FreeBSD 6 is coming too fast
Peter Schultz
pmes at bis.midco.net
Mon Apr 25 09:29:16 PDT 2005
On Apr 25, 2005, at 10:19 AM, Danny Braniss wrote:
>> On Mon, Apr 25, 2005 at 08:44:34AM -0600, Scott Long wrote:
>>>> No, I'm not going to do it because of lack of knowledge, there are
>>>> people who have more experience with it than me.
>>>>
>>> Well, as I said in another email, switching to GCC 4 just because of
>>> dubious "25% faster" (faster at what? compiling? resulting
>>> generated
>>> code? crashing?) claims in the changelog is not a terribly good
>>> reason =-)
>>
>> 25% faster to compile the code, not running it.
>
> so that closes the argument! i can do a makeworld in about 25 minutes,
> cutting
> it down by 6 minutes will not make any real difference
>
I have a seven year old dual PII 350 machine here, and since the switch
to 6-CURRENT I have seen *significant* increases in performance. Great
job everyone!
That being said, compiling FreeBSD in 25 minutes is exactly what's
wrong and in my opinion a big reason the thread was started. Moore's
Law is nifty and all but it doesn't help create excellent software. I
say all FreeBSD developers should be forcing themselves to run old
hardware so they're compelled to write better/faster code. Of course
the slow hardware isn't necessary, but it'll prompt coders to go back
to the matter at hand instead of just running bad code brute force with
massive processors.
Pete...
More information about the freebsd-current
mailing list