[Bug target/114414] New: [14 Regression] 15-18% exec time slowdown of 433.milc on Zen2

[Bug target/114414] New: [14 Regression] 15-18% exec time slowdown of 433.milc on Zen2

[pheeck at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114414

            Bug ID: 114414
           Summary: [14 Regression] 15-18% exec time slowdown of 433.milc
                    on Zen2
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Keywords: missed-optimization, needs-bisection
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: pheeck at gcc dot gnu.org
            Blocks: 26163
  Target Milestone: ---
              Host: x86_64-linux
            Target: x86_64-linux

As seen here

https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=685.70.0

there is an 18% slowdown of the 433.milc SPEC 2006 benchmark between commits

g:74e8cc28eda9b1d7
g:81ee1298b47d3f3b

The compilation options are -O2 -flto PGO. The CPU is AMD Zen2. I've replicated
this slowdown on another Zen2 machine. There the slowdown was only 15%.

Btw here is the same plot but with comparisons to GCC13 and GCC12

https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=685.70.0&plot.6=921.70.0&plot.7=762.70.0


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26163
[Bug 26163] [meta-bug] missed optimization in SPEC (2k17, 2k and 2k6 and 95)

[Bug target/114414] [14 Regression] 15-18% exec time slowdown of 433.milc on Zen2

[pheeck at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114414

Filip Kastl <pheeck at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |14.0

[Bug target/114414] [14 Regression] 15-18% exec time slowdown of 433.milc on Zen2

[law at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114414

Jeffrey A. Law <law at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |law at gcc dot gnu.org
           Priority|P3                          |P2

[Bug target/114414] [14 Regression] 15-18% exec time slowdown of 433.milc on Zen2

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114414

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
I'll note the performance is now again close to that of 12/13 but improvements
have been lost (so it might be not called a 14 regression).

[Bug target/114414] 15-18% exec time slowdown of 433.milc on Zen2

[pheeck at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114414

Filip Kastl <pheeck at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[14 Regression] 15-18% exec |15-18% exec time slowdown
                   |time slowdown of 433.milc   |of 433.milc on Zen2
                   |on Zen2                     |

--- Comment #2 from Filip Kastl <pheeck at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #1)
> I'll note the performance is now again close to that of 12/13 but
> improvements have been lost (so it might be not called a 14 regression).

I misunderstood what [14 Regression] means exactly. I'll remove it from the
bug's name.

[Bug target/114414] 15-18% exec time slowdown of 433.milc on Zen2

[pheeck at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114414

Filip Kastl <pheeck at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED

--- Comment #3 from Filip Kastl <pheeck at gcc dot gnu.org> ---
It looks like the benchmark went back to its original values so I'm closing
this PR.