Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!labrea!decwrl!sun!shukra!ram
From: ram@shukra.UUCP
Newsgroups: comp.arch
Subject: The minute Barrier
Message-ID: <27023@sun.uucp>
Date: Tue, 1-Sep-87 19:18:47 EDT
Article-I.D.: sun.27023
Posted: Tue Sep  1 19:18:47 1987
Date-Received: Thu, 3-Sep-87 04:25:11 EDT
Sender: news@sun.uucp
Lines: 30
Keywords: Kernel Make times



   ---------------------------------------------------------------------------
   DISCLAIMER: The numbers quoted here in no way reflects the MIPS/FLOPS/CLUNKS
   in a machine.  Its just FYI
   ---------------------------------------------------------------------------

   I guess I screwed up in my original statement about kernel make time on
   sequent.  The < 3hr is for kernel+utilities(/bin/, /usr/bin, /usr/ucb, /etc
   etc) + other extras that sequent has.  Phil@sequent pointed this out
   and claimed that it takes 2h 17m on a 20 processor NS32032.  Kernel (4.2)
   build is 12 minutes for sequent and 4.5 mts on Encore (again a 4.2 based
   32[0/3]32) - an 18 processor machine (I guess the factor three difference
   should be due to differences in kernel functionality and maybe processors)
   With a 386 based 30 processor sequent symmetric - Phil is aiming at the 1
   minute barrier.

   Note:[ Amdahl advertises that it makes it in < 3 mts].

   Anybody have figures for other machines with single digit times.

   Here are the days of touch *.c *.h;make -DKERNEL;/etc/reboot and 
   watch  Ben Johnson complete a 100 m dash.


---------------------
   Renu Raman				ARPA:ram@sun.com
   Sun Microsystems			UUCP:{ucbvax,seismo,hplabs}!sun!ram
   M/S 5-40, 2500 Garcia Avenue,
   Mt. View,  CA 94043