[Bioclusters] Re: [Biodevelopers] Altivec enhanced codes?

Chris Dwan (CCGB) bioclusters@bioinformatics.org
Wed, 5 Nov 2003 20:30:19 -0600 (CST)


A bit more information on the Sencel / Altivec question (also known as
"everything I know"):

- The enhanced -> non-enhanced code in question is the distribution of
  FASTA, from Bill Pearson at UVA:
  ftp://ftp.virginia.edu/pub/fasta/fasta34t23b1.shar.Z

- The comment that tipped me off was in dropgsw.c:
  ------------------------------------------------
  /* ...
     25-Feb-2003 - modified to support Altivec parallel Smith-Waterman

     22-Sep-2003 - removed Altivec support at request of Sencel lawyers
  */

- "dropgsw.h" has a bunch of ifdef'ed SW_ALTIVEC stuff, and makes
  reference to "altivec.c", which is not in the distribution.

- In response to a query I sent, I received a very polite note from
  Sencel (from Torbjrn Rognes in fact, holder of the patent in question)
  indicating that while they do not presently have a product which makes
  use of the altivec enhancements, they plan to relase one soon.  He says
  that the intent is that it be free for academic use.

  They also plan to release (sell) a library of altivec using functions so
  that developers can embed this Sencel product (a Smith-Waterman
  implemented on SIMD) in their own codes.

- I have dropped a note to Dr. Pearson (who I know not at all)
  asking if he's willing to share any more information on what
  prompted the removal of the code.

- I have sent a note to some of the life sciences folks at Apple asking
  their opinion.

That's what I know.

-C