[Info-vax] clock problems with OpenVMS x86 on VirtualBox

[Jan-Erik Söderholm]

Den 2023-05-08 kl. 00:44, skrev David Turner:
> Like I have always suspected.
> I think there is more of an emulator in the OpenVMS x86 code than has been 
> admitted.
> The faster the CPU on the virtualbox the faster your "native X86" code is 
> gonna run
> Anyone care to elaborate or prove this?


I should not take the writing below as anything of a value.
Just the line saying "it takes about >> 15 seconds for MONITOR
SYSTEM to initiate its display" says that something is really
wrong with that environment. On my laptop with VirtualBox,
MONITOR starts in around 1 sec. I cannot see any difference
between my VirtualBox environment and a real AlphaServer DS20e.

What I can see, is a slight lag in the clock. It looses approx
1 second in 2.5 minutes. But it is not consitent, a few minutes
later, my VirtualBox VMS time was back i phase with our real
production DS20e again. And over a longer timespan, it seems to
keep the time fine.

When looking at the MON screen, I can see that the running clock
at the top right, sometimes jumps over a second and advances 2
secons "in one second", so to speak.

I have not done any NNTP setup.
I have not done that command regarding the clock in the docs.

Regards,
Jan-Erik.



> 
> 
> DT
> 
> 
> On 5/6/2023 12:49 PM, Craig A. Berry wrote:
>>
>> I am running OpenVMS x86_64 E9.2-1 under VirtualBox 7.0.8 on macOS
>> Ventura 13.3.1 (a).  The host is a 2019 MacBook Pro with 2.3 GHz 8-Core
>> Intel Core i9.  The clock isn't working right, most easily seen by the
>> fact that the following two commands were typed exactly one minute apart:
>>
>> $ sh time
>>    5-MAY-2023 19:07:35
>> $ sh time
>>    5-MAY-2023 19:07:45
>>
>> So the system advances its clock about 10 seconds for every 60 seconds
>> of actual time.
>>
>> Another way of illustrating the problem is by running the vups.com from
>> <https://emuvm.com/download/vups-com-benchmark/>, which says:
>>
>>  Approximate System VUPs Rating : 10430.2 ( min: 950.4 max: 55125.0 )
>>
>> Divide that number by 6 and it might almost be credible based on numbers
>> other people have been posting, though a factor of 10-15 would be more
>> believable.  The system is *not* in fact fast at all -- it takes about
>> 15 seconds for MONITOR SYSTEM to initiate its display when nothing at
>> all is running.
>>
>> I have not seen this anywhere in the VSI documentation, but this blog post:
>>
>> <https://raymii.org/s/blog/OpenVMS_9.2_for_x86_Getting_Started.html>
>>
>> says it's necessary to set the clock characteristics of the virtual
>> machine like so:
>>
>> $ vboxmanage modifyvm <vmname> --hpet on
>>
>> I have done so but observe no difference (everything posted above was
>> after running this command and restarting VirtualBox).  Supposedly this
>> setting "Enables . . . a High Precision Event Timer (HPET) which can
>> replace the legacy system timers," whatever that means. I guess one
>> possibility is that the command, while it returned no errors, also
>> didn't take effect for some reason.  I cannot tell from the following
>> output whether the non-default setting is in effect:
>>
>> $ vboxmanage debugvm <vmname> info hpet
>> HPET status:
>>  config=0000000000000003     isr=0000000000000000
>>  offset=fffffff020435102 counter=0000000000000000 frequency=69841279 fs
>>  legacy-mode=on   timer-count=4
>> Timers:
>>  0: comparator=0000001022973a28 accumulator=0000000000022f4d (9999944 ns)
>>         config=ffffffff0000003c irq=0 en per cap_per cap_64
>>  1: comparator=00000000ffffffff accumulator=0000000000000000 (0 ns)
>>         config=ffffffff00000000 irq=8
>>  2: comparator=00000000ffffffff accumulator=0000000000000000 (0 ns)
>>         config=ffffffff00000000 irq=0
>>  3: comparator=00000000ffffffff accumulator=0000000000000000 (0 ns)
>>         config=ffffffff00000000 irq=0
>>
>>
>> Has anyone sen anything like this or has any ideas on how to debug/fix
>> it?  The only next step I can think of is to try VMWare Player and see
>> if it works better than VirtualBox.
>>
>