[Info-vax] Kittson question

[Stanley F. Quayle]

On Wednesday, December 31, 2014 6:06:51 PM UTC-5, JF Mezei wrote:
> But isn't the system ID and system model stored in ROM (which the
> emulator would emulate, right ?)

Yes. My product intercepts the SYS$GETSYI calls, which are the documented way to get system values. Sure, I could hack the ROM file which comes with the product, but that might make the virtual VAX not boot at all. The right bytes do move with different VAX models, as you said.

> in a file which is I assume the emulator loads
> into what would be ROM memory on the real hardware.

Yep.

> But would the emulated instance ever see DSSI devices  ? If not, when it
> boots, it would not configure any DSSI devices and thus not need any
> DSSI support. Right ?

The only way for the emulated VAX to see physical DSSI devices is for a physical VAX to MSCP-serve them.

> Or does your emulator actually simulate DSSI disks/bus ? If so, what is
> the advantage of this versus simulating standard RA disk ?

Some people actually want to completely replace their system without changing anything (duh). So, you can make a DSSI cluster -- as I said previously, the cluster and disk access is done using IP across the host (Windows, etc.) network interface.

CHARON-VAX can emulate any model of disk, but that doesn't make it DSSI by default. Your configuration file would have to specify how the virtual disk is accessed.

> If you simulate DSSI bus, how would the underlying Windows instance
> carry DSSI traffic to a real DSSI bus for real VAX hardsware sitting
> next to the emulated instance ?

That's why I said that you can't connect to physical hardware. There's nothing on a physical DSSI device that expects to get IP traffic from a host system. But unless you need to keep physical DSSI hardware (why?), that's not a problem.

> how do the Windows hosts exchange DSSI traffic, DSSI over IP ?

Yep.

> Is there any advantage of simulating DSSI versus letting the VMS
> instances just plain old SCS over Ethernet ? (since I assume the
> simulated DSSI would travel on Ethernet between Windows instances ?)

It's just a question of keeping the virtual implementation the same as the physical one was. It's actually kind of a pain to set up. Clustering over Ethernet is a snap. I have 6 nodes right here -- a mix of CHARON-VAX, CHARON-VAX, VAXstations, DPWS 500au, Integrity rx2600. It all plays together as a NI cluster.

> So the emulator really simulates a 24 bit path to memory controller by
> doing an AND of a 32 (or 64) bit value to only keep the first 24 bits
> FOR EACH MEMORY ACCESS ? Isn't that a lot of overhead ?

It works just like the physical machine being emulated, which is the point. Translating each VAX instruction is the definition of overhead. Fortunately, that PC on your desk is thousands of times faster that your VAX. Your emulated MicroVAX II, regardless of how "inefficient" you might think this is, is many times faster that the physical one.

I tell people, "If you like your VAX, you'll love your virtual VAX. If you hate your VAX, you'll hate your virtual VAX just as much, but faster."