[PATCH i386 AVX512] [48/n] Add shuffles (pd, 32x4, etc.).
Kirill Yukhin
kirill.yukhin@gmail.com
Tue Sep 23 11:22:00 GMT 2014
Hello,
Patch in the bottom extends more shuffle patterns.
Bootstrapped.
AVX-512* tests on top of patch-set all pass
under simulator.
Is it ok for trunk?
gcc/
* config/i386/i386.c
(ix86_expand_args_builtin): Handle CODE_FOR_sse2_shufpd,
CODE_FOR_sse2_sse2_shufpd_mask, CODE_FOR_sse2_avx512dq_shuf_f64x2_mask,
CODE_FOR_sse2_avx512dq_shuf_i64x2_mask,
CODE_FOR_sse2_avx512vl_shuf_i32x4_mask and
CODE_FOR_sse2_avx512vl_shuf_f32x4_mask.
* config/i386/sse.md
(define_expand "avx512dq_shuf_<shuffletype>64x2_mask"): New.
(define_insn
"<mask_codefor>avx512dq_shuf_<shuffletype>64x2_1<mask_name>"): Ditto.
(define_expand "avx512vl_shuf_<shuffletype>32x4_mask"): Ditto.
(define_insn
"<mask_codefor>avx512vl_shuf_<shuffletype>32x4_1<mask_name>"): Ditto.
(define_expand "avx512vl_pshufdv3_mask"): Ditto.
(define_insn "avx2_pshufd_1<mask_name>"): Add masking.
(define_expand "avx512vl_pshufd_mask"): New.
(define_insn "sse2_pshufd_1<mask_name>"): Add masking.
--
Thanks, K
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index af61628..a6c888a 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -34111,6 +34111,12 @@ ix86_expand_args_builtin (const struct builtin_description *d,
case CODE_FOR_avx512f_vinserti32x4_mask:
case CODE_FOR_avx512f_vextractf32x4_mask:
case CODE_FOR_avx512f_vextracti32x4_mask:
+ case CODE_FOR_sse2_shufpd:
+ case CODE_FOR_sse2_shufpd_mask:
+ case CODE_FOR_avx512dq_shuf_f64x2_mask:
+ case CODE_FOR_avx512dq_shuf_i64x2_mask:
+ case CODE_FOR_avx512vl_shuf_i32x4_mask:
+ case CODE_FOR_avx512vl_shuf_f32x4_mask:
error ("the last argument must be a 2-bit immediate");
return const0_rtx;
diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md
index 2dd79d0..f377a5e 100644
--- a/gcc/config/i386/sse.md
+++ b/gcc/config/i386/sse.md
@@ -11336,6 +11336,51 @@
(set_attr "prefix" "evex")
(set_attr "mode" "XI")])
+(define_expand "avx512dq_shuf_<shuffletype>64x2_mask"
+ [(match_operand:VI8F_256 0 "register_operand")
+ (match_operand:VI8F_256 1 "register_operand")
+ (match_operand:VI8F_256 2 "nonimmediate_operand")
+ (match_operand:SI 3 "const_0_to_3_operand")
+ (match_operand:VI8F_256 4 "register_operand")
+ (match_operand:QI 5 "register_operand")]
+ "TARGET_AVX512DQ"
+{
+ int mask = INTVAL (operands[3]);
+ emit_insn (gen_avx512dq_shuf_<shuffletype>64x2_1_mask
+ (operands[0], operands[1], operands[2],
+ GEN_INT (((mask >> 0) & 1) * 2 + 0),
+ GEN_INT (((mask >> 0) & 1) * 2 + 1),
+ GEN_INT (((mask >> 1) & 1) * 2 + 4),
+ GEN_INT (((mask >> 1) & 1) * 2 + 5),
+ operands[4], operands[5]));
+ DONE;
+})
+
+(define_insn "<mask_codefor>avx512dq_shuf_<shuffletype>64x2_1<mask_name>"
+ [(set (match_operand:VI8F_256 0 "register_operand" "=v")
+ (vec_select:VI8F_256
+ (vec_concat:<ssedoublemode>
+ (match_operand:VI8F_256 1 "register_operand" "v")
+ (match_operand:VI8F_256 2 "nonimmediate_operand" "vm"))
+ (parallel [(match_operand 3 "const_0_to_3_operand")
+ (match_operand 4 "const_0_to_3_operand")
+ (match_operand 5 "const_4_to_7_operand")
+ (match_operand 6 "const_4_to_7_operand")])))]
+ "TARGET_AVX512VL
+ && (INTVAL (operands[3]) == (INTVAL (operands[4]) - 1)
+ && INTVAL (operands[5]) == (INTVAL (operands[6]) - 1))"
+{
+ int mask;
+ mask = INTVAL (operands[3]) / 2;
+ mask |= (INTVAL (operands[5]) - 4) / 2 << 1;
