Vareck Bostrom writes:
>In message <199411270655.RAA08311@splode.mame.mu.OZ.AU>, matthew green writes:
>>
>> Sigh.  When I rebooted with the new kernel and started X, I saw an immediate
>> improvement.  X took just over 1/2 as long to start up.  then a few hours
>> later, I decided to try out the dhry tests again.  I almost wish I hadn't.
>
> I'd like to confirm that Theo's fix, while it appears to make the machine
> somewhat faster (perhaps because of an overactive imagination on my part)
> dhrystone test does not show any improvment, I still show dhry-dynamic of
> 8800 on a SPARCstation IPC.

Just as a coincidence, I decided to take the U.S. Thanksgiving holiday weekend
as a chance to rebuild X11 R6 from scratch on both NetBSD/SPARC 1.0 and SunOS
4.1.3.  I was surprised at the results:

"isolar"	is a SPARCstation-1 clone (20 MHz) with 24 Mb of RAM; gcc 2.6.2
"netbsd4me"	is a SPARCstation IPC     (25 MHz) with 36 Mb of RAM; gcc 2.4.5

isolar:1:475 [/usr/X11R6/src/xc] % uname -a ; egrep PST world.log
SunOS isolar 4.1.3 2 sun4c
Sun Nov 27 14:38:50 PST 1994
Sun Nov 27 22:31:17 PST 1994

Total elapsed time: less than 8 hours

netbsd4me:1:48 [/usr/X11R6/src/xc] % uname -a ; egrep PST world.log
NetBSD netbsd4me 1.0 NetBSD 1.0 (SPARC_SCSI3) #0: Mon Oct 24 15:23:27 MDT 1994
     deraadt@croc:/croc.usera/rel/src/sys/arch/sparc/compile/SPARC_SCSI3 sparc
Thu Nov 24 20:51:06 PST 1994
Fri Nov 25 07:40:29 PST 1994

Total elapsed time: almost 10 hours and 50 minutes

To make matters worse, I had the SunOS build make both Fresco and the X
Keyboard Extension (XKB) to boot, which the NetBSD system didn't do; *and* the
SunOS machine ran a good healthy "du" down my entire /var/spool/news tree
during this time.

Modulo other things running on each machine, I certainly expected that an IPC
with 36 Mb should handily whup a poor old SS1 with 24 Mb in this race ...

Short of inserting "date" commands before each subdirectory "make" command, I
don't have any explanation for why the NetBSD IPC came in so far behind.

	- Greg