7 Jun
2011
7 Jun
'11
7:22 p.m.
On 6/7/11 5:43 PM, Dave wrote:
I hope to eventually be able to run this on my
real System/370 at the
new building in PA.
Be very interested in seeing the results from this on real hardware.
When I ran Spice II under VM/370R6 I got slightly different results to
the sample print outs provided. I know it was one digit in the last
place so could just be normal floating point jitter, on the other hand
it could be something in the Hercules floating point code not adding
enough guard bytes. Any comment
It's easily many months off, but when I do bring up that machine, I'm
sure I'll inform (i.e., "brag to" ;)) everyone here.
-Dave
--
Dave McGuire
Port Charlotte, FL
7 Jun
7 Jun
7:54 p.m.
Date: Tue, 7 Jun 2011 19:22:57 -0400
From: mcguire at neurotica.com
To: hercules-os360 at yahoogroups.com
Subject: Re: [hercules-os360] fun electronics+vintage success
CC: dave.g4ugm at gmail.com; midatlanticretro at yahoogroups.com; cctalk at
classiccmp.org
On 6/7/11 5:43 PM, Dave wrote:
I built this too a couple months ago, using gcc in fortran mode.
Only problem I had was the 'time' function to measure the simulation run time.
Its not in the gcc intrinsic lib, so I just stubbed it out.
My version was 2g.6
Randy
I hope to eventually be able to run this on
my real System/370 at the
new building in PA.
Be very interested in seeing the results from this on real hardware.
When I ran Spice II under VM/370R6 I got slightly different results to
the sample print outs provided. I know it was one digit in the last
place so could just be normal floating point jitter, on the other hand
it could be something in the Hercules floating point code not adding
enough guard bytes. Any comment
It's easily many months off, but when I do bring up that machine, I'm
sure I'll inform (i.e., "brag to" ;)) everyone here.
-Dave
--
Dave McGuire
Port Charlotte, FL
9:33 p.m.
On 6/7/11 7:54 PM, Randy Dawson wrote:
I built this too a couple months ago, using gcc in
fortran mode.
Neat! Wow, I'm shocked that GNU FORTRAN could digest that ancient
FORTRAN IV code! Did you have to change much in the source?
My efforts ended up with a single JCL file containing the FORTRAN and
assembler sources and three jobsteps that result in an executable.
Kinda neat to encapsulate it all that way. (speaking from the standpoint
of a 25-year UNIX veteran, that is)
Only problem I had was the 'time' function to
measure the simulation
run time. Its not in the gcc intrinsic lib, so I just stubbed it
out.
"When in doubt, comment it out!" ;)
The time-related stuff is what choked it for me as well, initially.
The assembler assists use DIAGNOSE 00C to get the TOD clock values,
which assembles just fine of course, but very quickly results in an
ABEND when run under MVS. ;)
I hadn't really gone through the assembler sources when I first got
it built but it wouldn't run. I figured I'd need to understand the
FORTRAN<->assembler linkage stuff to debug it, so I went off in that
direction. It was then that I was able to decipher enough of the two
offending functions (SECOND and DATIME) to notice the DIAGNOSE calls,
and said, "ahh, this is for VM/CMS!" I had thought it was an MVS
program all along. I've used SPICE for years, but only under UNIX, and
never the old FORTRAN version.
My version was 2g.6
That's the one I'm working with as well. It's the last of the SPICE 2
line, as far as I'm aware.
Fun stuff!
-Dave
--
Dave McGuire
Port Charlotte, FL
4865
days inactive
4866
days old
2 comments
2 participants
participants (2)
-
mcguire＠neurotica.com -
rdawson16＠hotmail.com