@@ -588,6 +588,36 @@ arm_strsbu_qualifiers[SIMD_MAX_BUILTIN_ARGS]
qualifier_unsigned};
#define STRSBU_QUALIFIERS (arm_strsbu_qualifiers)
+static enum arm_type_qualifiers
+arm_ldrgu_qualifiers[SIMD_MAX_BUILTIN_ARGS]
+ = { qualifier_unsigned, qualifier_pointer, qualifier_unsigned};
+#define LDRGU_QUALIFIERS (arm_ldrgu_qualifiers)
+
+static enum arm_type_qualifiers
+arm_ldrgs_qualifiers[SIMD_MAX_BUILTIN_ARGS]
+ = { qualifier_none, qualifier_pointer, qualifier_unsigned};
+#define LDRGS_QUALIFIERS (arm_ldrgs_qualifiers)
+
+static enum arm_type_qualifiers
+arm_ldrs_qualifiers[SIMD_MAX_BUILTIN_ARGS]
+ = { qualifier_none, qualifier_pointer};
+#define LDRS_QUALIFIERS (arm_ldrs_qualifiers)
+
+static enum arm_type_qualifiers
+arm_ldru_qualifiers[SIMD_MAX_BUILTIN_ARGS]
+ = { qualifier_unsigned, qualifier_pointer};
+#define LDRU_QUALIFIERS (arm_ldru_qualifiers)
+
+static enum arm_type_qualifiers
+arm_ldrgbs_qualifiers[SIMD_MAX_BUILTIN_ARGS]
+ = { qualifier_none, qualifier_unsigned, qualifier_immediate};
+#define LDRGBS_QUALIFIERS (arm_ldrgbs_qualifiers)
+
+static enum arm_type_qualifiers
+arm_ldrgbu_qualifiers[SIMD_MAX_BUILTIN_ARGS]
+ = { qualifier_unsigned, qualifier_unsigned, qualifier_immediate};
+#define LDRGBU_QUALIFIERS (arm_ldrgbu_qualifiers)
+
/* End of Qualifier for MVE builtins. */
/* void ([T element type] *, T, immediate). */
@@ -1716,6 +1716,20 @@ typedef struct { uint8x16_t val[4]; } uint8x16x4_t;
#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)
+#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_s8(__base) __arm_vldrbq_s8(__base)
+#define vldrbq_u8(__base) __arm_vldrbq_u8(__base)
+#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_s16(__base) __arm_vldrbq_s16(__base)
+#define vldrbq_u16(__base) __arm_vldrbq_u16(__base)
+#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 vldrbq_s32(__base) __arm_vldrbq_s32(__base)
+#define vldrbq_u32(__base) __arm_vldrbq_u32(__base)
+#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)
#endif
__extension__ extern __inline void
@@ -11106,6 +11120,105 @@ __arm_vstrwq_scatter_base_u32 (uint32x4_t __addr, const int __offset, uint32x4_t
{
__builtin_mve_vstrwq_scatter_base_uv4si (__addr, __offset, __value);
}
+
+__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 int8x16_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+__arm_vldrbq_s8 (int8_t const * __base)
+{
+ return __builtin_mve_vldrbq_sv16qi ((__builtin_neon_qi *) __base);
+}
+
+__extension__ extern __inline uint8x16_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+__arm_vldrbq_u8 (uint8_t const * __base)
+{
+ return __builtin_mve_vldrbq_uv16qi ((__builtin_neon_qi *) __base);
+}
+
+__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 int16x8_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+__arm_vldrbq_s16 (int8_t const * __base)
+{
+ return __builtin_mve_vldrbq_sv8hi ((__builtin_neon_qi *) __base);
+}
+
+__extension__ extern __inline uint16x8_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+__arm_vldrbq_u16 (uint8_t const * __base)
+{
+ return __builtin_mve_vldrbq_uv8hi ((__builtin_neon_qi *) __base);
+}
+
+__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_vldrbq_s32 (int8_t const * __base)
+{
+ return __builtin_mve_vldrbq_sv4si ((__builtin_neon_qi *) __base);
+}
+
+__extension__ extern __inline uint32x4_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+__arm_vldrbq_u32 (uint8_t const * __base)
+{
+ return __builtin_mve_vldrbq_uv4si ((__builtin_neon_qi *) __base);
+}
+
+__extension__ extern __inline int32x4_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+__arm_vldrwq_gather_base_s32 (uint32x4_t __addr, const int __offset)
+{
+ return __builtin_mve_vldrwq_gather_base_sv4si (__addr, __offset);
+}
+
+__extension__ extern __inline uint32x4_t
+__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
+__arm_vldrwq_gather_base_u32 (uint32x4_t __addr, const int __offset)
+{
+ return __builtin_mve_vldrwq_gather_base_uv4si (__addr, __offset);
+}
+
#if (__ARM_FEATURE_MVE & 2) /* MVE Floating point. */
