public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug tree-optimization/93328] missed optimization opportunity in deserialization code
       [not found] <bug-93328-4@http.gcc.gnu.org/bugzilla/>
@ 2020-03-20  2:42 ` pacoarjonilla at yahoo dot es
  2020-03-23  5:35 ` pacoarjonilla at yahoo dot es
                   ` (2 subsequent siblings)
  3 siblings, 0 replies; 4+ messages in thread
From: pacoarjonilla at yahoo dot es @ 2020-03-20  2:42 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93328

Paco Arjonilla <pacoarjonilla at yahoo dot es> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pacoarjonilla at yahoo dot es

--- Comment #2 from Paco Arjonilla <pacoarjonilla at yahoo dot es> ---
Using shifts and bitwise does not optimize either in GCC master, but it does
optimize in CLang.

Take this code:

#include <cstdint>
using type = std::uint32_t;
type foo(type v){
    type r  = (v << 24) & 0xFF000000;
         r += (v << 8)  & 0x00FF0000;
         r += (v >> 8)  & 0x0000FF00;
         r += (v >> 24) & 0x000000FF;
    return r;
}

CLang assembler output: ( optimize with -O2 )

        mov     eax, edi
        bswap   eax
        ret


GCC master assembler output: ( optimize with -O2 )
        mov     eax, edi
        mov     edx, edi
        shr     eax, 24
        sal     edx, 24
        add     edx, eax
        mov     eax, edi
        shr     edi, 8
        sal     eax, 8
        and     edi, 65280
        and     eax, 16711680
        add     eax, edx
        add     eax, edi
        ret

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug tree-optimization/93328] missed optimization opportunity in deserialization code
       [not found] <bug-93328-4@http.gcc.gnu.org/bugzilla/>
  2020-03-20  2:42 ` [Bug tree-optimization/93328] missed optimization opportunity in deserialization code pacoarjonilla at yahoo dot es
@ 2020-03-23  5:35 ` pacoarjonilla at yahoo dot es
  2020-03-30 15:46 ` boris_oncev at hotmail dot com
  2020-03-30 15:48 ` boris_oncev at hotmail dot com
  3 siblings, 0 replies; 4+ messages in thread
From: pacoarjonilla at yahoo dot es @ 2020-03-23  5:35 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93328

--- Comment #3 from Paco Arjonilla <pacoarjonilla at yahoo dot es> ---
But this gets optimized indeed!

#include <cstdint>
using type = std::uint32_t;
type foo(type v){
    type r  = ((v << 24) & 0xFF000000)
            | ((v << 8)  & 0x00FF0000)
            | ((v >> 8)  & 0x0000FF00)
            | ((v >> 24) & 0x000000FF);
    return r;
}

GCC -O2:

foo(unsigned int):
        mov     eax, edi
        bswap   eax
        ret

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug tree-optimization/93328] missed optimization opportunity in deserialization code
       [not found] <bug-93328-4@http.gcc.gnu.org/bugzilla/>
  2020-03-20  2:42 ` [Bug tree-optimization/93328] missed optimization opportunity in deserialization code pacoarjonilla at yahoo dot es
  2020-03-23  5:35 ` pacoarjonilla at yahoo dot es
@ 2020-03-30 15:46 ` boris_oncev at hotmail dot com
  2020-03-30 15:48 ` boris_oncev at hotmail dot com
  3 siblings, 0 replies; 4+ messages in thread
From: boris_oncev at hotmail dot com @ 2020-03-30 15:46 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93328

Boris <boris_oncev at hotmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |boris_oncev at hotmail dot com

--- Comment #4 from Boris <boris_oncev at hotmail dot com> ---
*** Bug 94403 has been marked as a duplicate of this bug. ***

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug tree-optimization/93328] missed optimization opportunity in deserialization code
       [not found] <bug-93328-4@http.gcc.gnu.org/bugzilla/>
                   ` (2 preceding siblings ...)
  2020-03-30 15:46 ` boris_oncev at hotmail dot com
@ 2020-03-30 15:48 ` boris_oncev at hotmail dot com
  3 siblings, 0 replies; 4+ messages in thread
From: boris_oncev at hotmail dot com @ 2020-03-30 15:48 UTC (permalink / raw)
  To: gcc-bugs

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93328

--- Comment #5 from Boris <boris_oncev at hotmail dot com> ---
full code:
https://godbolt.org/z/zjNqYV

template <typename T>
auto reverse(T num) {

    // misses optimization when num is int32_t OK for int64_t
    auto* bytes = reinterpret_cast<char*>(&num);

    // misses optimization for both 32 and 64 bit ints
    //auto* bytes = reinterpret_cast<std::byte*>(&num);

    constexpr auto size = sizeof(num);
    for (int i = 0; i < size / 2; i++) {
        std::swap(bytes[i], bytes[size-i-1]);
    }

    return num;
}

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2020-03-30 15:48 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
     [not found] <bug-93328-4@http.gcc.gnu.org/bugzilla/>
2020-03-20  2:42 ` [Bug tree-optimization/93328] missed optimization opportunity in deserialization code pacoarjonilla at yahoo dot es
2020-03-23  5:35 ` pacoarjonilla at yahoo dot es
2020-03-30 15:46 ` boris_oncev at hotmail dot com
2020-03-30 15:48 ` boris_oncev at hotmail dot com

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).