[Simh] Deathrow OpenVMS SIMH Node.. Hosted by OpenVMS?
Bill Korendyk
bill at korendyk.net
Wed Mar 11 17:41:59 EDT 2009
Hi Dave,
Again, thanks for the response.
On Wed, 11 Mar 2009 12:58:43 -0500 "Hittner, David T (IT)" <david.hittner at ngc.com> wrote:
> I'm not ruling out OpenVMS as a SIMH VAX platform. Far from it. Use it
> if you want to.
>
> Simulation of another CPU eats an astonishing number of host CPU cycles
> to perform the simulation work.
>
> I'm trying to set your (and other's) expectation level, based on my
> simulation experiences. Context is important here - maybe it's because I
> already have some real VAXes, but I don't WANT another "smaller"
> MicroVAX in my cluster. I already have a physical MicroVAX 4000/VLC in
> my cluster, and to be frank, it's performance sucks at 5.8 VUPS,
> particularly when compared to my physical MicroVAX 4000-500 at 22 VUPS.
Sadly, my expectations are very well grounded, which has me
accepting many of the shortcomings of both original VAXen and
of simulations. But that's another discussion.
I just didn't want others to be left with the impression that
OpenVMS would not make an "acceptable" SimH host. That assumes
I am the only one unable to get networking to work. :-}
> When I ran a SIMH VAX on my Alpha PWS 500au (which runs at 155 VUPS), I
> get 1.1 VUPS on the SIMH VAX. For comparison, when I ran the SIMH VAX on
> my 3.0 GHz PC, it simulated 1.89 VUPS.
I'm assuming a transposition here, and it is really 18.9 VUPs.
> If you have newer iron like the multi-issue Alpha ESxx/GSxx servers (or
> even a DS10!!) just sitting around that you can run simulations on, go
> for it. If you are looking at _purchasing_ a low-cost platform for home
> use to run VAX software, I'm merely suggesting that you might want to
> take the cheaper path to Hobbyist VAX performance, which is a fast PC
> running SIMH VAX - or many times, just a used VAX.
I have too many old VAXen already. I'm just trying to avoid adding
yet another box to my "home data centre" aka "my electric furnace."
> I'd be happy to look at your Alpha-based SIMH VAX VUPS numbers for
> comparison with mine; maybe my measuring and estimating sucks. :-))
Yes. I'm curious as to how you measured, because...
> FWIW, my low-end HP Integrity rx2620 [at work] running OpenVMS clocks in
> at 533 VUPS. I've never tried running SIMH it, but I'm sure it would be
> much better than the Alpha (at 155 VUPS); but then again, just try to
> buy an rx2620 on Ebay ... $$$. The semi-affordable, semi-supported
> rx2600 is pretty reasonable on Ebay at $300-$2500, but is _much_ slower
> than the rx2620.
I get rather different numbers. A sampling:
VUPs
SimH/VAX on AlphaServer DS10 5/466 2.0
SimH/VAX on AlphaServer DS10 6/600 2.8
VAX 4000/100 29.0
Compaq AlphaServer DS10 5/466 557.2
Compaq AlphaServer DS10 6/600 753.6
AlphaServer ES45 Model 2 1312.1
HP Integrity rx2600 (900Mhz/1.5MB) 4189.3
HP Integrity rx2600 (1.4Ghz/1.5MB) 6259.5
HP Integrity rx2620 (1.6Ghz/3.0MB) 7572.5
These numbers are from program(s) I've used since the late '70s
to compare performance of the "new" VAXen with PDP-11s. Through
the years the values were normalized to a VAX-11/780 (ie. 1 VUP)
and have compared favorably through the VAX line, and into the
early Alphas.
I'm not saying my numbers are more accurate, but it would be
interesting to see if a SimH/VAX on Integrity scales as it
does on Alpha... from my numbers, one might expect 20 VUPs
from SimH/VAX on an rx2620. Now if it wasn't so expensive...
Hmm. I have access to an rx2620, but it doesn't have compilers.
Now if I could get my hands on a SimH/VAX executable, I could
try it and see... ;-)
\bill
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Bill Korendyk Bill at korendyk.net "Just call me Bill."
--------------------------------------------------------------------
"Where am I going... and why am I in a handbasket?"
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
More information about the Simh
mailing list