procyberian/gcc
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

[AArch64] Refactor vector max and min RTL and builtins.

Browse source 
gcc/
	* config/aarch64/aarch64-simd-builtins.def
	(reduc_smax_): New.
	(reduc_smin_): Likewise.
	(reduc_umax_): Likewise.
	(reduc_umin_): Likewise.
	(reduc_smax_nan_): Likewise.
	(reduc_smin_nan_): Likewise.
	(fmax): Remove.
	(fmin): Likewise.
	(smax): Update for V2SF, V4SF and V2DF modes.
	(smin): Likewise.
	(smax_nan): New.
	(smin_nan): Likewise.
	* config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
	(<su><maxmin><mode>3): ...This, refactor.
	(s<maxmin><mode>3): New.
	(<maxmin_uns><mode>3): Likewise.
	(reduc_<maxmin_uns>_<mode>): Refactor.
	(reduc_<maxmin_uns>_v4sf): Likewise.
	(reduc_<maxmin_uns>_v2si): Likewise.
	(aarch64_<fmaxmin><mode>: Remove.
	* config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
	new builtin names.
	(vmin<q>_f<32,64>): Likewise.
	* config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
	(FMAXMIN): New.
	(su): Add mappings for smax, smin, umax, umin.
	(maxmin): New.
	(FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
	(FMAXMIN): Rename as...
	(FMAXMIN_UNS): ...This.
	(maxminv): Remove.
	(fmaxminv): Likewise.
	(fmaxmin): Likewise.
	(maxmin_uns): New.
	(maxmin_uns_op): Likewise.

From-SVN: r198497
This commit is contained in:
James Greenhalgh 2013-05-01 15:16:14 +00:00 • committed by James Greenhalgh
parent 92d9038e4f
commit 998eaf975b
5 changed files with 143 additions and 78 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

39
gcc/ChangeLog
View file

@ -1,3 +1,42 @@
2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
* config/aarch64/aarch64-simd-builtins.def
(reduc_smax_): New.
(reduc_smin_): Likewise.
(reduc_umax_): Likewise.
(reduc_umin_): Likewise.
(reduc_smax_nan_): Likewise.
(reduc_smin_nan_): Likewise.
(fmax): Remove.
(fmin): Likewise.
(smax): Update for V2SF, V4SF and V2DF modes.
(smin): Likewise.
(smax_nan): New.
(smin_nan): Likewise.
* config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
(<su><maxmin><mode>3): ...This, refactor.
(s<maxmin><mode>3): New.
(<maxmin_uns><mode>3): Likewise.
(reduc_<maxmin_uns>_<mode>): Refactor.
(reduc_<maxmin_uns>_v4sf): Likewise.
(reduc_<maxmin_uns>_v2si): Likewise.
(aarch64_<fmaxmin><mode>: Remove.
* config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
new builtin names.
(vmin<q>_f<32,64>): Likewise.
* config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
(FMAXMIN): New.
(su): Add mappings for smax, smin, umax, umin.
(maxmin): New.
(FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
(FMAXMIN): Rename as...
(FMAXMIN_UNS): ...This.
(maxminv): Remove.
(fmaxminv): Likewise.
(fmaxmin): Likewise.
(maxmin_uns): New.
(maxmin_uns_op): Likewise.
2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
* config/aarch64/arm_neon.h

