About a numenta type network on an fpga, if it were just local proximal dendrites (ones closeby) you could probably route them directly a->b. And if the network were to "evolve" you could recompile the network and update the FPGA to directly reflect it and take advantage of the programmability an ASIC doesnt have, even tho that would take a fair bit of time normally, maybe up to 10 minutes? if u used the average software they put out (Xilinx and Altera). They put out REALLY big units, if you want to go for 10 million Look up Tables (https://www.xilinx.com/support/documentation/data_sheets/ds890-ultrascale-overview.pdf) (https://au.element14.com/xilinx/ek-u1-vcu129-g/eval-kit-virtex-ultrascale-fpga/dp/3580744?gclid=CjwKCAjwjdOIBhA_EiwAHz8xmyAS8L4gj6Dsy5c8Cifw6FDIgfA_5pUkqlal9ucBG6gSjNRUTpM0CBoCX8cQAvD_BwE&mckv=_dc|pcrid|501152839127|pkw||pmt||slid||product|3580744|pgrid|123988903131|ptaid|pla-293946777986|&CMP=KNC-GAU-GEN-SMART-SHOPPING) -$ 30,000 :)
That thing would kick alot of butt, but it is very expensive. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T1dc2a5f2d1d491d4-M5d14541596f7c494314f25de Delivery options: https://agi.topicbox.com/groups/agi/subscription
