From: "Andre Vieira (lists)" <andre.simoesdiasvieira@arm.com>
To: Kyrylo Tkachov <Kyrylo.Tkachov@arm.com>,
"gcc-patches@gcc.gnu.org" <gcc-patches@gcc.gnu.org>
Cc: Richard Sandiford <Richard.Sandiford@arm.com>,
Richard Earnshaw <Richard.Earnshaw@arm.com>,
Richard Biener <rguenther@suse.de>
Subject: Re: [PATCH 2/3] arm: Remove unnecessary zero-extending of MVE predicates before use [PR 107674]
Date: Fri, 27 Jan 2023 09:58:03 +0000 [thread overview]
Message-ID: <433b8286-f54a-1a4a-e194-4ffbe0851a74@arm.com> (raw)
In-Reply-To: <PAXPR08MB692609F668931DE1C568A52693CF9@PAXPR08MB6926.eurprd08.prod.outlook.com>
On 26/01/2023 15:06, Kyrylo Tkachov wrote:
> Hi Andre,
>
>> -----Original Message-----
>> From: Andre Vieira (lists) <andre.simoesdiasvieira@arm.com>
>> Sent: Tuesday, January 24, 2023 1:54 PM
>> To: gcc-patches@gcc.gnu.org
>> Cc: Richard Sandiford <Richard.Sandiford@arm.com>; Richard Earnshaw
>> <Richard.Earnshaw@arm.com>; Richard Biener <rguenther@suse.de>;
>> Kyrylo Tkachov <Kyrylo.Tkachov@arm.com>
>> Subject: [PATCH 2/3] arm: Remove unnecessary zero-extending of MVE
>> predicates before use [PR 107674]
>>
>> Hi,
>>
>> This patch teaches GCC that zero-extending a MVE predicate from 16-bits
>> to 32-bits and then only using 16-bits is a no-op.
>> It does so in two steps:
>> - it lets gcc know that it can access any MVE predicate mode using any
>> other MVE predicate mode without needing to copy it, using the
>> TARGET_MODES_TIEABLE_P hook,
>> - it teaches simplify_subreg to optimize a subreg with a vector
>> outermode, by replacing this outermode with a same-sized integer mode
>> and trying the avalailable optimizations, then if successful it
>> surrounds the result with a subreg casting it back to the original
>> vector outermode.
>>
>> This removes the unnecessary zero-extending shown on PR 107674 (though
>> it's a sign-extend there), that was introduced in gcc 11.
>>
>> Bootstrapped on aarch64-none-linux-gnu and regression tested on
>> arm-none-eabi and armeb-none-eabi for armv8.1-m.main+mve.fp.
>>
>> OK for trunk?
>>
>> gcc/ChangeLog:
>>
>> PR target/107674
>> * conig/arm/arm.cc (arm_hard_regno_mode_ok): Use new MACRO.
>> (arm_modes_tieable_p): Make MVE predicate modes tieable.
>> * config/arm/arm.h (VALID_MVE_PRED_MODE): New define.
>> * simplify-rtx.cc (simplify_context::simplify_subreg): Teach
>> simplify_subreg to simplify subregs where the outermode is not
>> scalar.
>
> The arm changes look ok to me. We'll want a midend maintainer to have a look at simplify-rtx.cc
>
>>
>> gcc/testsuite/ChangeLog:
>>
>> * gcc.target/arm/mve/mve_vpt.c: Change to remove unecessary
>> zero-extend.
>
> diff --git a/gcc/testsuite/gcc.target/arm/mve/mve_vpt.c b/gcc/testsuite/gcc.target/arm/mve/mve_vpt.c
> index 26a565b79dd1348e361b3aa23a1d6e6d13bffce8..8e562a9f065eff157f63ebd5acf9af0a2155b5c5 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/mve_vpt.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/mve_vpt.c
> @@ -16,9 +16,6 @@ void test0 (uint8_t *a, uint8_t *b, uint8_t *c)
> ** vldrb.8 q2, \[r0\]
> ** vldrb.8 q1, \[r1\]
> ** vcmp.i8 eq, q2, q1
> -** vmrs r3, p0 @ movhi
> -** uxth r3, r3
> -** vmsr p0, r3 @ movhi
> ** vpst
> ** vaddt.i8 q3, q2, q1
> ** vpst
>
> Ah I see, that's the testcase from patch 1/3 that I criticized :)
> Maybe if we just scan for absence of an uxth, vmrs and vmsr it will be more robust?
> Thanks,
> Kyrill
I could, but I would rather not. I have a patch series waiting for GCC
14 that does further improvements to this (and other VPST codegen)
sequences and if I do scan for 'absence' of an instruction I have to
break them up into single tests each. Also it wouldn't then fail if we
start spilling the predicate directly to memory for instance. Like I
mentioned in the previous patch, the sequence is unlikely to be able to
change through scheduling (other than maybe the reordering of the loads
through some bad luck, but I could make it robust to that).
next prev parent reply other threads:[~2023-01-27 9:58 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-24 13:31 [PATCH 0/3] arm: Fix regressions around MVE predicate codegen Andre Vieira (lists)
2023-01-24 13:40 ` [PATCH 1/3] arm: Fix sign of MVE predicate mve_pred16_t [PR 107674] Andre Vieira (lists)
2023-01-24 13:48 ` Andre Vieira (lists)
2023-01-26 15:02 ` Kyrylo Tkachov
2023-01-26 15:03 ` Kyrylo Tkachov
2023-01-27 9:54 ` Andre Vieira (lists)
2023-01-27 9:56 ` Kyrylo Tkachov
2023-01-30 16:38 ` Andre Vieira (lists)
2023-01-30 16:40 ` Kyrylo Tkachov
2023-01-24 13:54 ` [PATCH 2/3] arm: Remove unnecessary zero-extending of MVE predicates before use " Andre Vieira (lists)
2023-01-26 15:06 ` Kyrylo Tkachov
2023-01-27 9:58 ` Andre Vieira (lists) [this message]
2023-01-27 9:59 ` Kyrylo Tkachov
2023-01-30 16:41 ` Andre Vieira (lists)
2023-01-30 23:17 ` Richard Sandiford
2023-01-31 6:15 ` Richard Sandiford
2023-01-24 13:56 ` [PATCH 3/3] arm: Fix MVE predicates synthesis [PR 108443] Andre Vieira (lists)
2023-01-25 17:40 ` Andre Vieira (lists)
2023-01-31 9:53 ` Kyrylo Tkachov
2023-01-31 11:38 ` Andre Vieira (lists)
2023-01-31 16:44 ` Kyrylo Tkachov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=433b8286-f54a-1a4a-e194-4ffbe0851a74@arm.com \
--to=andre.simoesdiasvieira@arm.com \
--cc=Kyrylo.Tkachov@arm.com \
--cc=Richard.Earnshaw@arm.com \
--cc=Richard.Sandiford@arm.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=rguenther@suse.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).