22
gcc/config/aarch64/aarch64-simd-builtins.def
View file

@ -234,18 +234,26 @@
BUILTIN_VSDQ_I_DI (BINOP, cmgtu, 0)
BUILTIN_VSDQ_I_DI (BINOP, cmtst, 0)
/* Implemented by aarch64_<fmaxmin><mode>. */
BUILTIN_VDQF (BINOP, fmax, 0)
BUILTIN_VDQF (BINOP, fmin, 0)
/* Implemented by aarch64_addv<mode>. */
BUILTIN_VDQF (UNOP, addv, 0)
/* Implemented by <maxmin><mode>3. */
BUILTIN_VDQ_BHSI (BINOP, smax, 3)
BUILTIN_VDQ_BHSI (BINOP, smin, 3)
/* Implemented by reduc_<maxmin_uns>_<mode>. */
BUILTIN_VDQIF (UNOP, reduc_smax_, 10)
BUILTIN_VDQIF (UNOP, reduc_smin_, 10)
BUILTIN_VDQ_BHSI (UNOP, reduc_umax_, 10)
BUILTIN_VDQ_BHSI (UNOP, reduc_umin_, 10)
BUILTIN_VDQF (UNOP, reduc_smax_nan_, 10)
BUILTIN_VDQF (UNOP, reduc_smin_nan_, 10)
/* Implemented by <maxmin><mode>3.
smax variants map to fmaxnm,
smax_nan variants map to fmax. */
BUILTIN_VDQIF (BINOP, smax, 3)
BUILTIN_VDQIF (BINOP, smin, 3)
BUILTIN_VDQ_BHSI (BINOP, umax, 3)
BUILTIN_VDQ_BHSI (BINOP, umin, 3)
BUILTIN_VDQF (BINOP, smax_nan, 3)
BUILTIN_VDQF (BINOP, smin_nan, 3)
/* Implemented by <frint_pattern><mode>2. */
BUILTIN_VDQF (UNOP, btrunc, 2)

