Re: Ada Shootout program for K-Nucleotide (patches)

[jonathan <johnscpg@googlemail.com>]

Now that I've measured IO overheard, I can
say more about the efficiency of the multitasking
knuckeotide.
The optimal result would be about 11 sec on
4 cores. The best I measure is about 13.2 seconds -
hard to improve on.

Here is the best timing I get on 4 cores. (I put
in the optimization I described previously;
results are a bit improved over yesterday. The
actual times fluctuate wildly. This is the
lower bound.)

real    0m13.174s
user    0m32.294s
sys     0m0.420s

3.7 sec go to IO, which won't parallelize.
Those 3.7 sec can't be reduced by adding more tasks.

The remaining 28.6 sec we hope to see reduced to
28.6 / 4 = 7.15 sec on 4 cores in the perfect limit.
That means the best result we can hope for
is a running time of 3.7 + 7.15 = 10.85 sec.

Jonathan