REGISTER | LOGIN
Breaking News
Blog

Architectures Battle for Deep Learning

Linley Gwennap
10/31/2017 03:41 PM EDT

 3 comments   post a comment
View Comments: Newest First | Oldest First | Threaded View
Misha177
User Rank
Rookie
What about FPAA?
Misha177   11/7/2017 10:24:20 AM
NO RATINGS
Can be used differentional equations for DNN. And one of solution of differentional equations is FPAA - Field programming analog array What about it? Interesting.  

Kiers
User Rank
Rookie
Suits the cloud paradigm
Kiers   11/4/2017 10:38:57 PM
NO RATINGS
Interesting that google have "released" it's Tensor silicon ONLY for it's cloud.  SO here we have a "chipmaker" RENTING the chip-as-a-service! They don't want to SELL it in chip form.  All this suits the cloud providers who want to lock in customers as tenants not as landlords! How long before a Texas Instruments or ARM or Samsung, or Intel jump in and SELL silicon for those customers who seek to be landlords?

Don Herres
User Rank
Author
time to market
Don Herres   11/2/2017 12:26:05 PM
NO RATINGS
It seems you could add a scale from right to left on the graph of time to market.

Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed