* [Bug target/97147] GCC uses vhaddpd which is bad for latency
2020-09-21 12:56 [Bug target/97147] New: GCC uses vhaddpd which is bad for latency rguenth at gcc dot gnu.org
@ 2020-09-22 6:43 ` crazylht at gmail dot com
2021-08-17 3:03 ` crazylht at gmail dot com
` (6 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: crazylht at gmail dot com @ 2020-09-22 6:43 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97147
--- Comment #1 from Hongtao.liu <crazylht at gmail dot com> ---
(In reply to Richard Biener from comment #0)
> typedef double v2df __attribute__((vector_size(16)));
> double foo (v2df x, double y)
> {
> return x[0] + x[1] + y;
> }
> double bar (v2df x, double y)
> {
> return y + x[0] + x[1];
> }
>
> with -O2 -mavx2 -mtune=znver2 ends up generating
>
> foo:
> .LFB0:
> .cfi_startproc
> vhaddpd %xmm0, %xmm0, %xmm0
> vaddsd %xmm1, %xmm0, %xmm0
> ret
>
> bar:
> .LFB1:
> .cfi_startproc
> vmovapd %xmm0, %xmm2
> vaddsd %xmm1, %xmm0, %xmm0
> vunpckhpd %xmm2, %xmm2, %xmm2
> vaddsd %xmm2, %xmm0, %xmm0
> ret
>
> where bar should be a _lot_ better according to Agner which says
> that vhaddpd has a 4 uops, a latency of 7 cycles and a throughput of only
> one per two cycles while both vunpckhpd and vaddsd fare a lot better here.
> Coffee-lake isn't much better here.
>
> Maybe we want to disable the V2DF instructions for most tunings somehow?
Bar is also better on CLX and ICL.
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug target/97147] GCC uses vhaddpd which is bad for latency
2020-09-21 12:56 [Bug target/97147] New: GCC uses vhaddpd which is bad for latency rguenth at gcc dot gnu.org
2020-09-22 6:43 ` [Bug target/97147] " crazylht at gmail dot com
@ 2021-08-17 3:03 ` crazylht at gmail dot com
2021-08-17 6:58 ` rguenth at gcc dot gnu.org
` (5 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: crazylht at gmail dot com @ 2021-08-17 3:03 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97147
--- Comment #2 from Hongtao.liu <crazylht at gmail dot com> ---
Disable (define_insn "*sse3_haddv2df3_low" and (define_insn
"*sse3_hsubv2df3_low" seems to be ok.
But for foo1.
v2df foo1 (v2df x, v2df y)
{
v2df a;
a[0] = x[0] + x[1];
a[1] = y[0] + y[1];
return a;
}
it's
vhaddpd %xmm1, %xmm0, %xmm0
ret
vs
movapd xmm2, xmm0
unpckhpd xmm2, xmm2
addsd xmm0, xmm2
movapd xmm2, xmm1
unpckhpd xmm1, xmm1
addsd xmm1, xmm2
unpcklpd xmm0, xmm1
ret
and note w/o vhaddpd, codegen can be optimized to
movapd xmm2, xmm0
unpcklpd xmm2, xmm1
unpckhpd xmm0, xmm1
addpd xmm0, xmm2
ret
Guess maybe it's better done in gimple level?
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug target/97147] GCC uses vhaddpd which is bad for latency
2020-09-21 12:56 [Bug target/97147] New: GCC uses vhaddpd which is bad for latency rguenth at gcc dot gnu.org
2020-09-22 6:43 ` [Bug target/97147] " crazylht at gmail dot com
2021-08-17 3:03 ` crazylht at gmail dot com
@ 2021-08-17 6:58 ` rguenth at gcc dot gnu.org
2021-08-17 7:17 ` crazylht at gmail dot com
` (4 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: rguenth at gcc dot gnu.org @ 2021-08-17 6:58 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97147
--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Hongtao.liu from comment #2)
> Disable (define_insn "*sse3_haddv2df3_low" and (define_insn
> "*sse3_hsubv2df3_low" seems to be ok.
> But for foo1.
>
> v2df foo1 (v2df x, v2df y)
> {
> v2df a;
> a[0] = x[0] + x[1];
> a[1] = y[0] + y[1];
> return a;
> }
>
> it's
>
> vhaddpd %xmm1, %xmm0, %xmm0
> ret
>
> vs
>
> movapd xmm2, xmm0
> unpckhpd xmm2, xmm2
> addsd xmm0, xmm2
> movapd xmm2, xmm1
> unpckhpd xmm1, xmm1
> addsd xmm1, xmm2
> unpcklpd xmm0, xmm1
> ret
>
> and note w/o vhaddpd, codegen can be optimized to
>
> movapd xmm2, xmm0
> unpcklpd xmm2, xmm1
> unpckhpd xmm0, xmm1
> addpd xmm0, xmm2
> ret
>
> Guess maybe it's better done in gimple level?
On GIMPLE we see the testcase basically unchanged from what the source does:
_1 = BIT_FIELD_REF <x_7(D), 64, 0>;
_2 = BIT_FIELD_REF <x_7(D), 64, 64>;
_3 = _1 + _2;
a_9 = BIT_INSERT_EXPR <a_8(D), _3, 0>;
_4 = BIT_FIELD_REF <y_10(D), 64, 0>;
_5 = BIT_FIELD_REF <y_10(D), 64, 64>;
_6 = _4 + _5;
a_11 = BIT_INSERT_EXPR <a_9, _6, 64>;
return a_11;
vectorization fails in SLP discovery because we essentially see two lanes
operating on different vectors and we don't implement a way to shuffle
them together.
I think the full hadd define_insns are OK to keep, they really have special
arrangements (esp. the SFmode variants). But the reductions to scalar
(*_low) seem unnecessary and penaltizing (maybe we can guard use of those
with a -mtune-ctl?).
I also see we're missing patterns for h{add,sub}ps (not sure if we can manage
to get combine to synthesize it).
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug target/97147] GCC uses vhaddpd which is bad for latency
2020-09-21 12:56 [Bug target/97147] New: GCC uses vhaddpd which is bad for latency rguenth at gcc dot gnu.org
` (2 preceding siblings ...)
2021-08-17 6:58 ` rguenth at gcc dot gnu.org
@ 2021-08-17 7:17 ` crazylht at gmail dot com
2021-08-17 7:36 ` rguenther at suse dot de
` (3 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: crazylht at gmail dot com @ 2021-08-17 7:17 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97147
--- Comment #4 from Hongtao.liu <crazylht at gmail dot com> ---
(In reply to Richard Biener from comment #3)
> (In reply to Hongtao.liu from comment #2)
> > Disable (define_insn "*sse3_haddv2df3_low" and (define_insn
> > "*sse3_hsubv2df3_low" seems to be ok.
> > But for foo1.
> >
> > v2df foo1 (v2df x, v2df y)
> > {
> > v2df a;
> > a[0] = x[0] + x[1];
> > a[1] = y[0] + y[1];
> > return a;
> > }
> >
> > it's
> >
> > vhaddpd %xmm1, %xmm0, %xmm0
> > ret
> >
> > vs
> >
> > movapd xmm2, xmm0
> > unpckhpd xmm2, xmm2
> > addsd xmm0, xmm2
> > movapd xmm2, xmm1
> > unpckhpd xmm1, xmm1
> > addsd xmm1, xmm2
> > unpcklpd xmm0, xmm1
> > ret
> >
> > and note w/o vhaddpd, codegen can be optimized to
> >
> > movapd xmm2, xmm0
> > unpcklpd xmm2, xmm1
> > unpckhpd xmm0, xmm1
> > addpd xmm0, xmm2
> > ret
> >
> > Guess maybe it's better done in gimple level?
>
> On GIMPLE we see the testcase basically unchanged from what the source does:
>
> _1 = BIT_FIELD_REF <x_7(D), 64, 0>;
> _2 = BIT_FIELD_REF <x_7(D), 64, 64>;
> _3 = _1 + _2;
> a_9 = BIT_INSERT_EXPR <a_8(D), _3, 0>;
> _4 = BIT_FIELD_REF <y_10(D), 64, 0>;
> _5 = BIT_FIELD_REF <y_10(D), 64, 64>;
> _6 = _4 + _5;
> a_11 = BIT_INSERT_EXPR <a_9, _6, 64>;
> return a_11;
>
> vectorization fails in SLP discovery because we essentially see two lanes
> operating on different vectors and we don't implement a way to shuffle
> them together.
>
> I think the full hadd define_insns are OK to keep, they really have special
> arrangements (esp. the SFmode variants). But the reductions to scalar
> (*_low) seem unnecessary and penaltizing (maybe we can guard use of those
> with a -mtune-ctl?).
>
Yes, i'm add a tune to enabled v2df vector reduction and defaut disabled for
all processors.