+ operands[3] = GEN_INT (mask);
+ return "vshuf<shuffletype>64x2\t{%3, %2, %1, %0<mask_operand7>|%0<mask_operand7>, %1, %2, %3}";
+}
+ [(set_attr "type" "sselog")
+ (set_attr "length_immediate" "1")
+ (set_attr "prefix" "evex")
+ (set_attr "mode" "XI")])
+
(define_expand "avx512f_shuf_<shuffletype>64x2_mask"
[(match_operand:V8FI 0 "register_operand")
(match_operand:V8FI 1 "register_operand")
@@ -11394,6 +11439,64 @@
(set_attr "prefix" "evex")
(set_attr "mode" "<sseinsnmode>")])
+(define_expand "avx512vl_shuf_<shuffletype>32x4_mask"
+ [(match_operand:VI4F_256 0 "register_operand")
+ (match_operand:VI4F_256 1 "register_operand")
+ (match_operand:VI4F_256 2 "nonimmediate_operand")
+ (match_operand:SI 3 "const_0_to_3_operand")
+ (match_operand:VI4F_256 4 "register_operand")
+ (match_operand:QI 5 "register_operand")]
+ "TARGET_AVX512VL"
+{
+ int mask = INTVAL (operands[3]);
+ emit_insn (gen_avx512vl_shuf_<shuffletype>32x4_1_mask
+ (operands[0], operands[1], operands[2],
+ GEN_INT (((mask >> 0) & 1) * 4 + 0),
+ GEN_INT (((mask >> 0) & 1) * 4 + 1),
+ GEN_INT (((mask >> 0) & 1) * 4 + 2),
+ GEN_INT (((mask >> 0) & 1) * 4 + 3),
+ GEN_INT (((mask >> 1) & 1) * 4 + 8),
+ GEN_INT (((mask >> 1) & 1) * 4 + 9),
+ GEN_INT (((mask >> 1) & 1) * 4 + 10),
+ GEN_INT (((mask >> 1) & 1) * 4 + 11),
+ operands[4], operands[5]));
+ DONE;
+})
+
+(define_insn "<mask_codefor>avx512vl_shuf_<shuffletype>32x4_1<mask_name>"
+ [(set (match_operand:VI4F_256 0 "register_operand" "=v")
+ (vec_select:VI4F_256
+ (vec_concat:<ssedoublemode>
+ (match_operand:VI4F_256 1 "register_operand" "v")
+ (match_operand:VI4F_256 2 "nonimmediate_operand" "vm"))
+ (parallel [(match_operand 3 "const_0_to_7_operand")
+ (match_operand 4 "const_0_to_7_operand")
+ (match_operand 5 "const_0_to_7_operand")
+ (match_operand 6 "const_0_to_7_operand")
+ (match_operand 7 "const_8_to_15_operand")
+ (match_operand 8 "const_8_to_15_operand")
+ (match_operand 9 "const_8_to_15_operand")
+ (match_operand 10 "const_8_to_15_operand")])))]
+ "TARGET_AVX512VL
+ && (INTVAL (operands[3]) == (INTVAL (operands[4]) - 1)
+ && INTVAL (operands[3]) == (INTVAL (operands[5]) - 2)
+ && INTVAL (operands[3]) == (INTVAL (operands[6]) - 3)
+ && INTVAL (operands[7]) == (INTVAL (operands[8]) - 1)
+ && INTVAL (operands[7]) == (INTVAL (operands[9]) - 2)
+ && INTVAL (operands[7]) == (INTVAL (operands[10]) - 3))"
+{
+ int mask;
+ mask = INTVAL (operands[3]) / 4;
+ mask |= (INTVAL (operands[7]) - 8) / 4 << 1;
+ operands[3] = GEN_INT (mask);
+
+ return "vshuf<shuffletype>32x4\t{%3, %2, %1, %0<mask_operand11>|%0<mask_operand11>, %1, %2, %3}";
+}
+ [(set_attr "type" "sselog")
+ (set_attr "length_immediate" "1")
+ (set_attr "prefix" "evex")
+ (set_attr "mode" "<sseinsnmode>")])
+
(define_expand "avx512f_shuf_<shuffletype>32x4_mask"
[(match_operand:V16FI 0 "register_operand")
(match_operand:V16FI 1 "register_operand")
@@ -11554,6 +11657,28 @@
(set_attr "length_immediate" "1")
(set_attr "mode" "XI")])
+(define_expand "avx512vl_pshufdv3_mask"
+ [(match_operand:V8SI 0 "register_operand")
+ (match_operand:V8SI 1 "nonimmediate_operand")
+ (match_operand:SI 2 "const_0_to_255_operand")
+ (match_operand:V8SI 3 "register_operand")
+ (match_operand:QI 4 "register_operand")]
+ "TARGET_AVX512VL"
+{
+ int mask = INTVAL (operands[2]);
+ emit_insn (gen_avx2_pshufd_1_mask (operands[0], operands[1],
+ GEN_INT ((mask >> 0) & 3),
