Re: make genattrtab generate smaller code

[Andreas Jaeger <aj at suse dot de>]

Michael Matz <matz@suse.de> writes:

> Hi,
>
> On Mon, 8 Sep 2003 law@redhat.com wrote:
>
>> I believe the "canonical" test for genattrtab optimizations was always
>> the MIPS port.  IIRC the MIPS port was consistently the most
>> problematical.
>>
>> You might repeat your test with a MIPS target and see what happens.
>
> I did now with a x86->mips-elf compiler on the (basically same) kdecore.ii
> file:
>
> kdecore.ii mips, HEAD
> real    4m4.732s
> user    4m0.640s
> sys     0m3.940s
>
> kdecore.ii mips, no genattrtab optimization
> real    4m5.499s
> user    4m1.490s
> sys     0m3.870s
>
> (this is with an unoptimized compiler)
>
>   535992 2003-09-08 17:07 insn-attrtab.new.c
>   269240 2003-09-08 17:07 insn-attrtab.new.o
>   625268 2003-09-08 16:16 insn-attrtab.old.c
>   366092 2003-09-08 16:16 insn-attrtab.old.o
>
> genattrtab doesn't take any noticable time while optimizing the tree in
> both cases.  So it doesn't really make any big difference on MIPS.

Does it generate the sampe kdecore file?  In that case I think you
passed the test ;-)

Andreas
-- 
 Andreas Jaeger, aj@suse.de, http://www.suse.de/~aj
  SuSE Linux AG, Deutschherrnstr. 15-19, 90429 Nürnberg, Germany
   GPG fingerprint = 93A3 365E CE47 B889 DF7F  FED1 389A 563C C272 A126

Attachment: pgp00000.pgp
Description: PGP signature