On Tue, 2006-08-08 at 09:55 +0100, Matthew Allum wrote:
> On Tue, 2006-08-08 at 09:06 +0100, Michael Meeks wrote:
> > Did someone post a clear profile ( painting the problem sideways-on ;-)
> > before optimising, a link to that appreciated,
>
> oprofile on ARM ( pxa250 ) stats focus on text rendering comparing GTK
> 2.6 to 2.8; http://www.o-hand.com/~jorn/pango-benchmarks/
Ah - interesting; so poking at:
http://www.o-hand.com/~jorn/pango-benchmarks/28/full-report.txt
I see a ton of vmlinux stuff apparently emulating the FPU - but,
unfortunately, it doesn't appear to tell me where those calls came from
(?) is there a pretty picture that shows that ? [ something like
'sysprof' output would do that I guess ].
Also - I notice my old friend:
1 0.0679 ld-2.3.90.so ld-2.3.90.so
_dl_relocate_object
1 0.0679 ld-2.3.90.so ld-2.3.90.so _dl_fixup
1470 99.8641 ld-2.3.90.so ld-2.3.90.so
_dl_lookup_symbol_x
1236 0.5132 ld-2.3.90.so ld-2.3.90.so do_lookup_x
1236 83.9674 ld-2.3.90.so ld-2.3.90.so
do_lookup_x [self]
197 13.3832 ld-2.3.90.so ld-2.3.90.so strcmp
39 2.6495 ld-2.3.90.so ld-2.3.90.so
_dl_name_match_p
A small but useful part of my research in this area just got merged
up-stream ( to binutils / glibc ) implementing a new, far more efficient
hash for symbol lookup[1]. The punch line is if you care enough, you
should back-port the glibc & binutils patch, and use
-Wl,--hash-style=both or --hash-style=gnu if you want to re-do your ABI
to save some space :-)
HTH,
Michael.
[1] - http://sourceware.org/ml/binutils/2006-07/msg00052.html - a ton of
development discussion.
--
[EMAIL PROTECTED] <><, Pseudo Engineer, itinerant idiot
_______________________________________________
Performance-list mailing list
[email protected]
http://mail.gnome.org/mailman/listinfo/performance-list
