Re: default CPU target for ix86 based ports
Jan-Benedict Glaw said:
>...and up to now, I haven't seen real hard numbers that show that
>optimizing for i486 does really make anything noticeable faster. From
I've given such numbers for the use of i486 instructions. The speed
increase applies to specific applications only and is quite large in
those applications. I don't think there is any significant use of the
special i486 instructions in any other applications (GCC didn't generate
them, last I checked), so it shouldn't slow down anything else on i386
at all.
The --tune option makes a marginal difference generally in *either*
direction (and does not affect compatibility) so I think it's best to
tune for the vast majority of machines. --tune=i686 tunes for Pentium
Pro and up, which is the vast majority of machines.
>my point of view, you're making something marginally (at best...)
>faster but giving up i386 compatibility (relying on a hackish emulator
>which isn't right now available for the latest kernel).
i386 compatibility already requires the emulator for all of C++, and
hence for apt. This isn't going to change since it would change the
C++ ABI.
Making the emulator available for the latest kernel is certainly
important. Volunteering? ;-)
>Maybe we'd go another way and build two distributions - i386 as well as
>i486 or i586. I bet there are still i386 machines out there, but
Feel free; this appears to have been rejected by the Debian project in
favor of the emulator. (I can certainly understand why, since it's a
lot easier!)
>they're not updated that often. Please don't cut their update pathes...
--
Nathanael Nerode <neroden at gcc.gnu.org>
http://home.twcny.rr.com/nerode/neroden/fdl.html
Reply to: