[Info-vax] What choices are available to the OpenVMS Process Scheduler once all the CPUs are in use?

[Johnny Billquist]

On 2016-02-11 09:21, Fred.Zwarts wrote:
> "abrsvc"  schreef in bericht
> news:2babd44a-eb5f-4d22-b8a2-f16c75fcc737 at googlegroups.com...
>>
>> On Wednesday, February 10, 2016 at 1:40:06 PM UTC-5, Carl Friedberg
>> wrote:
>>> I would try turning off hyper-threads. I've never seen it make more
>>> than a
>>> few percent better perofrmance, and it can make things worse.
>>> Remember the
>>> VAX 782 (with 2CPUs, performance, IIRC, was of order .8 VUP. Hoff will
>>> remember better.
>>>
>>
>> I remember that well.  There was an analysis tool that would indicate
>> whether or not your particular application would benefit from the
>> additional CPU.  Please note that the 782 was an ASMP machine
>> (Asynchronous) and would only benefit compute bound applications.
>
> I remember that the A stood Asymmetric, not Asynchronous. Asymmetric,
> because I/O (and some other kernel stuff) was done by the first
> processor only. Does my memory fail?

No, you are correct. And even more, the 11/782 was more asymmetric than 
that. The second processor was totally a slave to the primary one. It 
was a compute resource. Whenever a system call, trapping, or anything at 
all happened, it was managed by the primary CPU. The secondary did 
essentially not run anything of VMS. All scheduling, all I/O, all 
process management, time keeping, you name it... All done only on the 
primary. The primary could kick off the secondary to run user code for a 
specific process, but as soon as the process needed any attention it was 
back on the primary.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol