MindModeling Updates Dec 02 2015

Message boards : News : MindModeling Updates Dec 02 2015

Author	Message
Brandon Project administrator Project developer Project tester Send message Joined: 5 Jan 15 Posts: 267 Credit: 1,456,117 RAC: 0	Message 3585 - Posted: 2 Dec 2015, 16:37:20 UTC Last modified: 2 Dec 2015, 16:37:20 UTC
	Hello Everyone and Happy Holidays, We just finished upgrading our Server Side BOINC code as well as a large amount of database upgrades, and work is now flowing again. This upgrade will now allow us to push out new apps more easily, and with the new scheduler we should see some possible efficiency gains due to the inclusion of OpenMP. We have also released a new ACT-R Lisp application running the latest version of CCL lisp. This release also includes a number of updates to the application that should help resolve many of the processing errors we have been seeing recently. We have more updates on the way, thank you all for your support, and happy crunching! Brandon
	ID: 3585 · Rating: 0 · rate: / Reply Quote

Coleslaw Send message Joined: 31 Oct 08 Posts: 29 Credit: 1,127,398 RAC: 0	Message 3586 - Posted: 4 Dec 2015, 0:42:31 UTC - in response to Message 3585. Last modified: 4 Dec 2015, 0:42:31 UTC
	This upgrade will now allow us to push out new apps more easily, and with the new scheduler we should see some possible efficiency gains due to the inclusion of OpenMP. Does this mean the possibility of GPU applications? ____________
	ID: 3586 · Rating: 0 · rate: / Reply Quote

Brandon Project administrator Project developer Project tester Send message Joined: 5 Jan 15 Posts: 267 Credit: 1,456,117 RAC: 0	Message 3600 - Posted: 15 Dec 2015, 15:17:40 UTC Last modified: 15 Dec 2015, 15:18:24 UTC
	We have considered building a gpu application in the past, however none of our modelers are currently working with a cognitive architecture that can take advantage of a gpu application. If this changes I will definitely let you guys know and we will be very happy to add a gpu application to our project.
	ID: 3600 · Rating: 0 · rate: / Reply Quote

exalpha Send message Joined: 23 Mar 14 Posts: 23 Credit: 505,507 RAC: 0	Message 3601 - Posted: 18 Dec 2015, 23:29:19 UTC - in response to Message 3600. Last modified: 18 Dec 2015, 23:29:19 UTC
	A GPU is designed on extreme parallel processing. I would have thought the brain is very parallel in processing. Are the cognitive models corresponding with parallel processing?
	ID: 3601 · Rating: 0 · rate: / Reply Quote

The Physician Send message Joined: 31 Dec 15 Posts: 1 Credit: 0 RAC: 0	Message 3605 - Posted: 31 Dec 2015, 11:11:38 UTC Last modified: 31 Dec 2015, 11:11:38 UTC
	Hello, I have just joined the project. I am glad that it has been recently updated. Kindly, I have a question: 1- Is BOINC compatible with El Capitan (Mac OSX) 2- Does Mind Modelling need GPU? Mine is Intel 4000 which is not quite supported by Boinc. Thanks! Waiting for your reply :) ____________ Pay a visit to my two projects: PPN and IGCSE WORLD!. Thanks!
	ID: 3605 · Rating: 0 · rate: / Reply Quote

exalpha Send message Joined: 23 Mar 14 Posts: 23 Credit: 505,507 RAC: 0	Message 3606 - Posted: 31 Dec 2015, 23:30:23 UTC - in response to Message 3601. Last modified: 31 Dec 2015, 23:30:23 UTC
	A GPU is designed on extreme parallel processing. I would have thought the brain is very parallel in processing. Are the cognitive models corresponding with parallel processing? What I mean is the gpu has been used in at least image object detection, classifying and natural language processing. http://devblogs.nvidia.com/parallelforall/deep-learning-for-computer-vision-with-matlab-and-cudnn/ So at least it can be applied in these areas. The brain is believed to have a similar structure from one part to another. It does not matter what the function of that area is. It just matters what it is connected to that determines the function. So this may mean the gpu could be universal for most brain functions.
	ID: 3606 · Rating: 0 · rate: / Reply Quote

