*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
*****
Just in case some folks on the list don't realise, Tim is the original author of the Autoquant / Autodeblur software, even though he is now no longer connected with it.
Guy
-----Original Message-----
From: Confocal Microscopy List [mailto:[log in to unmask]] On Behalf Of Tim Holmes
Sent: Wednesday, 28 March 2012 3:57 AM
To: [log in to unmask]
Subject: Re: Grfx cards for AutoQuant/AutoDeblur
*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
*****
Esteban,
We have experience in using NVIDIA Cuda cards in software development projects in other areas, like computed tomography. At least in our case, for our projects, you HAVE to use NVIDIA because they are the only ones that are CUDA compatible. CUDA is the language used to program the cards that gives them the "parallel processing" capability. At least in our case, for our projects, we automatically detect if the NVIDIA card is there and if it is CUDA compatible. Then, if so, we execute the algorithms on the cards rather than the CPU. If any other card is there, besides and NVIDIA CUDA card, then we execute the algorithms on the slower CPU. I can't speak for the Media Cybernetics product, but I would guess (not really knowing) that they must do soething along those lines. If so, you would only get the speed advantage with an NVIDIA CUDA compatible card. Other cards, like ATI Radeon, have their own languages they use fore speeding up graphics which are not CUDA. That's why they don't work for CUDA programs.
I am not working for Media Cybernetics, and I do not know if this would be their official answer from their support group. I am guessing what the answer would be based on our experience using CUDA and NVIDIA cards in other projects.
Tim Holmes, D.Sc.
CEO
Lickenbrock Technologies, LLC
-----Original Message-----
From: Confocal Microscopy List [mailto:[log in to unmask]] On Behalf Of G. Esteban Fernandez
Sent: Tuesday, March 27, 2012 11:20 AM
To: [log in to unmask]
Subject: Grfx cards for AutoQuant/AutoDeblur
*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
*****
Hi all,
I'm purchasing a new computer to be used for deconvolving confocal data with Media Cybernetics's AutoQuant/AutoDeblur software (ver.
X1.4.1 maybe upgrading to X3). I know MediaCy recommends NVIDIA graphics cards for their CUDA parallel processing ability but my computer people want to purchase a different card, ostensibly because we get a deal on the particular brand they want. The card (GIGABYTE
GV-R797D5-3GD-B) does have parallel processors but they're not branded as CUDA, I don't know enough to determine if that makes a difference; it does support openGL. I'd appreciate it if people would share their experiences running AutoDeblur with non-NVIDIA (non-CUDA) cards.
Thanks,
Esteban
-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2012.0.1913 / Virus Database: 2114/4897 - Release Date: 03/27/12
|