> I also see we're missing patterns for h{add,sub}ps (not sure if we can manage
> to get combine to synthesize it).
You mean (define_insn "sse3_h<insn>v4sf3"?
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug target/97147] GCC uses vhaddpd which is bad for latency
2020-09-21 12:56 [Bug target/97147] New: GCC uses vhaddpd which is bad for latency rguenth at gcc dot gnu.org
` (3 preceding siblings ...)
2021-08-17 7:17 ` crazylht at gmail dot com
@ 2021-08-17 7:36 ` rguenther at suse dot de
2021-08-18 3:26 ` cvs-commit at gcc dot gnu.org
` (2 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: rguenther at suse dot de @ 2021-08-17 7:36 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97147
--- Comment #5 from rguenther at suse dot de <rguenther at suse dot de> ---
On Tue, 17 Aug 2021, crazylht at gmail dot com wrote:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97147
>
> --- Comment #4 from Hongtao.liu <crazylht at gmail dot com> ---
> (In reply to Richard Biener from comment #3)
> > I also see we're missing patterns for h{add,sub}ps (not sure if we can manage
> > to get combine to synthesize it).
>
> You mean (define_insn "sse3_h<insn>v4sf3"?
Yeah... stupid macros ;)
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug target/97147] GCC uses vhaddpd which is bad for latency
2020-09-21 12:56 [Bug target/97147] New: GCC uses vhaddpd which is bad for latency rguenth at gcc dot gnu.org
` (4 preceding siblings ...)
2021-08-17 7:36 ` rguenther at suse dot de
@ 2021-08-18 3:26 ` cvs-commit at gcc dot gnu.org
2021-08-18 3:27 ` crazylht at gmail dot com
2022-01-11 12:07 ` rguenth at gcc dot gnu.org
7 siblings, 0 replies; 9+ messages in thread
From: cvs-commit at gcc dot gnu.org @ 2021-08-18 3:26 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97147
--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by hongtao Liu <liuhongt@gcc.gnu.org>:
https://gcc.gnu.org/g:97d51c1764554fcef05fe94ee6445f5d2252bcff
commit r12-2981-g97d51c1764554fcef05fe94ee6445f5d2252bcff
Author: liuhongt <hongtao.liu@intel.com>
Date: Tue Aug 17 13:11:26 2021 +0800
Add x86 tune to enable v2df vector reduction by paddpd.
The tune is disabled by default.
gcc/ChangeLog:
PR target/97147
* config/i386/i386.h (TARGET_V2DF_REDUCTION_PREFER_HADDPD):
New macro.
* config/i386/sse.md (*sse3_haddv2df3_low): Add
TARGET_V2DF_REDUCTION_PREFER_HADDPD.
(*sse3_hsubv2df3_low): Ditto.
* config/i386/x86-tune.def
(X86_TUNE_V2DF_REDUCTION_PREFER_HADDPD): New tune.
gcc/testsuite/ChangeLog:
PR target/97147
* gcc.target/i386/pr54400.c: Adjust testcase.
* gcc.target/i386/pr94147.c: New test.
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug target/97147] GCC uses vhaddpd which is bad for latency
2020-09-21 12:56 [Bug target/97147] New: GCC uses vhaddpd which is bad for latency rguenth at gcc dot gnu.org
` (5 preceding siblings ...)
2021-08-18 3:26 ` cvs-commit at gcc dot gnu.org
@ 2021-08-18 3:27 ` crazylht at gmail dot com
2022-01-11 12:07 ` rguenth at gcc dot gnu.org
7 siblings, 0 replies; 9+ messages in thread
From: crazylht at gmail dot com @ 2021-08-18 3:27 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97147
--- Comment #7 from Hongtao.liu <crazylht at gmail dot com> ---
Fixed in GCC12.
^ permalink raw reply [flat|nested] 9+ messages in thread
* [Bug target/97147] GCC uses vhaddpd which is bad for latency
2020-09-21 12:56 [Bug target/97147] New: GCC uses vhaddpd which is bad for latency rguenth at gcc dot gnu.org
` (6 preceding siblings ...)
2021-08-18 3:27 ` crazylht at gmail dot com
@ 2022-01-11 12:07 ` rguenth at gcc dot gnu.org
7 siblings, 0 replies; 9+ messages in thread
From: rguenth at gcc dot gnu.org @ 2022-01-11 12:07 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97147
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |RESOLVED
Resolution|--- |FIXED
Known to work| |12.0
--- Comment #8 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.
^ permalink raw reply [flat|nested] 9+ messages in thread