On Saturday 16 February 2002 08:14 pm, Sridhar Dhanapalan wrote:
> On Sat, 16 Feb 2002 18:13:39 -0500, Tom Brinkman
> > Nothin scientific, I used the clock on the microwave
> > (computer's in the kitchen) to test mozilla, and kernel compiles
> > ... ie, � a minute ;~> i686, or athlon, proved slower more times
> > than not. Specially with kernels compiled for i686 or athlon,
> > rather than Mandrake's 'stock' i586 kernels. Even against testing
> > usin 'optimized' 'trimmed down' kernels to run the tests.
>
> I think that the more you 'optimise' code, the longer it can take to
> compile.
Well, in the case of the kernel, I used the 'stock' Mandrake
.config and only hand edited for cpu type. Compile times (default
/boot/.config) were unchanged. Triming out some uneeded stuff with
'make xconfig' produced a smaller kernel. Using that kernel optimized
only for athlon, then a compile of the kernel (using the default
.config again) was actually a touch slower.
So I'm stickin to my story. In my experience over the years,
optimizing for cpu, and or 'lighter' kernels .... doesn't improve
performance. At least not enough to make the whole proccess worth
while. YMMV
>
> BTW, perhaps you should look into the 'time' command? :)
I did, but it wouldn't work with a kernel compile, eg, 'make dep &&
make clean && make bzImage && make modules' I suspect the multiple
commands are the reason.
--
Tom Brinkman Corpus Christi, Texas, USA
Want to buy your Pack or Services from MandrakeSoft?
Go to http://www.mandrakestore.com
