[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Gcl-devel] recent commit
From: |
Camm Maguire |
Subject: |
[Gcl-devel] recent commit |
Date: |
Thu, 28 Aug 2003 13:36:29 -0400 |
User-agent: |
SEMI/1.14.3 (Ushinoya) FLIM/1.14.3 (Unebigory ōmae) APEL/10.3 Emacs/21.2 (i386-debian-linux-gnu) MULE/5.0 (SAKAKI) |
[ This seems to have gotten lost from the list. Reposted now. CM]
reassign 204789 libc6
thanks
Greetings all! A few comments on the latest upload onto HEAD and the
release branch:
--enable-static configuration option -- default on ia64, as a
workaround for current algorithm of runtime realized function
descriptors. I'll be releasing gcl/maxima/acl2 packages soon with
this in to cure 204789 for now. I'm reassigning the bug to libc6 in
case the ld.so ia64 maintainers want to take a look at this. The
details are in the BTS, but in summary, stored function addresses in
gcl executables ostensibly referring to functions in same said
executable (via function descriptors) and saved to a file via
unexec can wind up referring to incorrect function descriptor
addresses when the executable is run against compatible but
different shared library versions.
amd64-linux support -- e.g "Opteron". This is the first 64bit port
using bfd relocation. Brings us to 12 Linux ports.
O6->O3 -- I did some brief benchmarking -- O3 seems a bit larger and
slightly faster than O2. I'm open to reconsidering and changing O3 ->
O2 if someone can provide meaningful benchmarks to demonstrate an
improvement.
Take care,
-- Camm Maguire address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens." --
Baha'u'llah
Camm Maguire wrote:
>Greetings! I don't suppose you have the whole email I sent you here.
>For some reason it didn't get to address@hidden, and I'd like to
>get it into the archives. If you could send it to me I'd be most
> grateful. Take care,
>
>Erik Andersson <address@hidden> writes:
>
>
>>On Thu, 21 Aug 2003, Camm Maguire wrote:
>>
>>
>>>O6->O3 -- I did some brief benchmarking -- O3 seems a bit larger and
>>>slightly faster than O2. I'm open to reconsidering and changing O3 ->
>>>O2 if someone can provide meaningful benchmarks to demonstrate an
>>>improvement.
>>>
>>According to the GCC manual the only difference between O2 and O3 is
>>-frename-registers and -finline-functions. I think that providing
>>meaningful benchmarks is difficult because different processors react
>>differently to -finline-functions depending on things like the size of the
>>instruction cache. Something that you could also try would be Os which use
>>all O2 optimizations that don't typically increase code size.
>>
>>/Erik Andersson
>>
>>
>>_______________________________________________
>>Gcl-devel mailing list
>>address@hidden
>>http://mail.gnu.org/mailman/listinfo/gcl-devel
>>
>>
>>
>>
>
>
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- [Gcl-devel] recent commit,
Camm Maguire <=