__extension__ extern __inline void
@@ -17682,6 +17795,17 @@ extern void *__ARM_undef;
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)));})
+#define vldrbq_gather_offset(p0,p1) __arm_vldrbq_gather_offset(p0,p1)
+#define __arm_vldrbq_gather_offset(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_const_ptr][__ARM_mve_type_uint8x16_t]: __arm_vldrbq_gather_offset_s8 (__ARM_mve_coerce(__p0, int8_t const *), __ARM_mve_coerce(__p1, uint8x16_t)), \
+ int (*)[__ARM_mve_type_int8_t_const_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_s16 (__ARM_mve_coerce(__p0, int8_t const *), __ARM_mve_coerce(__p1, uint16x8_t)), \
+ int (*)[__ARM_mve_type_int8_t_const_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_s32 (__ARM_mve_coerce(__p0, int8_t const *), __ARM_mve_coerce(__p1, uint32x4_t)), \
+ int (*)[__ARM_mve_type_uint8_t_const_ptr][__ARM_mve_type_uint8x16_t]: __arm_vldrbq_gather_offset_u8 (__ARM_mve_coerce(__p0, uint8_t const *), __ARM_mve_coerce(__p1, uint8x16_t)), \
+ int (*)[__ARM_mve_type_uint8_t_const_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_u16 (__ARM_mve_coerce(__p0, uint8_t const *), __ARM_mve_coerce(__p1, uint16x8_t)), \
+ int (*)[__ARM_mve_type_uint8_t_const_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_u32 (__ARM_mve_coerce(__p0, uint8_t const *), __ARM_mve_coerce(__p1, uint32x4_t)));})
+
#endif /* MVE Floating point. */
#ifdef __cplusplus
@@ -691,3 +691,9 @@ 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)
+VAR3 (LDRGU, vldrbq_gather_offset_u, v16qi, v8hi, v4si)
+VAR3 (LDRGS, vldrbq_gather_offset_s, v16qi, v8hi, v4si)
+VAR3 (LDRS, vldrbq_s, v16qi, v8hi, v4si)
+VAR3 (LDRU, vldrbq_u, v16qi, v8hi, v4si)
+VAR1 (LDRGBS, vldrwq_gather_base_s, v4si)
+VAR1 (LDRGBU, vldrwq_gather_base_u, v4si)
@@ -192,7 +192,8 @@
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 VSTRWQSB_S VSTRWQSB_U
- VSTRBQSO_S VSTRBQSO_U VSTRBQ_S VSTRBQ_U])
+ VSTRBQSO_S VSTRBQSO_U VSTRBQ_S VSTRBQ_U VLDRBQGO_S
+ VLDRBQGO_U VLDRBQ_S VLDRBQ_U VLDRWQGB_S VLDRWQGB_U])
(define_mode_attr MVE_CNVT [(V8HI "V8HF") (V4SI "V4SF")
(V8HF "V8HI") (V4SF "V4SI")])
@@ -345,7 +346,9 @@
(VMLALDAVAXQ_P_S "s") (VMLALDAVAXQ_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")])
+ (VSTRBQSO_U "u") (VSTRBQ_S "s") (VSTRBQ_U "u")
+ (VLDRBQGO_S "s") (VLDRBQGO_U "u") (VLDRBQ_S "s")
+ (VLDRBQ_U "u") (VLDRWQGB_S "s") (VLDRWQGB_U "u")])
(define_int_attr mode1 [(VCTP8Q "8") (VCTP16Q "16") (VCTP32Q "32")
(VCTP64Q "64") (VCTP8Q_M "8") (VCTP16Q_M "16")
@@ -569,6 +572,9 @@
(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_int_iterator VLDRBGOQ [VLDRBQGO_S VLDRBQGO_U])
+(define_int_iterator VLDRBQ [VLDRBQ_S VLDRBQ_U])
+(define_int_iterator VLDRWGBQ [VLDRWQGB_S VLDRWQGB_U])
(define_insn "*mve_mov<mode>"
[(set (match_operand:MVE_types 0 "s_register_operand" "=w,w,r,w,w,r,w")
@@ -8002,3 +8008,65 @@
return "";
}
[(set_attr "length" "4")])
+
+;;
+;; [vldrbq_gather_offset_s vldrbq_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))
+ ]
+ "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 "length" "4")])
+
+;;
+;; [vldrbq_s vldrbq_u]
+;;
+(define_insn "mve_vldrbq_<supf><mode>"
+ [(set (match_operand:MVE_2 0 "s_register_operand" "=w")
+ (unspec:MVE_2 [(match_operand:<MVE_B_ELEM> 1 "memory_operand" "Us")]
+ VLDRBQ))
+ ]
+ "TARGET_HAVE_MVE"
+{
+ rtx ops[2];
+ int regno = REGNO (operands[0]);
+ ops[0] = gen_rtx_REG (TImode, regno);
+ ops[1] = operands[1];
+ output_asm_insn ("vldrb.<supf><V_sz_elem>\t%q0, %E1",ops);
+ return "";
+}
+ [(set_attr "length" "4")])
+
+;;
+;; [vldrwq_gather_base_s vldrwq_gather_base_u]
+;;
+(define_insn "mve_vldrwq_gather_base_<supf>v4si"
+ [(set (match_operand:V4SI 0 "s_register_operand" "=&w")
+ (unspec:V4SI [(match_operand:V4SI 1 "s_register_operand" "w")
+ (match_operand:SI 2 "immediate_operand" "i")]
+ VLDRWGBQ))
+ ]
+ "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, [%q1, %2]",ops);
+ return "";
+}
+ [(set_attr "length" "4")])
new file mode 100644
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+int16x8_t
+foo (int8_t const * base, uint16x8_t offset)
+{
