From: "Arnd Bergmann" <arnd@kernel.org>
To: "Ingo Molnar" <mingo@kernel.org>
Cc: "H. Peter Anvin" <hpa@zytor.com>,
linux-kernel@vger.kernel.org,
"Ahmed S . Darwish" <darwi@linutronix.de>,
"Andrew Cooper" <andrew.cooper3@citrix.com>,
"Ard Biesheuvel" <ardb@kernel.org>,
"Borislav Petkov" <bp@alien8.de>,
"Dave Hansen" <dave.hansen@linux.intel.com>,
"John Ogness" <john.ogness@linutronix.de>,
"Linus Torvalds" <torvalds@linux-foundation.org>,
"Peter Zijlstra" <peterz@infradead.org>,
"Thomas Gleixner" <tglx@linutronix.de>
Subject: Re: [PATCH 13/15] x86/cpu: Make CONFIG_X86_CX8 unconditional
Date: Mon, 28 Apr 2025 13:21:11 +0200 [thread overview]
Message-ID: <6c363c6f-7152-4d09-96db-861eda759a35@app.fastmail.com> (raw)
In-Reply-To: <aA9HaA2u-tdVA2ET@gmail.com>
On Mon, Apr 28, 2025, at 11:16, Ingo Molnar wrote:
> * Arnd Bergmann <arnd@kernel.org> wrote:
>>
>> b) always build with -march=i586 and leave only the -mtune
>> flags; see if anyone cares enough to even benchmark
>> and pick one of the other options if they can show
>> a meaningful regression over -march=i686 -mtune=
>
> That's actually a good idea IMO. I looked at the code generation with
> current compilers and it turns out that M686 is *substantially* worse
> in code generation than M586, as apparently the extra CMOV instructions
> bloat up the generated code:
>
> text data bss dec hex filename
> 15427023 7601010 1744896 24772929 17a0141 vmlinux.M586
> 16578295 7598826 1744896 25922017 18b89e1 vmlinux.M686
>
> - +7.5% increase in text size (5.6% according to bloatometer),
> - +2% increase in instruction count,
> - while number of branches increases by +1.3%.
>
> But it's not about CMOV: I checked about a dozen functions that end up
> using CMOV, and the 'conditional' part of CMOV does seem to reduce
> branches for those functions by a minor degree and ends up reducing
> their size as well. So CMOV helps, a bit.
>
> The substantial code bloat comes from some other aspect of GCC's
> march=i686 flag ... I bet it's primarily inlining: there's a 0.7%
> reduction in number of calls done.
I had tried the same thing already, but saw a different result,
For me, the i686 output is 0.2% smaller than the i586 one (both
-mtune=generic), using gcc-14.2. or just 0.1% with clang-21,
which is roughly what I expected:
text data bss dec hex filename
7454055 4158218 1695744 13308017 cb1071 build/tmp/vmlinux-i586
7433427 4154146 1695744 13283317 caaff5 build/tmp/vmlinux-i686
7318514 4052573 1687552 13058639 c7424f build/tmp/vmlinux-i586-clang
7309938 4052573 1687552 13050063 c720cf build/tmp/vmlinux-i686-clang
I do see a larger difference compared to other -mtune= options, here is
the same config with "clang-21 -march=i586 -mtune=i686" instead of
"-march=i586 -mtune=generic":
7254510 4056669 1687552 12998731 c6584b build/tmp/vmlinux
There is a good chance that the -mtune= optimizations totally
dwarf cmov not just in code size difference but also actual
performance, the bit I'm unsure about is whether we still need
to worry about any core where this is not the case (I'm guessing
not but have no way to prove that).
