[Simh] Questions on SIMH VAX

[Marc Chametzky]

I can give some input from my experience on a few of these things...

> Does anyone happen to know what kind of performance this would give?
> How many VUPS would such a setup offer?

I don't know how it would perform on your system as I'm not familiar 
with the performance characteristics of the Atom processor. On my SIMH 
setup, which is a virtualized Linux system running on an ESXi host using 
an AMD Phenom II X3 710 processor (limited to 2 GHz on a single 
processor), I get 13 VUPs (using SRI's PT_VAX.EXE).

> How well does the CPU throttleing work, especially if running something like
> the WASD webserver?

It seems to work very well for me. I seem to recall that I tweaked my 
sources to make some minor changes to the idle code from a discussion on 
this list that was a couple of years ago. While running PT_VAX (pegging 
the CPU in the VAX), the Linux host reported that the vax binary was 
consuming 99.8% of the available CPU, which is what it should do. Once 
PT_VAX finished, it immediately dropped down to about 23%.

I will note that I got better idle handling once I started running 
DECnet. I think that SIMH dealt better when there was some occasional 
activity (such as handling network packets) rather than a truly idle system.

> I assume that Volume Shadowing is supported, so that I can place disk images
> on seperate physical hard drives and turn on shadowing?

That should work just fine. There might be an issue with SIMH working 
properly if one of its virtual drive volumes becoming unavailable, 
though, so an advantage of volume shadowing (system reliability) 
probably goes away.

>  What about
> clustering with real physical VMS systems if I desire it?

I've done it in the past using SIMH VAX clustered with a hardware Alpha 
box. It worked just fine.

> How large of
> virtual disks does SIMH support?

Big. :-) Using a command like "set rqb0 rauser=N" in vax.ini, I'm able 
to use large drives including one that's 15 GB. I did need to tweak my 
sources in order to get large volumes working properly at some point in 
the past (an issue with the large file APIs), but that was some time ago 
and I don't recall whether it was still needed with the current sources.

--Marc