+ return vldrbq_gather_offset_s16 (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.s16" } } */
+
+int16x8_t
+foo1 (int8_t const * base, uint16x8_t offset)
+{
+ return vldrbq_gather_offset (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.s16" } } */
new file mode 100644
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+int32x4_t
+foo (int8_t const * base, uint32x4_t offset)
+{
+ return vldrbq_gather_offset_s32 (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.s32" } } */
+
+int32x4_t
+foo1 (int8_t const * base, uint32x4_t offset)
+{
+ return vldrbq_gather_offset (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.s32" } } */
new file mode 100644
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+int8x16_t
+foo (int8_t const * base, uint8x16_t offset)
+{
+ return vldrbq_gather_offset_s8 (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.u8" } } */
+
+int8x16_t
+foo1 (int8_t const * base, uint8x16_t offset)
+{
+ return vldrbq_gather_offset (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.u8" } } */
new file mode 100644
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+uint16x8_t
+foo (uint8_t const * base, uint16x8_t offset)
+{
+ return vldrbq_gather_offset_u16 (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.u16" } } */
+
+uint16x8_t
+foo1 (uint8_t const * base, uint16x8_t offset)
+{
+ return vldrbq_gather_offset (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.u16" } } */
new file mode 100644
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+uint32x4_t
+foo (uint8_t const * base, uint32x4_t offset)
+{
+ return vldrbq_gather_offset_u32 (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.u32" } } */
+
+uint32x4_t
+foo1 (uint8_t const * base, uint32x4_t offset)
+{
+ return vldrbq_gather_offset (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.u32" } } */
new file mode 100644
@@ -0,0 +1,21 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+uint8x16_t
+foo (uint8_t const * base, uint8x16_t offset)
+{
+ return vldrbq_gather_offset_u8 (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.u8" } } */
+
+uint8x16_t
+foo1 (uint8_t const * base, uint8x16_t offset)
+{
+ return vldrbq_gather_offset (base, offset);
+}
+
+/* { dg-final { scan-assembler "vldrb.u8" } } */
new file mode 100644
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+int16x8_t
+foo (int8_t const * base)
+{
+ return vldrbq_s16 (base);
+}
+
+/* { dg-final { scan-assembler "vldrb.s16" } } */
new file mode 100644
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+int32x4_t
+foo (int8_t const * base)
+{
+ return vldrbq_s32 (base);
+}
+
+/* { dg-final { scan-assembler "vldrb.s32" } } */
new file mode 100644
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+int8x16_t
+foo (int8_t const * base)
+{
+ return vldrbq_s8 (base);
+}
+
+/* { dg-final { scan-assembler "vldrb.s8" } } */
new file mode 100644
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+uint16x8_t
+foo (uint8_t const * base)
+{
+ return vldrbq_u16 (base);
+}
+
+/* { dg-final { scan-assembler "vldrb.u16" } } */
new file mode 100644
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+uint32x4_t
+foo (uint8_t const * base)
+{
+ return vldrbq_u32 (base);
+}
+
+/* { dg-final { scan-assembler "vldrb.u32" } } */
new file mode 100644
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+uint8x16_t
+foo (uint8_t const * base)
+{
+ return vldrbq_u8 (base);
+}
+
+/* { dg-final { scan-assembler "vldrb.u8" } } */
new file mode 100644
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+int32x4_t
+foo (uint32x4_t addr)
+{
+ return vldrwq_gather_base_s32 (addr, 4);
+}
+
+/* { dg-final { scan-assembler "vldrw.u32" } } */
new file mode 100644
@@ -0,0 +1,13 @@
+/* { dg-do compile } */
+/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */
+/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */
+
+#include "arm_mve.h"
+
+uint32x4_t
+foo (uint32x4_t addr)
+{
+ return vldrwq_gather_base_u32 (addr, 4);
+}
+
+/* { dg-final { scan-assembler "vldrw.u32" } } */
Hello, This patch supports the following MVE ACLE load intrinsics. vldrbq_gather_offset_u8, vldrbq_gather_offset_s8, vldrbq_s8, vldrbq_u8, vldrbq_gather_offset_u16, vldrbq_gather_offset_s16, vldrbq_s16, vldrbq_u16, vldrbq_gather_offset_u32, vldrbq_gather_offset_s32, vldrbq_s32, vldrbq_u32, vldrwq_gather_base_s32, vldrwq_gather_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 Regression tested on arm-none-eabi and found no regressions. Ok for trunk? Thanks, Srinath. gcc/ChangeLog: 2019-11-01 Andre Vieira <andre.simoesdiasvieira@arm.com> Mihail Ionescu <mihail.ionescu@arm.com> Srinath Parvathaneni <srinath.parvathaneni@arm.com> * config/arm/arm-builtins.c (LDRGU_QUALIFIERS): Define builtin qualifier. (LDRGS_QUALIFIERS): Likewise. (LDRS_QUALIFIERS): Likewise. (LDRU_QUALIFIERS): Likewise. (LDRGBS_QUALIFIERS): Likewise. (LDRGBU_QUALIFIERS): Likewise. * config/arm/arm_mve.h (vldrbq_gather_offset_u8): Define macro. (vldrbq_gather_offset_s8): Likewise. (vldrbq_s8): Likewise. (vldrbq_u8): Likewise. (vldrbq_gather_offset_u16): Likewise. (vldrbq_gather_offset_s16): Likewise. (vldrbq_s16): Likewise. (vldrbq_u16): Likewise. (vldrbq_gather_offset_u32): Likewise. (vldrbq_gather_offset_s32): Likewise. (vldrbq_s32): Likewise. (vldrbq_u32): Likewise. (vldrwq_gather_base_s32): Likewise. (vldrwq_gather_base_u32): Likewise. (__arm_vldrbq_gather_offset_u8): Define intrinsic. (__arm_vldrbq_gather_offset_s8): Likewise. (__arm_vldrbq_s8): Likewise. (__arm_vldrbq_u8): Likewise. (__arm_vldrbq_gather_offset_u16): Likewise. (__arm_vldrbq_gather_offset_s16): Likewise. (__arm_vldrbq_s16): Likewise. (__arm_vldrbq_u16): Likewise. (__arm_vldrbq_gather_offset_u32): Likewise. (__arm_vldrbq_gather_offset_s32): Likewise. (__arm_vldrbq_s32): Likewise. (__arm_vldrbq_u32): Likewise. (__arm_vldrwq_gather_base_s32): Likewise. (__arm_vldrwq_gather_base_u32): Likewise. (vldrbq_gather_offset): Define polymorphic variant. * config/arm/arm_mve_builtins.def (LDRGU_QUALIFIERS): Use builtin qualifier. (LDRGS_QUALIFIERS): Likewise. (LDRS_QUALIFIERS): Likewise. (LDRU_QUALIFIERS): Likewise. (LDRGBS_QUALIFIERS): Likewise. (LDRGBU_QUALIFIERS): Likewise. * config/arm/mve.md (VLDRBGOQ): Define iterator. (VLDRBQ): Likewise. (VLDRWGBQ): Likewise. (mve_vldrbq_gather_offset_<supf><mode>): Define RTL pattern. (mve_vldrbq_<supf><mode>): Likewise. (mve_vldrwq_gather_base_<supf>v4si): Likewise. gcc/testsuite/ChangeLog: 2019-11-01 Andre Vieira <andre.simoesdiasvieira@arm.com> Mihail Ionescu <mihail.ionescu@arm.com> Srinath Parvathaneni <srinath.parvathaneni@arm.com> * gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s16.c: New test. * gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s32.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s8.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u16.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u32.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u8.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_s16.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_s32.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_s8.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_u16.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_u32.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrbq_u8.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrwq_gather_base_s32.c: Likewise. * gcc.target/arm/mve/intrinsics/vldrwq_gather_base_u32.c: Likewise. ############### Attachment also inlined for ease of reply ############### diff --git a/gcc/config/arm/arm-builtins.c b/gcc/config/arm/arm-builtins.c index ec88199bb5e7e9c15a346061c70841f3086004ef..02ea297937b18099f33a50c808964d1dd7eac1b3 100644 --- a/gcc/config/arm/arm-builtins.c +++ b/gcc/config/arm/arm-builtins.c @@ -588,6 +588,36 @@ arm_strsbu_qualifiers[SIMD_MAX_BUILTIN_ARGS] qualifier_unsigned}; #define STRSBU_QUALIFIERS (arm_strsbu_qualifiers) +static enum arm_type_qualifiers +arm_ldrgu_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_unsigned, qualifier_pointer, qualifier_unsigned}; +#define LDRGU_QUALIFIERS (arm_ldrgu_qualifiers) + +static enum arm_type_qualifiers +arm_ldrgs_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_none, qualifier_pointer, qualifier_unsigned}; +#define LDRGS_QUALIFIERS (arm_ldrgs_qualifiers) + +static enum arm_type_qualifiers +arm_ldrs_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_none, qualifier_pointer}; +#define LDRS_QUALIFIERS (arm_ldrs_qualifiers) + +static enum arm_type_qualifiers +arm_ldru_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_unsigned, qualifier_pointer}; +#define LDRU_QUALIFIERS (arm_ldru_qualifiers) + +static enum arm_type_qualifiers +arm_ldrgbs_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_none, qualifier_unsigned, qualifier_immediate}; +#define LDRGBS_QUALIFIERS (arm_ldrgbs_qualifiers) + +static enum arm_type_qualifiers +arm_ldrgbu_qualifiers[SIMD_MAX_BUILTIN_ARGS] + = { qualifier_unsigned, qualifier_unsigned, qualifier_immediate}; +#define LDRGBU_QUALIFIERS (arm_ldrgbu_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 3d64033dd23636b041cbd6ecb3eb18c3b9db3aed..c123de1f3d37054b779a9ee908525c242b2bc4ab 100644 --- a/gcc/config/arm/arm_mve.h +++ b/gcc/config/arm/arm_mve.h @@ -1716,6 +1716,20 @@ typedef struct { uint8x16_t val[4]; } uint8x16x4_t; #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) +#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_s8(__base) __arm_vldrbq_s8(__base) +#define vldrbq_u8(__base) __arm_vldrbq_u8(__base) +#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_s16(__base) __arm_vldrbq_s16(__base) +#define vldrbq_u16(__base) __arm_vldrbq_u16(__base) +#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 vldrbq_s32(__base) __arm_vldrbq_s32(__base) +#define vldrbq_u32(__base) __arm_vldrbq_u32(__base) +#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) #endif __extension__ extern __inline void @@ -11106,6 +11120,105 @@ __arm_vstrwq_scatter_base_u32 (uint32x4_t __addr, const int __offset, uint32x4_t { __builtin_mve_vstrwq_scatter_base_uv4si (__addr, __offset, __value); } + +__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 int8x16_t +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vldrbq_s8 (int8_t const * __base) +{ + return __builtin_mve_vldrbq_sv16qi ((__builtin_neon_qi *) __base); +} + +__extension__ extern __inline uint8x16_t +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vldrbq_u8 (uint8_t const * __base) +{ + return __builtin_mve_vldrbq_uv16qi ((__builtin_neon_qi *) __base); +} + +__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 int16x8_t +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vldrbq_s16 (int8_t const * __base) +{ + return __builtin_mve_vldrbq_sv8hi ((__builtin_neon_qi *) __base); +} + +__extension__ extern __inline uint16x8_t +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vldrbq_u16 (uint8_t const * __base) +{ + return __builtin_mve_vldrbq_uv8hi ((__builtin_neon_qi *) __base); +} + +__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_vldrbq_s32 (int8_t const * __base) +{ + return __builtin_mve_vldrbq_sv4si ((__builtin_neon_qi *) __base); +} + +__extension__ extern __inline uint32x4_t +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vldrbq_u32 (uint8_t const * __base) +{ + return __builtin_mve_vldrbq_uv4si ((__builtin_neon_qi *) __base); +} + +__extension__ extern __inline int32x4_t +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vldrwq_gather_base_s32 (uint32x4_t __addr, const int __offset) +{ + return __builtin_mve_vldrwq_gather_base_sv4si (__addr, __offset); +} + +__extension__ extern __inline uint32x4_t +__attribute__ ((__always_inline__, __gnu_inline__, __artificial__)) +__arm_vldrwq_gather_base_u32 (uint32x4_t __addr, const int __offset) +{ + return __builtin_mve_vldrwq_gather_base_uv4si (__addr, __offset); +} + #if (__ARM_FEATURE_MVE & 2) /* MVE Floating point. */ __extension__ extern __inline void @@ -17682,6 +17795,17 @@ extern void *__ARM_undef; 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)));}) +#define vldrbq_gather_offset(p0,p1) __arm_vldrbq_gather_offset(p0,p1) +#define __arm_vldrbq_gather_offset(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_const_ptr][__ARM_mve_type_uint8x16_t]: __arm_vldrbq_gather_offset_s8 (__ARM_mve_coerce(__p0, int8_t const *), __ARM_mve_coerce(__p1, uint8x16_t)), \ + int (*)[__ARM_mve_type_int8_t_const_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_s16 (__ARM_mve_coerce(__p0, int8_t const *), __ARM_mve_coerce(__p1, uint16x8_t)), \ + int (*)[__ARM_mve_type_int8_t_const_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_s32 (__ARM_mve_coerce(__p0, int8_t const *), __ARM_mve_coerce(__p1, uint32x4_t)), \ + int (*)[__ARM_mve_type_uint8_t_const_ptr][__ARM_mve_type_uint8x16_t]: __arm_vldrbq_gather_offset_u8 (__ARM_mve_coerce(__p0, uint8_t const *), __ARM_mve_coerce(__p1, uint8x16_t)), \ + int (*)[__ARM_mve_type_uint8_t_const_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_u16 (__ARM_mve_coerce(__p0, uint8_t const *), __ARM_mve_coerce(__p1, uint16x8_t)), \ + int (*)[__ARM_mve_type_uint8_t_const_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_u32 (__ARM_mve_coerce(__p0, uint8_t const *), __ARM_mve_coerce(__p1, uint32x4_t)));}) + #endif /* MVE Floating point. */ #ifdef __cplusplus diff --git a/gcc/config/arm/arm_mve_builtins.def b/gcc/config/arm/arm_mve_builtins.def index 149c4115b6167e3f2f0eb43fbdeaa8708ba5d723..60e50767a2ac80a05303e3df8512f2612c8bc8ef 100644 --- a/gcc/config/arm/arm_mve_builtins.def +++ b/gcc/config/arm/arm_mve_builtins.def @@ -691,3 +691,9 @@ 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) +VAR3 (LDRGU, vldrbq_gather_offset_u, v16qi, v8hi, v4si) +VAR3 (LDRGS, vldrbq_gather_offset_s, v16qi, v8hi, v4si) +VAR3 (LDRS, vldrbq_s, v16qi, v8hi, v4si) +VAR3 (LDRU, vldrbq_u, v16qi, v8hi, v4si) +VAR1 (LDRGBS, vldrwq_gather_base_s, v4si) +VAR1 (LDRGBU, vldrwq_gather_base_u, v4si) diff --git a/gcc/config/arm/mve.md b/gcc/config/arm/mve.md index eb365166e934f32f7695e5354258151202f378c0..7b416f9a4e04f8c02547953d6019787817688a36 100644 --- a/gcc/config/arm/mve.md +++ b/gcc/config/arm/mve.md @@ -192,7 +192,8 @@ 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 VSTRWQSB_S VSTRWQSB_U - VSTRBQSO_S VSTRBQSO_U VSTRBQ_S VSTRBQ_U]) + VSTRBQSO_S VSTRBQSO_U VSTRBQ_S VSTRBQ_U VLDRBQGO_S + VLDRBQGO_U VLDRBQ_S VLDRBQ_U VLDRWQGB_S VLDRWQGB_U]) (define_mode_attr MVE_CNVT [(V8HI "V8HF") (V4SI "V4SF") (V8HF "V8HI") (V4SF "V4SI")]) @@ -345,7 +346,9 @@ (VMLALDAVAXQ_P_S "s") (VMLALDAVAXQ_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")]) + (VSTRBQSO_U "u") (VSTRBQ_S "s") (VSTRBQ_U "u") + (VLDRBQGO_S "s") (VLDRBQGO_U "u") (VLDRBQ_S "s") + (VLDRBQ_U "u") (VLDRWQGB_S "s") (VLDRWQGB_U "u")]) (define_int_attr mode1 [(VCTP8Q "8") (VCTP16Q "16") (VCTP32Q "32") (VCTP64Q "64") (VCTP8Q_M "8") (VCTP16Q_M "16") @@ -569,6 +572,9 @@ (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_int_iterator VLDRBGOQ [VLDRBQGO_S VLDRBQGO_U]) +(define_int_iterator VLDRBQ [VLDRBQ_S VLDRBQ_U]) +(define_int_iterator VLDRWGBQ [VLDRWQGB_S VLDRWQGB_U]) (define_insn "*mve_mov<mode>" [(set (match_operand:MVE_types 0 "s_register_operand" "=w,w,r,w,w,r,w") @@ -8002,3 +8008,65 @@ return ""; } [(set_attr "length" "4")]) + +;; +;; [vldrbq_gather_offset_s vldrbq_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)) + ] + "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 "length" "4")]) + +;; +;; [vldrbq_s vldrbq_u] +;; +(define_insn "mve_vldrbq_<supf><mode>" + [(set (match_operand:MVE_2 0 "s_register_operand" "=w") + (unspec:MVE_2 [(match_operand:<MVE_B_ELEM> 1 "memory_operand" "Us")] + VLDRBQ)) + ] + "TARGET_HAVE_MVE" +{ + rtx ops[2]; + int regno = REGNO (operands[0]); + ops[0] = gen_rtx_REG (TImode, regno); + ops[1] = operands[1]; + output_asm_insn ("vldrb.<supf><V_sz_elem>\t%q0, %E1",ops); + return ""; +} + [(set_attr "length" "4")]) + +;; +;; [vldrwq_gather_base_s vldrwq_gather_base_u] +;; +(define_insn "mve_vldrwq_gather_base_<supf>v4si" + [(set (match_operand:V4SI 0 "s_register_operand" "=&w") + (unspec:V4SI [(match_operand:V4SI 1 "s_register_operand" "w") + (match_operand:SI 2 "immediate_operand" "i")] + VLDRWGBQ)) + ] + "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, [%q1, %2]",ops); + return ""; +} + [(set_attr "length" "4")]) diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s16.