLDRQ_EXT_Z
- VLDRBQGO_S
- VLDRBQGO_U
+ VLDRGOQ
+ VLDRGOQ_Z
+ VLDRGOQ_EXT
+ VLDRGOQ_EXT_Z
VLDRWQGB_S
VLDRWQGB_U
- VLDRHQGO_S
- VLDRHQGO_U
VLDRHQGSO_S
VLDRHQGSO_U
VLDRDQGB_S
VLDRDQGB_U
- VLDRDQGO_S
- VLDRDQGO_U
VLDRDQGSO_S
VLDRDQGSO_U
- VLDRHQGO_F
VLDRHQGSO_F
VLDRWQGB_F
- VLDRWQGO_F
- VLDRWQGO_S
- VLDRWQGO_U
VLDRWQGSO_F
VLDRWQGSO_S
VLDRWQGSO_U
generated by cgit v1.1 at 2024-12-26 06:40:46 +0000