Re: task-safe hash table?

[tmoran@acm.org]

> >What about a window in which each task makes, say, n searches. n is the
> >window size. All misses are postponed till the window end. At that point
> ...
>   Perhaps a windowing or batching approach such as you suggest would
> distribute the Protected overhead over more useful work and make the
> concurrent approach faster.  I'll look into that this weekend.
   Dual tasking takes 2/3 as long as sequential.
   The k-nucleotide strings start half the time with 'A' or 'C' and half
the time with 'G' or 'T'.  So I tried using one hash table for the former
and one for the latter.  Then there is no synchronization needed (till
printing the results).  Running with the main program handling A+C and a
second task concurrently handling G+T drives CPU usage up to 96% and takes
2/3 the wall clock time, as compared to doing A+C followed sequentially by
G+T (which only takes CPU usage up to 50%, or all of one core on this
dual-core Pentium).
  So Ada multitasking on a multi-core CPU gives substantial improvements
for another of the shootout benchmarks (which I believe, unfortunately,
are officially measured on obsolete single-core CPUs).