14 Jan
2004
14 Jan
'04
9:06 p.m.
Jay,
Congrats on getting your system back up!
I have a favor to ask. Could you give us times for Access to run these HP benchmarks. HP
was kind enough to give us times for a number of other systems, but Access was not around
when these numbers were published. I guess we should know which CPU type your Main is
too.
Were you able to get the HP2000 Access running under SIMH? It sure sucks a lot less juice
than real hardware, and that translates directly into heat. The heat is reusable in the
winter (free heat) but in the summer you have to pay a second time to get rid of the heat
(AC).
I have a number of other HIB tapes that I need to get converted and tested. Are you in a
position to convert dot-tap files back to real (reel) tapes? If so I can post things for
you and send you a URL for them. Otherwise I would have to get some scratch tapes and cut
them here and mail them.
How much disk are you running? We'll have to set up HIBE tapes that will fit on your
system. We usually run 2 simulated 2883 disks for an Access simulation (50 MEG total) but
that could be pushed up if we need it.
Keep us posted.
Mike.
BTEST
10 REM HP benchmark taken from "E" IMS. Rekeyed by MWG.
20 REM
30 REM System Time to Run
40 REM C' 2 Min. 2.8 sec.
50 REM F 1 Min. 16.5 sec.
60 REM C 2 Min. 2.5 sec.
70 REM E 1 Min. 23.6 sec.
80 REM Access Unk. Jay? Want to help out here?
90 REM
100 READ L
110 IF L<0 THEN 999
120 DATA 60000.
130 DATA -1
140 PRINT "START=";TIM(0)
150 X=0
160 X=X+1
170 IF X<L THEN 160
180 PRINT "X=";X;"STOP=";TIM(0)
190 GOTO 100
999 END
FPTS9
1 REM HP benchmark from "E" IMS. Rekeyed by MWG
2 REM
3 REM system time to run
4 REM C' 1 min. 58.2 sec.
5 REM F 1 min. 16.4 sec.
6 REM C 1 min. 58.2 sec.
7 REM E 1 min. 15.4 sec.
8 REM Access Unk. <-- Jay? want to help here?
10 DATA 5000
20 READ L
30 K=0
40 PRINT "start=";TIM(0)
50 I=0
60 J=50
70 I=I+J+100*(I-J)^2
80 I=I+I+I+I+I+I+I+I+I+I+I+I
90 J=I-J+I-J+I-J+I-J
100 I=J^(-2)+I*I*I
110 K=K+1
120 IF K<L THEN 50
130 PRINT "STOP=";TIM(0)
140 END
15 Jan
15 Jan
3:06 a.m.
You wrote...
Congrats on getting your system back up!
Thanks! It has been pure hell to be honest. Many times it seemed to be up,
only to have something fail the day after I posted it was up. But I'm pretty
sure it's up for a good while now :)
I have a favor to ask. Could you give us times for
Access to run these HP
benchmarks. HP was kind enough to give us times for a number
of other
systems, but Access was not around when these numbers were published. I
guess we should know which CPU type your Main is too.
Sure, no problem. I will try to run it tonight. My main and IOP cpu's are
2100A and 2100S respectively. However, a project is under way to reverse
engineer the Access IOP microcode and port to the E machines, so at some
point I will likely have two 21MXE cpus instead. Speaking of which, I notice
in the comments you list times for C, C', F, E, etc... these aren't really
detailed enough. For example, E will run faster on a 21MX than a 2100 mainly
due to memory access times between core and semiconductor.
Were you able to get the HP2000 Access running under
SIMH? It sure sucks a
lot less juice than real hardware, and that translates
directly into heat.
The heat is reusable in the winter (free heat) but in the summer you have to
pay a second time to get rid of the heat (AC).
*GRIN* Most true. However, I'm not working with SIMH as I'm still working on
my own HP emulator "HPEMU". Mine emulates not just a cpu, but a whole
computer room full of separate (or connected) HP systems, printers, etc. I
made huge progress on it, then got stonewalled due to work issues. I hope to
get back to it and finish it off. It's not all that far from being done if I
could just find the time to get back into it.
I have a number of other HIB tapes that I need to get
converted and
tested. Are you in a position to convert dot-tap files back to real
(reel)
tapes? If so I can post things for you and send you a URL for them.
Otherwise I would have to get some scratch tapes and cut them here and mail
them.
I currently do not have a 1/2 tape drive hooked up to a PC to cut .tap
files. However, I'm working on that - right now it's top priority. I would
hope to have everything in place for that within a week or two.
How much disk are you running? We'll have to set
up HIBE tapes that will
fit on your system. We usually run 2 simulated 2883 disks
for an Access
simulation (50 MEG total) but that could be pushed up if we need it.
I currently have one 7906D (30mb) running, but with Access you loose head 3
as it only supports 7905's. I seem to recall that works out to something
like 25mb (removable and fixed platters are different densities, and "heads"
2 and 3 are actually on the same platter). However, I have a 7900A (10mb) in
a separate rack I can easily wheel in and connect, as well as another
working 7906D in a different separate rack I can wheel in and connect. So I
currently have 30mb ballpark, but can have 40 or 60mb easily.
Keep us posted.
Sure!
Jay West
---
[This E-mail scanned for viruses by Declude Virus]
2 p.m.
New subject: HP2000/Access - benchmarks
Mike wrote....
I have a favor to ask. Could you give us times for
Access to run these HP
benchmarks. HP was kind enough to give us times for a number
of other
systems, but Access was not around when these numbers were published. I
guess we should know which CPU type your Main is too.
BTEST ran in 1 Min 37.5 sec
FPTS9 ran in 1 Min 23.1 sec
Main CPU is a 2100A w/32K core
However.... I'm wondering how they got the timings listed in the source
comments? The TIM(0) function only prints minutes. This isn't helpful. So I
used a stopwatch, and timed from the "Start" message to the "Stop"
message.
I could have changed the code to do this, but didn't want to change anything
that might effect the timing. Access supports TIM(4), but according to my
pocket guide F does not. I'm not sure if "different" code in the OS (the
TIM
function) would have any bearing on results.
Also interesting - in a stack of papers I got from my high school years ago,
there was a benchmark printout that was run on two identical systems with
the exception that one was 2100A (Forest Park Community College), the other
was 21MX-M (St. Louis University High School). The difference wasn't HUGE,
but it was pretty noticeable. I suspect the 21MX-E would be even moreso, and
plus you have the option of several different types & speeds of memory.
Hopefully we can find out!
Anywho... that's the scoop!
Jay West
7495
days inactive
7495
days old
2 comments
2 participants
participants (2)
-
jwest＠classiccmp.org -
MGemeny＠pgcps.org