+ GEN_INT ((mask >> 2) & 3),
+ GEN_INT ((mask >> 4) & 3),
+ GEN_INT ((mask >> 6) & 3),
+ GEN_INT (((mask >> 0) & 3) + 4),
+ GEN_INT (((mask >> 2) & 3) + 4),
+ GEN_INT (((mask >> 4) & 3) + 4),
+ GEN_INT (((mask >> 6) & 3) + 4),
+ operands[3], operands[4]));
+ DONE;
+})
+
(define_expand "avx2_pshufdv3"
[(match_operand:V8SI 0 "register_operand")
(match_operand:V8SI 1 "nonimmediate_operand")
@@ -11573,10 +11698,10 @@
DONE;
})
-(define_insn "avx2_pshufd_1"
- [(set (match_operand:V8SI 0 "register_operand" "=x")
+(define_insn "avx2_pshufd_1<mask_name>"
+ [(set (match_operand:V8SI 0 "register_operand" "=v")
(vec_select:V8SI
- (match_operand:V8SI 1 "nonimmediate_operand" "xm")
+ (match_operand:V8SI 1 "nonimmediate_operand" "vm")
(parallel [(match_operand 2 "const_0_to_3_operand")
(match_operand 3 "const_0_to_3_operand")
(match_operand 4 "const_0_to_3_operand")
@@ -11589,7 +11714,8 @@
&& INTVAL (operands[2]) + 4 == INTVAL (operands[6])
&& INTVAL (operands[3]) + 4 == INTVAL (operands[7])
&& INTVAL (operands[4]) + 4 == INTVAL (operands[8])
- && INTVAL (operands[5]) + 4 == INTVAL (operands[9])"
+ && INTVAL (operands[5]) + 4 == INTVAL (operands[9])
+ && <mask_avx512vl_condition>"
{
int mask = 0;
mask |= INTVAL (operands[2]) << 0;
@@ -11598,13 +11724,31 @@
mask |= INTVAL (operands[5]) << 6;
operands[2] = GEN_INT (mask);
- return "vpshufd\t{%2, %1, %0|%0, %1, %2}";
+ return "vpshufd\t{%2, %1, %0<mask_operand10>|%0<mask_operand10>, %1, %2}";
}
[(set_attr "type" "sselog1")
- (set_attr "prefix" "vex")
+ (set_attr "prefix" "maybe_evex")
(set_attr "length_immediate" "1")
(set_attr "mode" "OI")])
+(define_expand "avx512vl_pshufd_mask"
+ [(match_operand:V4SI 0 "register_operand")
+ (match_operand:V4SI 1 "nonimmediate_operand")
+ (match_operand:SI 2 "const_0_to_255_operand")
+ (match_operand:V4SI 3 "register_operand")
+ (match_operand:QI 4 "register_operand")]
+ "TARGET_AVX512VL"
+{
+ int mask = INTVAL (operands[2]);
+ emit_insn (gen_sse2_pshufd_1_mask (operands[0], operands[1],
+ GEN_INT ((mask >> 0) & 3),
+ GEN_INT ((mask >> 2) & 3),
+ GEN_INT ((mask >> 4) & 3),
+ GEN_INT ((mask >> 6) & 3),
+ operands[3], operands[4]));
+ DONE;
+})
+
(define_expand "sse2_pshufd"
[(match_operand:V4SI 0 "register_operand")
(match_operand:V4SI 1 "nonimmediate_operand")
@@ -11620,15 +11764,15 @@
DONE;
})
-(define_insn "sse2_pshufd_1"
- [(set (match_operand:V4SI 0 "register_operand" "=x")
+(define_insn "sse2_pshufd_1<mask_name>"
+ [(set (match_operand:V4SI 0 "register_operand" "=v")
(vec_select:V4SI
- (match_operand:V4SI 1 "nonimmediate_operand" "xm")
+ (match_operand:V4SI 1 "nonimmediate_operand" "vm")
(parallel [(match_operand 2 "const_0_to_3_operand")
(match_operand 3 "const_0_to_3_operand")
(match_operand 4 "const_0_to_3_operand")
(match_operand 5 "const_0_to_3_operand")])))]
- "TARGET_SSE2"
+ "TARGET_SSE2 && <mask_avx512vl_condition>"
{
int mask = 0;
mask |= INTVAL (operands[2]) << 0;
@@ -11637,11 +11781,11 @@
mask |= INTVAL (operands[5]) << 6;
operands[2] = GEN_INT (mask);
- return "%vpshufd\t{%2, %1, %0|%0, %1, %2}";
+ return "%vpshufd\t{%2, %1, %0<mask_operand6>|%0<mask_operand6>, %1, %2}";
}
[(set_attr "type" "sselog1")
(set_attr "prefix_data16" "1")
- (set_attr "prefix" "maybe_vex")
+ (set_attr "prefix" "<mask_prefix2>")
(set_attr "length_immediate" "1")
(set_attr "mode" "TI")])
More information about the Gcc-patches
mailing list