Coleslaw Send message Joined: 31 Oct 08 Posts: 29 Credit: 1,127,398 RAC: 0	Message 3607 - Posted: 5 Jan 2016, 14:17:49 UTC - in response to Message 3605. Last modified: 5 Jan 2016, 14:17:49 UTC
	Hello, I have just joined the project. I am glad that it has been recently updated. Kindly, I have a question: 1- Is BOINC compatible with El Capitan (Mac OSX) 2- Does Mind Modelling need GPU? Mine is Intel 4000 which is not quite supported by Boinc. Thanks! Waiting for your reply :) 1. Yes 2a. Look at Brandon's post above... 2b. Have you tried running your Intel 4000 GPU at SETI, Collatz, Gerasim or Einstein? ____________
	ID: 3607 · Rating: 0 · rate: / Reply Quote

Brandon Project administrator Project developer Project tester Send message Joined: 5 Jan 15 Posts: 267 Credit: 1,456,117 RAC: 0	Message 3608 - Posted: 5 Jan 2016, 15:31:45 UTC Last modified: 5 Jan 2016, 15:35:00 UTC
	Hello The Physician and Welcome!!! To answer your questions, I believe BOINC should work with El Capatain, from what I have read in the forums on Berkeley's website, as long as you install the latest version of BOINC. Here is a link to someone who had an issue with it and a posted responce: https://boinc.berkeley.edu/dev/forum_thread.php?id=10566 To answer your second question, Coleslaw is correct but I will answer anyway, at this time we do not support a gpu application, but if that changes in the future I will let everyone know. Happy New Year, hope this helps, and Happy Crunching! Brandon
	ID: 3608 · Rating: 0 · rate: / Reply Quote

exalpha Send message Joined: 23 Mar 14 Posts: 23 Credit: 505,507 RAC: 0	Message 3612 - Posted: 9 Jan 2016, 5:33:52 UTC - in response to Message 3606. Last modified: 9 Jan 2016, 5:33:52 UTC
	To answer myself. As it has been already stated none of the cognitive models are compatible with gpu processing ATM. This may or probably is synonymous with parallel processing. So I may have a wrong understanding of MindModeling, in that it is a search for most likely cognitive models. Not actually simulating cognitive models. If it is iterative search, each step depending on previous results then it cannot be paralell. I don't know of answering myself this way is wrong or correct.
	ID: 3612 · Rating: 0 · rate: / Reply Quote

Brandon Project administrator Project developer Project tester Send message Joined: 5 Jan 15 Posts: 267 Credit: 1,456,117 RAC: 0	Message 3613 - Posted: 11 Jan 2016, 23:02:01 UTC Last modified: 11 Jan 2016, 23:02:01 UTC
	Hello exalpha, Your initial understanding was correct we do simulate cognitive models. To be more specific, our project specializes in parallel parameter evaluations of cognitive models, meaning: we run many independent simulations under varying conditions(both for the model of the environment and the model of the human). Currently there are no cognitive architectures that support gpus. If this changes we will create a gpu application and I will definitely let everyone know. Hope this helps. Thanks for your support and happy crunching, Brandon
	ID: 3613 · Rating: 0 · rate: / Reply Quote

Ousermaatre Send message Joined: 25 Aug 12 Posts: 5 Credit: 90,467 RAC: 0	Message 3616 - Posted: 13 Jan 2016, 18:38:12 UTC - in response to Message 3613. Last modified: 13 Jan 2016, 18:38:12 UTC
	Hi Brandon, personal messaging works, because I sent you a message and I have not received answers?
	ID: 3616 · Rating: 0 · rate: / Reply Quote

Brandon Project administrator Project developer Project tester Send message Joined: 5 Jan 15 Posts: 267 Credit: 1,456,117 RAC: 0	Message 3618 - Posted: 15 Jan 2016, 13:54:28 UTC Last modified: 15 Jan 2016, 13:54:28 UTC
	Yes, it does, and I apologize. Jack or I should be getting back with you shortly.
	ID: 3618 · Rating: 0 · rate: / Reply Quote

Message boards : News : MindModeling Updates Dec 02 2015