Re: Embedded Concurrency in Ada

Re: Embedded Concurrency in Ada

[Robert Dewar]

In article <8E65D75A8synoptikdamudderfuck@news>,
  synoptik@home.com (Andrew Logue) wrote:
> Hi All,
> What is the maximum number of Ada tasks supported by the
> kernel?  If there
> is no set limit, how can I determine the "point of no return",
> where task
> management overhead consumes, say 20% of total CPU power.

Surely this is something best asked to the vendor support
organization?


Sent via Deja.com http://www.deja.com/
Before you buy.

Embedded Concurrency in Ada

[Andrew Logue]

Hi All,

I have an Ada specific question that is specific to embedded systems based 
on the Rational (verdix) VADSNet kernel.  The target platform has a single 
40 MHZ processor.  The Vads kernel provides system calls (memory 
management, hardware interrupt management, task scheduling, etc.) and 
system elaboration for a single Ada program.  I believe the task 
management/scheduling is implemented in C, but probably doesn't matter 
anyway.

What is the maximum number of Ada tasks supported by the kernel?  If there 
is no set limit, how can I determine the "point of no return", where task 
management overhead consumes, say 20% of total CPU power.  This question 
assumes that heap and code space are unlimited.  At what point does context 
switching/stack management overhead become detrimental to system 
performance, especially in a real-time system?

Obviously the hard-line answer to this question lies with system timing 
requirements, hardware constraints, etc.  But is there a way to gather some 
metrics on how much system CPU resources are "wasted" or eaten up by the 
kernel itself?  

Regards,

Andrew Logue
Systems Architect/Bug Squisher
Computing Devices Canada
andrew.logue@cdcgy.com
synoptik@home.com