[Bug c/30287] New: -fregmove produces slow code

[Bug c/30287] New: -fregmove produces slow code

[safari-ml-gcc-bugzilla-vgvbszlod4jrh2zwsefgqp4frqg4zc7fgobtxnvg at u dot safari dot iki dot fi]

$ gcc aes256.i -O2 ; ./a.out
min: 651 avg: 686 extra: 644 discarded: 6766/10000

$ gcc aes256.i -O2 -fno-regmove ; ./a.out
min: 385 avg: 387 extra: 756 discarded: 66/10000

aes_encrypt is the function which is being benchmarked...
aes256.i.bz2 is pretty big but I can't help it.

My CPU is Pentium D.


other gcc versions
regmove        1      0
gcc-2.95.3    572    387
gcc-3.0.4     570    376
gcc-3.1.1     565    365
gcc-3.2.3     569    367
gcc-3.3.5     477    351
gcc-3.4.6     494    350
gcc-4.0.3     651    385
gcc-4.1.1-20061220 (Red Hat 4.1.1-48)
              596    355

icc-9.1        ?     342


-- 
           Summary: -fregmove produces slow code
           Product: gcc
           Version: 4.0.3
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: safari-ml-gcc-bugzilla-
                    vgvbszlod4jrh2zwsefgqp4frqg4zc7fgobtxnvg at u dot safari
                    dot iki dot fi
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30287

[Bug c/30287] -fregmove produces slow code

[safari-ml-gcc-bugzilla-vgvbszlod4jrh2zwsefgqp4frqg4zc7fgobtxnvg at u dot safari dot iki dot fi]

------- Comment #1 from safari-ml-gcc-bugzilla-vgvbszlod4jrh2zwsefgqp4frqg4zc7fgobtxnvg at u dot safari dot iki dot fi  2006-12-24 15:53 -------
Created an attachment (id=12842)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=12842&action=view)
aes_encode performance test


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30287

[Bug rtl-optimization/30287] -fregmove produces slow code

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2006-12-24 16:32 -------
The only thing regmove does is makes the RA work harder as it make the life
range of the psedu registers longer.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30287

[Bug rtl-optimization/30287] -fregmove produces slow code

[steven at gcc dot gnu dot org]

------- Comment #3 from steven at gcc dot gnu dot org  2007-04-04 12:12 -------
I want to understand this better, so I'll investigate this.


-- 

steven at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |steven at gcc dot gnu dot
                   |dot org                     |org
             Status|UNCONFIRMED                 |ASSIGNED
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2007-04-04 12:12:31
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30287

[Bug rtl-optimization/30287] -fregmove produces slow code

[steven at gcc dot gnu dot org]

------- Comment #4 from steven at gcc dot gnu dot org  2007-05-25 20:35 -------
Something in the backward pass increases the register pressure, inducing extra
spills.  Not much can be done about this except for tuning the transformations
involved.


-- 

steven at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|steven at gcc dot gnu dot   |unassigned at gcc dot gnu
                   |org                         |dot org
             Status|ASSIGNED                    |NEW


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30287