Re: Need some light on using Ada or not

[Brian Drummond <brian_drummond@btconnect.com>]

On Sun, 20 Feb 2011 07:45:33 -0800 (PST), jonathan <johnscpg@googlemail.com>
wrote:

>On Feb 20, 2:34�pm, Brian Drummond <brian_drumm...@btconnect.com>
>wrote:
>
>> I removed virtually the entire body of the program into a single task.
>> This change alone doubles the "CPU" time. There appears to be a 100% penalty
>> associated simply with running the original program from within a second task.
>
>
>I noticed that too.  As soon as I declare a task, (you
>don't have to use it for anything) then run-time doubles.
>
>So as far as I can tell, anything with gnat tasks (p-threads)
>has this behavior. The C program that uses p-threads (C#5)
>has the same behavior.

Thanks for the confirmation. I'll have to do some reading about Annex E, and
PolyOrb...

Meanwhile, slightly better news...

on my 4-core machine (AMD Phenom 955X4) the original single-CPU code runs in 30
seconds, and moving to a single task increases runtime to about 50 seconds.

Using 4 tasks, the CPU time remains about 50 seconds, but the elapsed time is
reduced to 18 seconds ( 9 tasks give 400% CPU usage for 12 seconds, then 100%
for a further 6). Assuming this scales to the reference machine (single-CPU 37
seconds, but Intel rather than AMD) would give a runtime around 23s, and move
Ada up from 16th to 10th place.  

Or if I can employ finer grained tasking to balance the load better... 
well, 16 seconds would give us 6th place, 15s would give 4th place.

- Brian