[Info-vax] Real time performance, was: Re: VSI: "Official 8.4-1H1 Launch"

[clairgrant71 at gmail.com]

On Monday, June 8, 2015 at 4:22:42 AM UTC-4, Jan-Erik Soderholm wrote:
> JF Mezei skrev den 2015-06-08 07:00:
> > On 15-06-07 23:47, Robert A. Brooks wrote:
> >
> >> It's not as clear-cut as you think . . .
> >
> >
> > Seemed clear to me (and I am not kmowm to understand things quickly)
> >
> > I asked how the engineers get the initial code tested, some x86
> > emulators, running it on x86 servers etc.
> >
> > The asnwer I got was that they would be doing it (the early boot
> > sequences) on a VM (the KVM discussion ensued).
> 
> I read it as that they would *also* use the KVM environment and
> the KVM tools in the porting process. But not exclusively.
> 
> I don't know how well this KVM emulates the real hardware.
> If it is exaclty the same system image that boots on both
> real hardvare and the KVM envionment, and so on.
> 
> >
> > I don't know why there is so much discussion.
> 
> *You* should know, it's mainly your discussion. :-)
> 
> > If the answer was clear to
> > me, it should have been extremely clear to everyone else.
> >
> 
> You usualy just make up the parts you didn't understand.
> For those that do understand it was clear that the picture
> is way more complex then what you seems to think.

Yes, "also" is what I was trying to convey, sorry if I did not make that clear. We will be pursuing real HW and KVM simultaneously because we want VMS to support both. Having a totally SW environment like KVM for debugging is appealing but getting VMS to run as a KVM guest will likely not be simple. This has nothing to do with the CPU architecture. Modern day hypervisors expect their guests to be linux or Windows. VMS will almost certainly behave in a few ways they are not expecting and we will have to figure those out as we go along. For example, for many years now VMS has run as a guest on HP-UX's HPVM. That took a bit of work but we got there. I expect KVM or any of the others to be a similar effort.