From: Palmer Dabbelt <palmer@dabbelt.com>
To: rzinsly@ventanamicro.com
Cc: gcc-patches@gcc.gnu.org, jlaw@ventanamicro.com, rzinsly@ventanamicro.com
Subject: Re: [PATCH] RISC-V: Produce better code with complex constants [PR95632] [PR106602]
Date: Thu, 08 Dec 2022 09:53:01 -0800 (PST) [thread overview]
Message-ID: <mhng-e3367d70-4b53-4aff-bbc8-09bea1e0296a@palmer-ri-x1c9a> (raw)
In-Reply-To: <20221207205517.526182-1-rzinsly@ventanamicro.com>
On Wed, 07 Dec 2022 12:55:17 PST (-0800), rzinsly@ventanamicro.com wrote:
> Due to RISC-V limitations on operations with big constants combine
> is failing to match such operations and is not being able to
> produce optimal code as it keeps splitting them. By pretending we
> can do those operations we can get more opportunities for
> simplification of surrounding instructions.
I saw Jeff's comments. This is always the kind of thing that worries
me: we're essentially lying to the optimizer in order to trick it into
generating better code, which might just make it generate worse code.
It's always easy to see a small example that improves, but those could
be wiped out by secondary effects in real code. So I'd usually want to
have some benchmarking for a patch like this.
That said, if this is just the standard way of doing things then maybe
it's just fine?
> 2022-12-06 Raphael Moreira Zinsly <rzinsly@ventanamicro.com>
> Jeff Law <jlaw@ventanamicro.com>
>
> gcc/Changelog:
> PR target/95632
> PR target/106602
> * config/riscv/riscv.md: New pattern to simulate complex
> const_int loads.
>
> gcc/testsuite/ChangeLog:
> * gcc.target/riscv/pr95632.c: New test.
> * gcc.target/riscv/pr106602.c: Likewise.
> ---
> gcc/config/riscv/riscv.md | 16 ++++++++++++++++
> gcc/testsuite/gcc.target/riscv/pr106602.c | 14 ++++++++++++++
> gcc/testsuite/gcc.target/riscv/pr95632.c | 15 +++++++++++++++
> 3 files changed, 45 insertions(+)
> create mode 100644 gcc/testsuite/gcc.target/riscv/pr106602.c
> create mode 100644 gcc/testsuite/gcc.target/riscv/pr95632.c
>
> diff --git a/gcc/config/riscv/riscv.md b/gcc/config/riscv/riscv.md
> index df57e2b0b4a..0a9b5ec22b0 100644
> --- a/gcc/config/riscv/riscv.md
> +++ b/gcc/config/riscv/riscv.md
> @@ -1667,6 +1667,22 @@
> MAX_MACHINE_MODE, &operands[3], TRUE);
> })
>
> +;; Pretend to have the ability to load complex const_int in order to get
> +;; better code generation around them.
> +(define_insn_and_split ""
> + [(set (match_operand:GPR 0 "register_operand" "=r")
> + (match_operand:GPR 1 "splittable_const_int_operand" "i"))]
> + "cse_not_expected"
> + "#"
> + "&& 1"
> + [(const_int 0)]
> +
> +{
> + riscv_move_integer (operands[0], operands[0], INTVAL (operands[1]),
> + <GPR:MODE>mode, TRUE);
> + DONE;
> +})
There's some comments from Jakub on this, I don't see any additional
issues with the code (aside from the "does it help" stuff from above).
> +
> ;; 64-bit integer moves
>
> (define_expand "movdi"
> diff --git a/gcc/testsuite/gcc.target/riscv/pr106602.c b/gcc/testsuite/gcc.target/riscv/pr106602.c
> new file mode 100644
> index 00000000000..83b70877012
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/riscv/pr106602.c
> @@ -0,0 +1,14 @@
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -march=rv64gc" } */
There's a DG hook to limit this to 64-bit targets, that way it'll run
with whatever target is being tested.
> +
> +unsigned long
> +foo2 (unsigned long a)
> +{
> + return (unsigned long)(unsigned int) a << 6;
> +}
> +
> +/* { dg-final { scan-assembler-times "slli\t" 1 } } */
> +/* { dg-final { scan-assembler-times "srli\t" 1 } } */
> +/* { dg-final { scan-assembler-not "\tli\t" } } */
> +/* { dg-final { scan-assembler-not "addi\t" } } */
> +/* { dg-final { scan-assembler-not "and\t" } } */
> diff --git a/gcc/testsuite/gcc.target/riscv/pr95632.c b/gcc/testsuite/gcc.target/riscv/pr95632.c
> new file mode 100644
> index 00000000000..bd316ab1d7b
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/riscv/pr95632.c
> @@ -0,0 +1,15 @@
> +/* { dg-do compile } */
> +/* { dg-options "-O2 -march=rv32imafc -mabi=ilp32f" } */
Is there a reason to make this rv32-only? Unless I'm missing something
this should generate pretty much the same code for rv64.
> +
> +unsigned short
> +foo (unsigned short crc)
> +{
> + crc ^= 0x4002;
> + crc >>= 1;
> + crc |= 0x8000;
> +
> + return crc;
> +}
> +
> +/* { dg-final { scan-assembler-times "srli\t" 1 } } */
> +/* { dg-final { scan-assembler-not "slli\t" } } */
next prev parent reply other threads:[~2022-12-08 17:53 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-12-07 20:55 Raphael Moreira Zinsly
2022-12-07 21:13 ` Jeff Law
2022-12-07 21:30 ` Jakub Jelinek
2022-12-08 17:39 ` Palmer Dabbelt
2022-12-08 17:53 ` Palmer Dabbelt [this message]
2022-12-08 18:15 ` Jeff Law
2022-12-08 20:21 ` Palmer Dabbelt
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=mhng-e3367d70-4b53-4aff-bbc8-09bea1e0296a@palmer-ri-x1c9a \
--to=palmer@dabbelt.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=jlaw@ventanamicro.com \
--cc=rzinsly@ventanamicro.com \
/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).