[x86_64] Support pandn for V1TI mode (i.e. *andnotv1ti3).
Commit Message
This simple patch allows the i386 backend to generate pandn instructions
for V1TI mode. Currently, the testcase:
typedef unsigned __int128 v1ti __attribute__ ((__vector_size__ (16)));
v1ti andnot1(v1ti x, v1ti y) { return ~x & y; }
generates with -O2
pcmpeqd %xmm2, %xmm2
pxor %xmm2, %xmm0
pand %xmm1, %xmm0
ret
with this patch, we now generate:
pandn %xmm1, %xmm0
ret
It turns out that there are currently three (near) duplicates of the
logic for andn/pandn/vandn/vpandn in i386/sse.md: one for floating point
vectors (MODEF), one for integer vectors (VI) and a third for TFmode.
Rather than introduce a fourth copy, this patch introduces a new mode
iterator to share/reuse the TFmode define_insn to also handle V1TI.
This patch has been tested on x86_64-pc-linux-gnu with make bootstrap
and make -k check with no new failures. Ok for mainline?
2022-04-05 Roger Sayle <roger@nextmovesoftware.com>
gcc/ChangeLog
* config/i386/sse.md (ANDNOT_MODE): New mode iterator for TF and V1TI.
(*andnottf3): Replace with...
(*andnot<mode>3): New define_insn using ANDNOT_MODE.
gcc/testsuite/ChangeLog
* gcc.target/i386/sse2-v1ti-andnot.c: New test case.
Thanks in advance,
Roger
--
Comments
On Wed, Apr 6, 2022 at 5:56 AM Roger Sayle <roger@nextmovesoftware.com> wrote:
>
>
>
> This simple patch allows the i386 backend to generate pandn instructions
>
> for V1TI mode. Currently, the testcase:
>
>
>
> typedef unsigned __int128 v1ti __attribute__ ((__vector_size__ (16)));
>
> v1ti andnot1(v1ti x, v1ti y) { return ~x & y; }
>
>
>
> generates with -O2
>
>
>
> pcmpeqd %xmm2, %xmm2
>
> pxor %xmm2, %xmm0
>
> pand %xmm1, %xmm0
>
> ret
>
>
>
> with this patch, we now generate:
>
>
>
> pandn %xmm1, %xmm0
>
> ret
>
>
>
> It turns out that there are currently three (near) duplicates of the
>
> logic for andn/pandn/vandn/vpandn in i386/sse.md: one for floating point
>
> vectors (MODEF), one for integer vectors (VI) and a third for TFmode.
>
> Rather than introduce a fourth copy, this patch introduces a new mode
>
> iterator to share/reuse the TFmode define_insn to also handle V1TI.
>
>
>
> This patch has been tested on x86_64-pc-linux-gnu with make bootstrap
>
> and make -k check with no new failures. Ok for mainline?
>
Ok.
>
>
>
>
> 2022-04-05 Roger Sayle <roger@nextmovesoftware.com>
>
>
>
> gcc/ChangeLog
>
> * config/i386/sse.md (ANDNOT_MODE): New mode iterator for TF and V1TI.
>
> (*andnottf3): Replace with...
>
> (*andnot<mode>3): New define_insn using ANDNOT_MODE.
>
>
>
> gcc/testsuite/ChangeLog
>
> * gcc.target/i386/sse2-v1ti-andnot.c: New test case.
>
>
>
>
>
> Thanks in advance,
>
> Roger
>
> --
>
>
>
@@ -4923,11 +4923,14 @@
]
(const_string "<ssevecmode>")))])
-(define_insn "*andnottf3"
- [(set (match_operand:TF 0 "register_operand" "=x,x,v,v")
- (and:TF
- (not:TF (match_operand:TF 1 "register_operand" "0,x,v,v"))
- (match_operand:TF 2 "vector_operand" "xBm,xm,vm,v")))]
+;; Modes for andnot3 not covered by VI and MODEF.
+(define_mode_iterator ANDNOT_MODE [TF V1TI])
+
+(define_insn "*andnot<mode>3"
+ [(set (match_operand:ANDNOT_MODE 0 "register_operand" "=x,x,v,v")
+ (and:ANDNOT_MODE
+ (not:ANDNOT_MODE (match_operand:ANDNOT_MODE 1 "register_operand" "0,x,v,v"))
+ (match_operand:ANDNOT_MODE 2 "vector_operand" "xBm,xm,vm,v")))]
"TARGET_SSE"
{
char buf[128];
new file mode 100644
@@ -0,0 +1,11 @@
+/* { dg-do compile { target int128 } } */
+/* { dg-options "-O2 -msse2" } */
+
+typedef __int128 v1ti __attribute__ ((__vector_size__ (16)));
+
+v1ti andnot1(v1ti x, v1ti y) { return ~x & y; }
+v1ti andnot2(v1ti x, v1ti y) { return x & ~y; }
+
+/* { dg-final { scan-assembler-times "pandn" 2 } } */
+/* { dg-final { scan-assembler-not "pcmpeqd" } } */
+/* { dg-final { scan-assembler-not "pxor" } } */