> -----Original Message----- > From: John Rose via AGI [mailto:[email protected]] > Sent: Saturday, November 29, 2014 9:36 AM > To: AGI > Subject: RE: [agi] The role of prediction [was What's preventing me...] > > > -----Original Message----- > > From: Ben Goertzel via AGI [mailto:[email protected]] > > Sent: Friday, November 28, 2014 10:28 PM > > > > Speeding that up by 10% wouldn't make much difference to us. > > Speeding that up by 100x would make a tremendous difference to us, > > because it would then let us perform MOSES-parameter-space search in a > > more automated way, rather than "largely by hand.".... Speeding that > > up by 10x would make a moderate difference to us.... > > > > FWIW I've been testing some code which is basically multicomponent > reaction-diffusion automata running on graphs (totalistic CA) and you can get > anywhere from 10x to 20x right away by implementing SSE and OpenCL or > OpenML. Not sure if MOSES uses that or has the execution model with that > type of concurrency... > > John >
Oops misstyped here I meant OpenMP (not OpenML) as in: https://gcc.gnu.org/wiki/openmp There is an OpenML but that's something else :) OpenMP can have somewhat better performance than OpenCL but OpenCL *seems* to be more widely supported. John ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
