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

[jonathan <johnscpg@googlemail.com>]

On Aug 2, 5:00 pm, jonathan <johns...@googlemail.com> wrote:
> The new version worked like a charm.  I checked the web page, and
> the output is correct.  Your latest version gave an additional
> 70% speedup on my machine, so:
>
>      time ./knucleotide < fasta25.dat    # 25  Meg data file
>
> ran in 4.2 seconds (fastest result), and
>
>      time ./knucleotide < fasta250.dat   # 250 Meg data file
>
> ran in  43 seconds.  I used
>
>    gnatmake -gnatnp -O3 -march=native knucleotide
>

And I neglected to say that the timings above are wallclock times.
The several tasks ran in parallel on the 4 cores - the actual CPU
time accumulated over the several cores was than twice the 43 seconds.
If this is not allowed by the rules (single core supposedly) then
old version is faster, because it has no tasking overhead.  On the
other hand, some of the other programs seem to be threaded.

j.