[Bug target/84251] [9/10/11/12 Regression] Performance regression in gcc 8/9/10/11/12 when comparing floating point numbers

[Bug target/84251] [9/10/11/12 Regression] Performance regression in gcc 8/9/10/11/12 when comparing floating point numbers

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84251

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|8.5                         |9.4

--- Comment #14 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 8 branch is being closed.

[Bug target/84251] [9/10/11/12 Regression] Performance regression in gcc 8/9/10/11/12 when comparing floating point numbers

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84251

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|9.4                         |9.5

--- Comment #15 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 9.4 is being released, retargeting bugs to GCC 9.5.

[Bug target/84251] [10/11/12/13 Regression] Performance regression in gcc 8/9/10/11/12 when comparing floating point numbers

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84251

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|9.5                         |10.4

--- Comment #16 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 9 branch is being closed

[Bug target/84251] [10/11/12/13 Regression] Performance regression in gcc 8/9/10/11/12 when comparing floating point numbers

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84251

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|10.4                        |10.5

--- Comment #17 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 10.4 is being released, retargeting bugs to GCC 10.5.

[Bug target/84251] [10/11/12/13/14 Regression] Performance regression in gcc 8/9/10/11/12 when comparing floating point numbers

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84251

--- Comment #18 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Seems like we could use the range while doing the expansion even?
Or some how mark the EQ_EXPR as not having to worry about NANs here during VRP.

[Bug target/84251] [11/12/13/14 Regression] Performance regression in gcc 8/9/10/11/12 when comparing floating point numbers

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84251

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|10.5                        |11.5

--- Comment #19 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 10 branch is being closed.

[Bug target/84251] [11/12/13/14 Regression] Performance regression in gcc 8/9/10/11/12 when comparing floating point numbers

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84251

--- Comment #20 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Hmm: we have:
```
(jump_insn 8 7 9 2 (set (pc)
        (if_then_else (ordered (reg:CCFP 17 flags)
                (const_int 0 [0]))
            (label_ref 12)
            (pc))) "/app/example.cpp":5:6 1055 {*jcc}
     (expr_list:REG_DEAD (reg:CCFP 17 flags)
        (int_list:REG_BR_PROB 1073741831 (nil)))
 -> 12)
(note 9 8 10 3 [bb 3] NOTE_INSN_BASIC_BLOCK)
(call_insn 10 9 12 3 (call (mem:QI (symbol_ref:DI ("abort") [flags 0x41] 
<function_decl 0x7f2aa863b500 __builtin_abort>) [0 __builtin_abort S1 A8])
        (const_int 0 [0])) "/app/example.cpp":6:5 1062 {*call}
     (expr_list:REG_CALL_DECL (symbol_ref:DI ("abort") [flags 0x41] 
<function_decl 0x7f2aa863b500 __builtin_abort>)
        (expr_list:REG_ARGS_SIZE (const_int 0 [0])
            (expr_list:REG_NORETURN (const_int 0 [0])
                (expr_list:REG_EH_REGION (const_int 0 [0])
                    (nil)))))
    (nil))
(code_label 12 10 13 4 2 (nil) [1 uses])
(note 13 12 15 4 [bb 4] NOTE_INSN_BASIC_BLOCK)
(insn 15 13 16 4 (set (reg:QI 91)
        (ordered:QI (reg:CCFP 17 flags)
            (const_int 0 [0]))) "/app/example.cpp":7:12 1050 {*setcc_qi}
     (expr_list:REG_DEAD (reg:CCFP 17 flags)
        (nil)))
```
We know in insn 16 at this point (ordered:QI (reg:CCFP 17 flags) (const_int 0
[0])) is 1. Isn't there a pass which does that?
Or is it because the differences in the modes (VOIDmode vs QImode) causing that
not to happen?