[Bug c++/99582] New: No intrinsics to access rcl or rcr instruction on x86_64

[Bug c++/99582] New: No intrinsics to access rcl or rcr instruction on x86_64

[unlvsur at live dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99582

            Bug ID: 99582
           Summary: No intrinsics to access rcl or rcr instruction on
                    x86_64
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: unlvsur at live dot com
  Target Milestone: ---

How to deal with this? Any compiler patterns to access this instruction?

[Bug target/99582] No intrinsics to access rcl or rcr instruction on x86_64

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99582

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
ROR can be open-coded, GCC tries hard to detect common patterns.

RCR is difficult because it involves the carry flag and thus the carry producer
would have to be explicit.  Or it would need to be an awkward intrinsic
like

__builtin_rcr ({unsigned, unsigned short, unsigned char} word, bool carry,
unsigned amnt);

and you'd need to hope the compiler can arrange 'carry' to be produced in
the carry flag ...

[Bug target/99582] No intrinsics to access rcl or rcr instruction on x86_64

[unlvsur at live dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99582

--- Comment #2 from cqwrteur <unlvsur at live dot com> ---
(In reply to Richard Biener from comment #1)
> ROR can be open-coded, GCC tries hard to detect common patterns.
> 
> RCR is difficult because it involves the carry flag and thus the carry
> producer
> would have to be explicit.  Or it would need to be an awkward intrinsic
> like
> 
> __builtin_rcr ({unsigned, unsigned short, unsigned char} word, bool carry,
> unsigned amnt);
> 
> and you'd need to hope the compiler can arrange 'carry' to be produced in
> the carry flag ...

so is that possible to __builtin_rcl and __builtin_rcr? I think a lot of
cryptography and multi-precision libraries need that for pow stuffs

[Bug target/99582] No intrinsics to access rcl or rcr instruction on x86_64

[amonakov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99582

Alexander Monakov <amonakov at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |amonakov at gcc dot gnu.org

--- Comment #3 from Alexander Monakov <amonakov at gcc dot gnu.org> ---
RCL and RCR are supported via microcode sequencer on Intel and involve many (9)
uops on modern AMD, so they are quite slow in comparison to simple
shifts/rotates. Would library developers still want to use them despite the
poor performance? Equivalent code with "classic" shifts should be more
efficient.

https://uops.info/html-instr/RCL_R64_CL.html

[Bug target/99582] No intrinsics to access rcl or rcr instruction on x86_64

[unlvsur at live dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99582

--- Comment #4 from cqwrteur <unlvsur at live dot com> ---
(In reply to Alexander Monakov from comment #3)
> RCL and RCR are supported via microcode sequencer on Intel and involve many
> (9) uops on modern AMD, so they are quite slow in comparison to simple
> shifts/rotates. Would library developers still want to use them despite the
> poor performance? Equivalent code with "classic" shifts should be more
> efficient.
> 
> https://uops.info/html-instr/RCL_R64_CL.html

here they want is to get the carry flag from rcl/rcr