Difference: JaysTermTwoJournal (7 vs. 8)

Revision 82011-01-18 - jayzhang

Line: 1 to 1
 
META TOPICPARENT name="NGSAlignerProject"

01/05/2011

Started reading more in-depth into CUDA. Made some notes (I'll post them later).
Line: 40 to 40
 To do:
  • GPU stuff.
Added:
>
>

01/18/2011

I started playing around with CUDA. I've decided to start a new, empty branch and write the CUDA stuff separately from the main NGSA library so I don't have to worry about getting CMake to compile CUDA stuff. So far, I've just been reading up on standard techniques, and I've also implemented a simple hamming distance "aligner". The aligner pretty much just copies the read and reference string into constant global memory, then threads compare the characters at each point (one thread per character) and assign a '1' or '0' based on the result of the comparison. The second phase just adds all the comparison results together to get the total number of mismatches. The addition uses a simple parallel reduction algorithm I found on these slides.

To do:

  • Start thinking about Smith-Waterman on the GPU?
  -- Main.jayzhang - 05 Jan 2011
 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2025 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback