Re: Clock Error

On Tue, Dec 19, 2023 at 11:02:47PM +0100, Johnny Billquist wrote:
> On 2023-12-19 15:34, Ken Wellsch wrote:
> > I have been running my own build release on simh, as I am sure others may be
> > doing, after seeing such encouraging news that the VAX port may finally be
> > self-hostable again.  The build has not yet completed but is well along:
> > 
> >    PID USERNAME    THR PRI NICE   SIZE    RES STATE    C   TIME    WCPU COMMAND
> >   6963 sim312        1 145   10   199M   150M CPU0     0 110.6H 100.00% vax
> 
> You and me both. And it's on the third day or so for me...
> 
> > I am now onto my 5th day running on the "classic" version of simh:
> > 
> >    VAX simulator V3.12-4
> > 
> > to which I added the small code fragment from open-simh for NEWBSD IDLE.
> > 
> > I did not add a network device and learned the hard way that you can't just
> > leave the default time after the install because some of the build config
> > fails when sanity checking the timestamps on files created.
> > 
> > So I set the system time from my watch (which is accurate to the second).
> > 
> > I just checked the date on the build box and compared it with another system.
> > 
> > My desktop:  Tue Dec 19 07:59:09 EST 2023
> > The Build:   Tue Dec 19 07:59:04 EST 2023
> 
> Interesting. I am running a somewhat recent simh (4.something), and it's
> *way* worse. So it would seem something was broken in the code compared to
> Bob's version. Or else it's just a question of host issues. I'm running
> under OSX...
> 
> Anyway.
> Desktop:           Tue Dec 19 23:00:23 CET 2023
> NetBSD under simh: Wed Dec 20 05:16:22 CET 2023
> 
> Time was originally pretty correct.
> 
> > I am running a cross-compiled copy of 10.99.10 with the last GCC patch applied:
> 
> Building 10.99.10 here, from a system currently on 10.99.7 (cross compiled a
> few days ago).
> 
> > NetBSD nb10sim312 10.99.10 NetBSD 10.99.10 (GENERIC) #1: Mon Dec 11 09:17:31 EST 2023  netbsd@z440:/home/netbsd/obj/sys/arch/vax/compile/GENERIC vax
> > 
> > So for what it is worth, it appears I am loosing about one second per day,
> > on a 100% loaded simh session, after nearly five days running.
> 
> So clearly what version of simh one is on, or possibly host, can make a big
> difference here. Yuck!

I borrowed a small HP desktop with a Pentium G7400 CPU - the single thread
speed is much higher than anything I own. The box is running FreeBSD 14.

I have been playing with both open-simh (4.1) and Bob's simh 3.  I just
ended up running my build on 3 rather than 4.

It probably is a crazy thought, but I am curious whether you have enabled
a network device and have it active or not.

> 
>   Johnny
> 
> -- 
> Johnny Billquist                  || "I'm on a bus
>                                   ||  on a psychedelic trip
> email: [email protected]             ||  Reading murder books
> pdp is alive!                     ||  tryin' to stay hip" - B. Idol