*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopyPost images on
http://www.imgur.com and include the link in your posting.
*****
Dear Confocal Listserv,
I had previously posted online at
http://works.bepress.com/gmcnamara/56/
and mentioned to the listserv that Marc Bruce and Manish Butte's NVidia
GPU Deconvolution
http://www.opticsinfobase.org/vjbo/fulltext.cfm?uri=oe-21-4-4766&id=249375http://tcell.stanford.edu/software.htmlhad a bug. Marc has developed a solution, described the issues, and sent
me results, that I have posted at
http://works.bepress.com/gmcnamara/62/I've noted in the past that B&B's GPU deconvolution on an NVidia TITAN
card is a lot faster than other deconvolution software I have available.
Some listservites might find of interest the comparison Ms. Vinita Popat
did in our (Cooper) lab in 2013, that I recently posted at
http://works.bepress.com/gmcnamara/57/Sincerely,
George
p.s. with respect to the Nobel Prize in Chemistry (Physics was just
awarded for blue LEDs leading to solid state blue light lasers), my
thanks to Stephen Baer for his post on his 1999 patent (filed in 1994).
Stephen's patent is available at
http://www.google.com/patents/US5866911On 10/10/2014 6:17 PM, McNamara,George T wrote:
>
>
> *From:* Marc Bruce
> *Sent:* Friday, October 10, 2014 6:01 PM
> *To:* McNamara,George T
> ***Subject:* Re: Hi Marc, thanks, bug fixed dataset looks good ... RE:
> Fixed bug
>
> Hi George,
>
> I can clarify a bit: NVIDIA's FFT libraries have multiple algorithms
> tailored to different input dimensions, and the FFT boundary effects
> introduced by these different algorithms appeared to various extents
> depending upon the input image. This is likely why you saw a bigger
> problem with 32 Z slices than 36 slices; the 36 slices gets processed
> by both 2- and 3-radix algorithms. The fix I've implemented
> specifically deals with the edges of the image in such a way that the
> boundary effects are at least significantly reduced if not eliminated
> completely.
>
> Marc
>
> --
>
> *MARC BRUCE, PHD*
>
> /CTO and Co-Founder/
>
> *MICROVOLUTION, LLC*
>
> M: 650-427-9122
>
--
George McNamara, Ph.D.
Single Cells Analyst
L.J.N. Cooper Lab
University of Texas M.D. Anderson Cancer Center
Houston, TX 77054
Tattletales
http://works.bepress.com/gmcnamara/42
Locked