Arnd
next prev parent reply other threads:[~2025-04-28 11:21 UTC|newest]
Thread overview: 70+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-04-25 8:41 [RFC PATCH 0/15] x86: Remove support for TSC-less and CX8-less CPUs Ingo Molnar
2025-04-25 8:41 ` [PATCH 01/15] x86/cpu: Remove M486/M486SX/ELAN support Ingo Molnar
2025-04-25 8:41 ` [PATCH 02/15] x86/cpu: Remove CONFIG_MWINCHIP3D/MWINCHIPC6 Ingo Molnar
2025-04-25 8:42 ` [PATCH 03/15] x86/cpu: Remove CPU_SUP_UMC_32 support Ingo Molnar
2025-04-25 8:42 ` [PATCH 04/15] x86/cpu: Remove TSC-less CONFIG_M586 support Ingo Molnar
2025-04-25 15:09 ` H. Peter Anvin
2025-04-25 8:42 ` [PATCH 05/15] x86/cpu, x86/platform, watchdog: Remove CONFIG_X86_RDC321X support Ingo Molnar
2025-04-25 8:42 ` [PATCH 06/15] x86/cpu: Remove the CONFIG_X86_INVD_BUG quirk Ingo Molnar
2025-04-25 8:42 ` [PATCH 07/15] x86/cpu, cpufreq: Remove AMD ELAN support Ingo Molnar
2025-04-25 8:42 ` [PATCH 08/15] x86/fpu: Remove MATH_EMULATION and related glue code Ingo Molnar
2025-04-25 8:42 ` [PATCH 09/15] x86/fpu: Remove the 'no387' boot option Ingo Molnar
2025-04-25 8:42 ` [PATCH 10/15] x86/fpu: Remove the math-emu/ FPU emulation library Ingo Molnar
2025-04-25 8:42 ` [PATCH 11/15] x86/cpu: Make CONFIG_X86_TSC unconditional Ingo Molnar
2025-04-25 8:42 ` [PATCH 12/15] x86: Remove !CONFIG_X86_TSC code Ingo Molnar
2025-04-25 8:42 ` [PATCH 13/15] x86/cpu: Make CONFIG_X86_CX8 unconditional Ingo Molnar
2025-04-25 12:10 ` Arnd Bergmann
2025-04-25 15:15 ` H. Peter Anvin
2025-04-25 21:54 ` Arnd Bergmann
2025-04-27 9:25 ` Ingo Molnar
2025-04-27 17:32 ` Arnd Bergmann
2025-04-27 21:26 ` H. Peter Anvin
2025-04-28 9:16 ` Ingo Molnar
2025-04-28 11:21 ` Arnd Bergmann [this message]
2025-04-29 10:22 ` Ingo Molnar
2025-04-29 12:10 ` Arnd Bergmann
2025-04-27 8:15 ` Ingo Molnar
2025-04-25 8:42 ` [PATCH 14/15] x86/percpu: Remove !CONFIG_X86_CX8 methods Ingo Molnar
2025-04-25 14:05 ` Uros Bizjak
2025-04-27 7:46 ` [PATCH -v2 " Ingo Molnar
2025-04-25 8:42 ` [PATCH 15/15] x86/atomics: " Ingo Molnar
2025-04-25 13:55 ` Arnd Bergmann
2025-04-25 14:10 ` Uros Bizjak
2025-04-25 15:00 ` Brian Gerst
2025-04-27 8:23 ` Ingo Molnar
2025-04-27 8:26 ` Ingo Molnar
2025-04-27 10:31 ` Uros Bizjak
2025-05-15 8:49 ` Ingo Molnar
2025-04-25 11:13 ` [RFC PATCH 0/15] x86: Remove support for TSC-less and CX8-less CPUs Arnd Bergmann
2025-04-26 8:26 ` Pavel Machek
2025-05-05 8:53 ` Maciej W. Rozycki
2025-05-05 12:48 ` H. Peter Anvin
2025-05-05 13:04 ` Maciej W. Rozycki
2025-05-05 19:57 ` H. Peter Anvin
2025-05-05 20:54 ` Borislav Petkov
2025-05-06 13:51 ` Maciej W. Rozycki
2025-05-06 14:16 ` Borislav Petkov
2025-05-08 14:51 ` Maciej W. Rozycki
2025-05-08 20:11 ` Borislav Petkov
2025-05-12 12:55 ` Maciej W. Rozycki
2025-05-12 13:48 ` Borislav Petkov
2025-05-12 17:29 ` Maciej W. Rozycki
2025-05-13 2:00 ` Linus Torvalds
2025-05-13 3:48 ` H. Peter Anvin
2025-05-13 5:43 ` John Paul Adrian Glaubitz
2025-05-13 21:55 ` Maciej W. Rozycki
2025-05-13 22:02 ` Linus Torvalds
2025-05-13 22:06 ` H. Peter Anvin
2025-05-15 16:32 ` Maciej W. Rozycki
2025-05-06 13:48 ` Maciej W. Rozycki
2025-05-06 13:54 ` John Paul Adrian Glaubitz
2025-05-05 15:59 ` Linus Torvalds
2025-05-06 13:53 ` Maciej W. Rozycki
2025-05-06 16:44 ` H. Peter Anvin
2025-05-06 17:11 ` Linus Torvalds
2025-05-06 17:51 ` H. Peter Anvin
2025-05-08 14:54 ` Maciej W. Rozycki
2025-05-08 14:53 ` Maciej W. Rozycki
2025-05-08 14:51 ` Maciej W. Rozycki
2025-05-15 8:56 [PATCH -v2 " Ingo Molnar
2025-05-15 8:57 ` [PATCH 13/15] x86/cpu: Make CONFIG_X86_CX8 unconditional Ingo Molnar
2025-12-14 8:46 [PATCH -v3 0/15] x86: Remove support for TSC-less and CX8-less CPUs Ingo Molnar
2025-12-14 8:47 ` [PATCH 13/15] x86/cpu: Make CONFIG_X86_CX8 unconditional Ingo Molnar
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=6c363c6f-7152-4d09-96db-861eda759a35@app.fastmail.com \
--to=arnd@kernel.org \
--cc=andrew.cooper3@citrix.com \
--cc=ardb@kernel.org \
--cc=bp@alien8.de \
--cc=darwi@linutronix.de \
--cc=dave.hansen@linux.intel.com \
--cc=hpa@zytor.com \
--cc=john.ogness@linutronix.de \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=peterz@infradead.org \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
/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
all inboxes | Powered by JetHome®