Re: FreeBSD 6 is coming too fast

From: /dev/null <null_at_dnswatch.com>
Date: Mon, 25 Apr 2005 03:54:04 -0700 (PDT)
Greetings Jon/ All,
 Thank you for the *informative* response!
    ...

> On 04/24/05 18:29, /dev/null wrote:
>> <snip>
>>
>> needed. All in all life on 5.x and the "upgrade" wasn't too bad. I will
>> say that there is ONE issue that I have found and have not yet solved.
>> It
>> now takes at least 2 times longer to build any of the ports. Performance
>> in other areas seems to be lagging as well. I have since upgraded one of
>> the 2 servers to 5.4-RC2 and have been chasing 5.x ever since hoping to
>> find the performance issues will dissappear.
>
> If you are running a UP system, it is expected that 4.x will outperform
> 5.x in many situations due to the focus on SMP.  Optimizing
> synchronization to increase performance is one of the main goals for 6.x
> (see the recent work on critical sections, for example).  This will
> allow us to scale well on SMP systems without pessimizing performance on
> UP systems.

Ah hah, I see. Well, it's funny you should bring up SMP. I bought one
about 2yrs. ago for the sole purpose of building an SMP FBSD kernel on
it, joined the list and started building. Turns out I, nor anyone on the
list could get one to work. Of all things, it was the AHA-2940 that killed
the whole thing. It just kept polling and resetting the drives. Never
brought 'em up. So now Win2k Advanced Server lives on it. GRRRRRRR >:(
Anyway, don't get me started. ;)
 But don't get me wrong. I'm truly greatful for your answer - REALLY.
I'm actually waiting for SIX to level out so I can take another stab at
it and remove another copy of Winblows. This time I'll try to take
advantage of the UDMA 100 on the SMP board, and when the AHA-2940UW
drivers start working on an SMP (in 6?), I can add the SCSI devices then.

>
> Another point to remember is that compilation times with GCC 3.4
> (default for recent 5.x) are much longer than those with 2.95 (default
> for 4.x), especially at higher optimization levels.  This is one of the
> main reasons why it takes longer to compile a port.

Interesting. Does this also apply to 5.0-RC2? I never noticed any
"slow-downs" until I loaded 5.3-STABLE.

>
> That said, in what specific areas are you seeing performance regressions?

Thus far, I haven't been able to pin it down. That is *something* seems
to start gobbling up CPU and/ or resources to the extent I notice slight
to moderate delays in things I'm doing. So far, only appears to occur
while I'm running X (Fluxbox + Filerunner + Xterm + NEdit and sometimes
Mozilla 1.7.7). But not the whole time, just for periods of time with
no (currently) apparent rhythm. I haven't yet been able to associate it
with anything other than X.
Anyone else?

-Chris

P.S. Thanks again!

>
> Jon
> _______________________________________________
> freebsd-current_at_freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe_at_freebsd.org"
>


////////////////////////////////////////////////////
 If only Western Electric had found a way to offer
binary licenses for the UNIX system back in 1974,
the UNIX system would be running on all PC's today
rather than DOS/Windows.
////////////////////////////////////////////////////
Received on Mon Apr 25 2005 - 08:54:13 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:32 UTC