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s16.c new file mode 100644 index 0000000000000000000000000000000000000000..f7763b8a1f668e25c5b34ed429dd5dc9a5455676 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s16.c @@ -0,0 +1,21 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +int16x8_t +foo (int8_t const * base, uint16x8_t offset) +{ + return vldrbq_gather_offset_s16 (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.s16" } } */ + +int16x8_t +foo1 (int8_t const * base, uint16x8_t offset) +{ + return vldrbq_gather_offset (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.s16" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s32.c new file mode 100644 index 0000000000000000000000000000000000000000..768a4325beae64496e174f55932554a961cbc886 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s32.c @@ -0,0 +1,21 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +int32x4_t +foo (int8_t const * base, uint32x4_t offset) +{ + return vldrbq_gather_offset_s32 (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.s32" } } */ + +int32x4_t +foo1 (int8_t const * base, uint32x4_t offset) +{ + return vldrbq_gather_offset (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.s32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s8.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s8.c new file mode 100644 index 0000000000000000000000000000000000000000..4c52b5a009507bf68b6bd83699cd78e85b291572 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_s8.c @@ -0,0 +1,21 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +int8x16_t +foo (int8_t const * base, uint8x16_t offset) +{ + return vldrbq_gather_offset_s8 (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.u8" } } */ + +int8x16_t +foo1 (int8_t const * base, uint8x16_t offset) +{ + return vldrbq_gather_offset (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.u8" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u16.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u16.c new file mode 100644 index 0000000000000000000000000000000000000000..44b310add69aa8e7b74cf2c501013b136d5da041 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u16.c @@ -0,0 +1,21 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +uint16x8_t +foo (uint8_t const * base, uint16x8_t offset) +{ + return vldrbq_gather_offset_u16 (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.u16" } } */ + +uint16x8_t +foo1 (uint8_t const * base, uint16x8_t offset) +{ + return vldrbq_gather_offset (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.u16" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u32.c new file mode 100644 index 0000000000000000000000000000000000000000..8415ea701327b2a296e1e28e37537045b69ff6a9 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u32.c @@ -0,0 +1,21 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +uint32x4_t +foo (uint8_t const * base, uint32x4_t offset) +{ + return vldrbq_gather_offset_u32 (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.u32" } } */ + +uint32x4_t +foo1 (uint8_t const * base, uint32x4_t offset) +{ + return vldrbq_gather_offset (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.u32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u8.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u8.c new file mode 100644 index 0000000000000000000000000000000000000000..9fbe1aa419d572ffad9b319d3ad24d9ac913b7e6 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_gather_offset_u8.c @@ -0,0 +1,21 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +uint8x16_t +foo (uint8_t const * base, uint8x16_t offset) +{ + return vldrbq_gather_offset_u8 (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.u8" } } */ + +uint8x16_t +foo1 (uint8_t const * base, uint8x16_t offset) +{ + return vldrbq_gather_offset (base, offset); +} + +/* { dg-final { scan-assembler "vldrb.u8" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_s16.