Re[2]: [Am-info] O/S 2 support - slightly off topic (my apologies)

[Erick Andrews]

On Sat, 16 Mar 2002 22:09:18 +0000, John Poltorak wrote:

>On Sat, Mar 16, 2002 at 03:35:08PM -0600, Eric M. Hopper wrote:

[...]

>> 
>> No, I mean cooperative multithreading.  As I said, almost nobody thinks
>> of cooperative multithreading as being that.  
>
>Well the concept is lost on me. I have no idea what it means at all.

Actually, I've never heard the term before either, but maybe it's common
in some circles.  I think I understand the meaning, though I'd prefer 
terms associated with releasing/blocking of mutex semaphores and the like.

>
>> Cooperative vs. preemptive
>> are CPU scheduling techniques.  Multitasking vs. multithreading are
>> about sharing memory between tasks, and not CPU scheduling at all.
>
>OS/2 is both multitasking and multithreading, where every task consists of 
>one or more threads. Those tasks share CPU not memory AFAIAA. The only 
>time that memory is shared is when it is overcommitted and a task is 
>swapped out to disk.
>

My understanding, too.  The "business end of the wedge" here is the serialization
of digital CPUs,  digital feedback, not storage.  The CPU is the guy who manages
the storage...lest we get into memory subsystems like some found in array processors, 
streaming graphic satellite data...which I don't think we should.  (That might beg
of Turing's concepts of "O-machines" or analogue devices that implement
Threshold/Voting Logic <g>.)

-- 
Erick Andrews