Re: Task Discriminants & Arrays

[mheaney@ni.net (Matthew Heaney)]

In article <33787506.41C67EA6@spam.innocon.com>, Jeff Carter
<carter@spam.innocon.com> wrote:


>As I understand it, task discriminants were intended to eliminate the
>"serial bottleneck" that results from having to call an initialization
>entry to provide information like this to task instances. However, as
>far as I can tell, it simply complicates matters: You have to declare an
>access type designating the task type, make your array type an array of
>these access types, then allocate the tasks at runtime:

Yeah, I thought of the heap-based solution, but was trying to avoid using
an allocator.  Perhaps you just have to do that.

>The use of discriminants may be more efficient and have less of a serial
>bottleneck. I'd be interested to know if this is the case, and if so,
>how much of an improvement one gains with various compilers.

I think it's true that more entries mean more overhead, because the
run-time executive must monitor activity on each entry queue.  In the case
of an Initialization entry, it means a loss of efficiency because that
entry will never get called more than once.

--------------------------------------------------------------------
Matthew Heaney
Software Development Consultant
<mailto:matthew_heaney@acm.org>
(818) 985-1271