94
gcc/config/aarch64/aarch64-simd.md
View file

@ -940,12 +940,12 @@
)
;; Max/Min operations.
(define_insn "<maxmin><mode>3"
(define_insn "<su><maxmin><mode>3"
[(set (match_operand:VQ_S 0 "register_operand" "=w")
(MAXMIN:VQ_S (match_operand:VQ_S 1 "register_operand" "w")
(match_operand:VQ_S 2 "register_operand" "w")))]
"TARGET_SIMD"
"<maxmin>\t%0.<Vtype>, %1.<Vtype>, %2.<Vtype>"
"<su><maxmin>\t%0.<Vtype>, %1.<Vtype>, %2.<Vtype>"
[(set_attr "simd_type" "simd_minmax")
(set_attr "simd_mode" "<MODE>")]
)
@ -1417,44 +1417,23 @@
;; only introduces MIN_EXPR/MAX_EXPR in fast math mode or when not honouring
;; NaNs.
(define_insn "smax<mode>3"
(define_insn "<su><maxmin><mode>3"
[(set (match_operand:VDQF 0 "register_operand" "=w")
(smax:VDQF (match_operand:VDQF 1 "register_operand" "w")
(FMAXMIN:VDQF (match_operand:VDQF 1 "register_operand" "w")
(match_operand:VDQF 2 "register_operand" "w")))]
"TARGET_SIMD"
"fmaxnm\\t%0.<Vtype>, %1.<Vtype>, %2.<Vtype>"
"f<maxmin>nm\\t%0.<Vtype>, %1.<Vtype>, %2.<Vtype>"
[(set_attr "simd_type" "simd_fminmax")
(set_attr "simd_mode" "<MODE>")]
)
(define_insn "smin<mode>3"
(define_insn "<maxmin_uns><mode>3"
[(set (match_operand:VDQF 0 "register_operand" "=w")
(smin:VDQF (match_operand:VDQF 1 "register_operand" "w")
(match_operand:VDQF 2 "register_operand" "w")))]
(unspec:VDQF [(match_operand:VDQF 1 "register_operand" "w")
(match_operand:VDQF 2 "register_operand" "w")]
FMAXMIN_UNS))]
"TARGET_SIMD"
"fminnm\\t%0.<Vtype>, %1.<Vtype>, %2.<Vtype>"
[(set_attr "simd_type" "simd_fminmax")
(set_attr "simd_mode" "<MODE>")]
)
;; FP 'across lanes' max and min ops.
(define_insn "reduc_s<fmaxminv>_v4sf"
[(set (match_operand:V4SF 0 "register_operand" "=w")
(unspec:V4SF [(match_operand:V4SF 1 "register_operand" "w")]
FMAXMINV))]
"TARGET_SIMD"
"f<fmaxminv>nmv\\t%s0, %1.4s";
[(set_attr "simd_type" "simd_fminmaxv")
(set_attr "simd_mode" "V4SF")]
)
(define_insn "reduc_s<fmaxminv>_<mode>"
[(set (match_operand:V2F 0 "register_operand" "=w")
(unspec:V2F [(match_operand:V2F 1 "register_operand" "w")]
FMAXMINV))]
"TARGET_SIMD"
"f<fmaxminv>nmp\\t%0.<Vtype>, %1.<Vtype>, %1.<Vtype>";
"<maxmin_uns_op>\\t%0.<Vtype>, %1.<Vtype>, %2.<Vtype>"
[(set_attr "simd_type" "simd_fminmax")
(set_attr "simd_mode" "<MODE>")]
)
@ -1609,26 +1588,58 @@
""
)
(define_insn "reduc_<maxminv>_<mode>"
;; 'across lanes' max and min ops.
(define_insn "reduc_<maxmin_uns>_<mode>"
[(set (match_operand:VDQV 0 "register_operand" "=w")
(unspec:VDQV [(match_operand:VDQV 1 "register_operand" "w")]
MAXMINV))]
"TARGET_SIMD"
"<maxminv>v\\t%<Vetype>0, %1.<Vtype>"
"<maxmin_uns_op>v\\t%<Vetype>0, %1.<Vtype>"
[(set_attr "simd_type" "simd_minmaxv")
(set_attr "simd_mode" "<MODE>")]
)
(define_insn "reduc_<maxminv>_v2si"
(define_insn "reduc_<maxmin_uns>_v2di"
[(set (match_operand:V2DI 0 "register_operand" "=w")
(unspec:V2DI [(match_operand:V2DI 1 "register_operand" "w")]
MAXMINV))]
"TARGET_SIMD"
"<maxmin_uns_op>p\\t%d0, %1.2d"
[(set_attr "simd_type" "simd_minmaxv")
(set_attr "simd_mode" "V2DI")]
)
(define_insn "reduc_<maxmin_uns>_v2si"
[(set (match_operand:V2SI 0 "register_operand" "=w")
(unspec:V2SI [(match_operand:V2SI 1 "register_operand" "w")]
MAXMINV))]
"TARGET_SIMD"
"<maxminv>p\\t%0.2s, %1.2s, %1.2s"
[(set_attr "simd_type" "simd_minmax")
"<maxmin_uns_op>p\\t%0.2s, %1.2s, %1.2s"
[(set_attr "simd_type" "simd_minmaxv")
(set_attr "simd_mode" "V2SI")]
)
(define_insn "reduc_<maxmin_uns>_<mode>"
[(set (match_operand:V2F 0 "register_operand" "=w")
(unspec:V2F [(match_operand:V2F 1 "register_operand" "w")]
FMAXMINV))]
"TARGET_SIMD"
"<maxmin_uns_op>p\\t%<Vetype>0, %1.<Vtype>"
[(set_attr "simd_type" "simd_fminmaxv")
(set_attr "simd_mode" "<MODE>")]
)
(define_insn "reduc_<maxmin_uns>_v4sf"
[(set (match_operand:V4SF 0 "register_operand" "=w")
(unspec:V4SF [(match_operand:V4SF 1 "register_operand" "w")]
FMAXMINV))]
"TARGET_SIMD"
"<maxmin_uns_op>v\\t%s0, %1.4s"
[(set_attr "simd_type" "simd_fminmaxv")
(set_attr "simd_mode" "V4SF")]
)
;; aarch64_simd_bsl may compile to any of bsl/bif/bit depending on register
;; allocation.
;; Operand 1 is the mask, operands 2 and 3 are the bitfields from which
@ -3442,17 +3453,6 @@
(set_attr "simd_mode" "DI")]
)
(define_insn "aarch64_<fmaxmin><mode>"
[(set (match_operand:VDQF 0 "register_operand" "=w")
(unspec:VDQF [(match_operand:VDQF 1 "register_operand" "w")
(match_operand:VDQF 2 "register_operand" "w")]
FMAXMIN))]
"TARGET_SIMD"
"<fmaxmin>\t%0.<Vtype>, %1.<Vtype>, %2.<Vtype>"
[(set_attr "simd_type" "simd_fminmax")
(set_attr "simd_mode" "<MODE>")]
)
;; sqrt
(define_insn "sqrt<mode>2"

12
gcc/config/aarch64/arm_neon.h
View file

@ -21544,7 +21544,7 @@ vld4q_f64 (const float64_t * __a)
__extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
vmax_f32 (float32x2_t __a, float32x2_t __b)
{
return __builtin_aarch64_fmaxv2sf (__a, __b);
return __builtin_aarch64_smax_nanv2sf (__a, __b);
}
__extension__ static __inline int8x8_t __attribute__ ((__always_inline__))
@ -21589,13 +21589,13 @@ vmax_u32 (uint32x2_t __a, uint32x2_t __b)
__extension__ static __inline float32x4_t __attribute__ ((__always_inline__))
vmaxq_f32 (float32x4_t __a, float32x4_t __b)
{
return __builtin_aarch64_fmaxv4sf (__a, __b);
return __builtin_aarch64_smax_nanv4sf (__a, __b);
}
__extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
vmaxq_f64 (float64x2_t __a, float64x2_t __b)
{
return __builtin_aarch64_fmaxv2df (__a, __b);
return __builtin_aarch64_smax_nanv2df (__a, __b);
}
__extension__ static __inline int8x16_t __attribute__ ((__always_inline__))
@ -21642,7 +21642,7 @@ vmaxq_u32 (uint32x4_t __a, uint32x4_t __b)
__extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
vmin_f32 (float32x2_t __a, float32x2_t __b)
{
return __builtin_aarch64_fminv2sf (__a, __b);
return __builtin_aarch64_smin_nanv2sf (__a, __b);
}
__extension__ static __inline int8x8_t __attribute__ ((__always_inline__))
@ -21687,13 +21687,13 @@ vmin_u32 (uint32x2_t __a, uint32x2_t __b)
__extension__ static __inline float32x4_t __attribute__ ((__always_inline__))
vminq_f32 (float32x4_t __a, float32x4_t __b)
{
return __builtin_aarch64_fminv4sf (__a, __b);
return __builtin_aarch64_smin_nanv4sf (__a, __b);
}
__extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
vminq_f64 (float64x2_t __a, float64x2_t __b)
{
return __builtin_aarch64_fminv2df (__a, __b);
return __builtin_aarch64_smin_nanv2df (__a, __b);
}
__extension__ static __inline int8x16_t __attribute__ ((__always_inline__))

