Re: buildworld times

From: Jason <jason_at_ec.rr.com>
Date: Sat, 28 Feb 2004 22:09:15 -0500
Vincent Poy wrote:

>On Sat, 28 Feb 2004, Jason wrote:
>
>  
>
>>Vincent Poy wrote:
>>
>>    
>>
>>>Hello all:
>>>
>>>	I was wondering what people are getting as the fastest make
>>>buildworld times since it seems like a P4m-2.6Ghz 2GB RAM, 7200RPM HDD
>>>machine when using make -j4 buildworld or just make buildworld takes 2.5
>>>hours.  I guess I'm trying to look for what the average times for
>>>buildworld is supposed to be for a Pentium 4 Northwood based platform.
>>>      
>>>
>>On my athlon xp running at 2ghz, 512mb 400mhz ram, 7200rpm dma 133 hd,
>>nforce2 I got 30-35min on a buildworld for current a week ago, but I had
>>to disable my nocyrpto option in make.conf to get mergemaster to work,
>>so that took me to 40-45min.  I used $make -j4 -s buildworld.  I'll add
>>my config file to show how to configure for 5.2.1 or current for speed.
>>Also I do like amd stuff better than intel, because I think the p4
>>netbust(yes I know it is netburst, but it seems to have fallen short of
>>expectations, at least for me)arch. sucks, so I wouldn't mind seeing
>>some buildworld benchmarchs and  statics to which is faster.
>>    
>>
>
>	I think it has more to do with the debug options as I can see from
>your CONFIG file.  How long does your buildkernel take?  Also, is there
>something to time the buildworld process since I notice it has a start
>time but it doesn't tell you at the end, how long it takes exactly.  I
>basically do a script /home/root/mw.out and then do the make buildworld
>under that.  All I can tell is that the buildworld started at:
>Script started on Sat Feb 28 01:53:16 2004
>You have mail.^M
>root_at_bigbang [1:53am][/usr/src] >> make -j4 u^H^[[Kbuildworld^M^M
>and then it ended at:
>===> usr.sbin/zzz^M
>gzip -cn /usr/src/usr.sbin/zzz/zzz.8 > zzz.8.gz^M
>root_at_bigbang [4:36am][/usr/src] >>
>
>	The original Willimette P4 was pretty poor in performance but I
>think the Northwood is better.  I'm running a Northwood Mobile P4 2.6Ghz
>CPU which is the maximum this machine supports so I don't really have a
>choice and this is a high-end notebook at the time as it pretty much has
>the same specs as the Dell Precision M50 with 2GB Memory which I did by
>changing the HD, video card and also the CPU and memory.  My desktop runs
>on a P4-3.2C with HyperThreading and 2GB and 7200rpm drives as well and
>that machine is pretty fast.
>
>
>Cheers,
>Vince - vince_at_WURLDLINK.NET - Vice President             ________   __ ____
>Unix Networking Operations - FreeBSD-Real Unix for Free / / / / |  / |[__  ]
>WurldLink Corporation                                  / / / /  | /  | __] ]
>San Francisco - Honolulu - Hong Kong                  / / / / / |/ / | __] ]
>HongKong Stars/Gravis UltraSound Mailing Lists Admin /_/_/_/_/|___/|_|[____]
>Almighty1_at_IRC - oahu.DAL.NET Hawaii's DALnet IRC Network Server Admin
>
>
>  
>
I agree with the debugging options making the biggest difference.   I 
think my buildkernel takes about or less than 10min, I also build my 
stuff in a terminal window with x running.  You will see a speed up if 
you drop into single user mode.  The system has a built in timer, just 
type time in front of a command and it times it for you.  Yes, the p4c 
is the last p4 before they really changed things.  The whatever(ALU?) 
units that used to run at 2x clock are now at core clock speed.  I think 
they also increased the pipeline again.  The current p4 has lower ipc 
than the last model, but is going to higher clock speed(supposedly) to 
run faster all together.  They did improve serval other parts like the 
branch predictor, which must be dead on to get good performance form 
such a deep pipeline.  As for ram, the 2gigs sounds nice but infact it 
will slow you down if it is not regestered.   Intel says it won't hurt 
speed to 1gig or more and to back this up they show how you loose only 
10% bandwidth or something less while not mentioning the access latinces 
go through the roof.  It has been a while since I last read up on intel 
stuff so I recommend if this stuff intrestest you to look into it more 
deeply yourself.
Received on Sat Feb 28 2004 - 18:08:42 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:45 UTC