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_s16.c new file mode 100644 index 0000000000000000000000000000000000000000..a63c7367e9ae24b78f4819aea95e6e5d0ec85891 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_s16.c @@ -0,0 +1,13 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +int16x8_t +foo (int8_t const * base) +{ + return vldrbq_s16 (base); +} + +/* { dg-final { scan-assembler "vldrb.s16" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_s32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_s32.c new file mode 100644 index 0000000000000000000000000000000000000000..040c58d44c5e6bf09ad61d13c44f53dfd987e8e4 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_s32.c @@ -0,0 +1,13 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +int32x4_t +foo (int8_t const * base) +{ + return vldrbq_s32 (base); +} + +/* { dg-final { scan-assembler "vldrb.s32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_s8.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_s8.c new file mode 100644 index 0000000000000000000000000000000000000000..e5cc3deb79b909a19cbbcafe9213c91b67ac337a --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_s8.c @@ -0,0 +1,13 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +int8x16_t +foo (int8_t const * base) +{ + return vldrbq_s8 (base); +} + +/* { dg-final { scan-assembler "vldrb.s8" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_u16.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_u16.c new file mode 100644 index 0000000000000000000000000000000000000000..7dedbde6eeae066fdc04e9d6fb6fe626af4580db --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_u16.c @@ -0,0 +1,13 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +uint16x8_t +foo (uint8_t const * base) +{ + return vldrbq_u16 (base); +} + +/* { dg-final { scan-assembler "vldrb.u16" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_u32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_u32.c new file mode 100644 index 0000000000000000000000000000000000000000..3c3ec6c4ffe6b014bf58d0b7c2a2d501ddf8301b --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_u32.c @@ -0,0 +1,13 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +uint32x4_t +foo (uint8_t const * base) +{ + return vldrbq_u32 (base); +} + +/* { dg-final { scan-assembler "vldrb.u32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_u8.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_u8.c new file mode 100644 index 0000000000000000000000000000000000000000..5328d99bab94d62bd032d4da3509cd2c55a7841a --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrbq_u8.c @@ -0,0 +1,13 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +uint8x16_t +foo (uint8_t const * base) +{ + return vldrbq_u8 (base); +} + +/* { dg-final { scan-assembler "vldrb.u8" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrwq_gather_base_s32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrwq_gather_base_s32.c new file mode 100644 index 0000000000000000000000000000000000000000..14831a91b65900d040f74b90c592bc195af42874 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrwq_gather_base_s32.c @@ -0,0 +1,13 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +int32x4_t +foo (uint32x4_t addr) +{ + return vldrwq_gather_base_s32 (addr, 4); +} + +/* { dg-final { scan-assembler "vldrw.u32" } } */ diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrwq_gather_base_u32.c b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrwq_gather_base_u32.c new file mode 100644 index 0000000000000000000000000000000000000000..7753ea6a7c9ec83b90f3e723683c156b20617e23 --- /dev/null +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vldrwq_gather_base_u32.c @@ -0,0 +1,13 @@ +/* { dg-do compile } */ +/* { dg-additional-options "-march=armv8.1-m.main+mve -mfloat-abi=hard -O2" } */ +/* { dg-skip-if "Skip if not auto" {*-*-*} {"-mfpu=*"} {"-mfpu=auto"} } */ + +#include "arm_mve.h" + +uint32x4_t +foo (uint32x4_t addr) +{ + return vldrwq_gather_base_u32 (addr, 4); +} + +/* { dg-final { scan-assembler "vldrw.u32" } } */