54
gcc/config/aarch64/iterators.md
View file

@ -163,7 +163,11 @@
[
UNSPEC_ASHIFT_SIGNED ; Used in aarch-simd.md.
UNSPEC_ASHIFT_UNSIGNED ; Used in aarch64-simd.md.
UNSPEC_FMAX ; Used in aarch64-simd.md.
UNSPEC_FMAXNMV ; Used in aarch64-simd.md.
UNSPEC_FMAXV ; Used in aarch64-simd.md.
UNSPEC_FMIN ; Used in aarch64-simd.md.
UNSPEC_FMINNMV ; Used in aarch64-simd.md.
UNSPEC_FMINV ; Used in aarch64-simd.md.
UNSPEC_FADDV ; Used in aarch64-simd.md.
UNSPEC_ADDV ; Used in aarch64-simd.md.
@ -223,8 +227,6 @@
UNSPEC_SSHLL ; Used in aarch64-simd.md.
UNSPEC_USHLL ; Used in aarch64-simd.md.
UNSPEC_ADDP ; Used in aarch64-simd.md.
UNSPEC_FMAX ; Used in aarch64-simd.md.
UNSPEC_FMIN ; Used in aarch64-simd.md.
UNSPEC_TBL ; Used in vector permute patterns.
UNSPEC_CONCAT ; Used in vector permute patterns.
UNSPEC_ZIP1 ; Used in vector permute patterns.
@ -535,6 +537,8 @@
;; Code iterator for variants of vector max and min.
(define_code_iterator MAXMIN [smax smin umax umin])
(define_code_iterator FMAXMIN [smax smin])
;; Code iterator for variants of vector max and min.
(define_code_iterator ADDSUB [plus minus])
@ -643,7 +647,9 @@
(define_code_attr su [(sign_extend "s") (zero_extend "u")
(sign_extract "s") (zero_extract "u")
(fix "s") (unsigned_fix "u")
(div "s") (udiv "u")])
(div "s") (udiv "u")
(smax "s") (umax "u")
(smin "s") (umin "u")])
;; Emit cbz/cbnz depending on comparison type.
(define_code_attr cbz [(eq "cbz") (ne "cbnz") (lt "cbnz") (ge "cbz")])
@ -652,10 +658,10 @@
(define_code_attr tbz [(eq "tbz") (ne "tbnz") (lt "tbnz") (ge "tbz")])
;; Max/min attributes.
(define_code_attr maxmin [(smax "smax")
(smin "smin")
(umax "umax")
(umin "umin")])
(define_code_attr maxmin [(smax "max")
(smin "min")
(umax "max")
(umin "min")])
;; MLA/MLS attributes.
(define_code_attr as [(ss_plus "a") (ss_minus "s")])
@ -677,7 +683,8 @@
(define_int_iterator MAXMINV [UNSPEC_UMAXV UNSPEC_UMINV
UNSPEC_SMAXV UNSPEC_SMINV])
(define_int_iterator FMAXMINV [UNSPEC_FMAXV UNSPEC_FMINV])
(define_int_iterator FMAXMINV [UNSPEC_FMAXV UNSPEC_FMINV
UNSPEC_FMAXNMV UNSPEC_FMINNMV])
(define_int_iterator HADDSUB [UNSPEC_SHADD UNSPEC_UHADD
UNSPEC_SRHADD UNSPEC_URHADD
@ -691,7 +698,7 @@
(define_int_iterator ADDSUBHN2 [UNSPEC_ADDHN2 UNSPEC_RADDHN2
UNSPEC_SUBHN2 UNSPEC_RSUBHN2])
(define_int_iterator FMAXMIN [UNSPEC_FMAX UNSPEC_FMIN])
(define_int_iterator FMAXMIN_UNS [UNSPEC_FMAX UNSPEC_FMIN])
(define_int_iterator VQDMULH [UNSPEC_SQDMULH UNSPEC_SQRDMULH])
@ -738,16 +745,27 @@
;; -------------------------------------------------------------------
;; Int Iterators Attributes.
;; -------------------------------------------------------------------
(define_int_attr maxminv [(UNSPEC_UMAXV "umax")
(UNSPEC_UMINV "umin")
(UNSPEC_SMAXV "smax")
(UNSPEC_SMINV "smin")])
(define_int_attr maxmin_uns [(UNSPEC_UMAXV "umax")
(UNSPEC_UMINV "umin")
(UNSPEC_SMAXV "smax")
(UNSPEC_SMINV "smin")
(UNSPEC_FMAX "smax_nan")
(UNSPEC_FMAXNMV "smax")
(UNSPEC_FMAXV "smax_nan")
(UNSPEC_FMIN "smin_nan")
(UNSPEC_FMINNMV "smin")
(UNSPEC_FMINV "smin_nan")])
(define_int_attr fmaxminv [(UNSPEC_FMAXV "max")
(UNSPEC_FMINV "min")])
(define_int_attr fmaxmin [(UNSPEC_FMAX "fmax")
(UNSPEC_FMIN "fmin")])
(define_int_attr maxmin_uns_op [(UNSPEC_UMAXV "umax")
(UNSPEC_UMINV "umin")
(UNSPEC_SMAXV "smax")
(UNSPEC_SMINV "smin")
(UNSPEC_FMAX "fmax")
(UNSPEC_FMAXNMV "fmaxnm")
(UNSPEC_FMAXV "fmax")
(UNSPEC_FMIN "fmin")
(UNSPEC_FMINNMV "fminnm")
(UNSPEC_FMINV "fmin")])
(define_int_attr sur [(UNSPEC_SHADD "s") (UNSPEC_UHADD "u")
(UNSPEC_SRHADD "sr") (UNSPEC_URHADD "ur")