[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Hurd performance [was: PPP]



On 20 Jul 1998, Gordon Matzigkeit wrote:

>  ML> I think that [Brent Fulgham] was speaking about my own web page at
>  ML> http://www.via.ecp.fr/~walken/hurd/compile.html... On my machine
>  ML> (PII/233 w/ 64M of ram) with a hurd 0.2 distribution, it takes 6
>  ML> hours to compile the glibc vs 75 mins if I cross-compile it from
>  ML> Linux.

> Something that just occurred to me: are you using a cross-GCC that was
> optimized in the same way as the Hurd native GCC?

ack. I didnt think about checking this. I didnt played with the gcc
building options, either : my gcc cross-compiler is built exactly as
described on my web page, using the default optimization options set by
./configure and the makefiles. I think that this includes optimization. my
native hurd gcc compiler is from the gnu-0.2 distribution, so I guess it
must have the debug information compiled in. I will check into this later
this week....

Still, the slowness of "make" when working in a big directory (compiling
the gnu libc for example) is extremely noticeable, too. It may take up to
30 seconds between two runs of gcc. I think this one is due to a slow
stat() implementation. 

Michel "Walken" LESPINASSE - Student at Ecole Centrale Paris (France)
           www              walken@via.ecp.fr
          (o o)            http://www.via.ecp.fr/~walken/
------oOO--(_)--OOo------ C0 9B E8 D8 44 43 D3 63  5A B4 BA 55 57 5B 19 6D
     "Just say know"     finger walken@zen.via.ecp.fr for complete PGP key



-- 
To UNSUBSCRIBE, email to debian-hurd-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org


Reply to: