THE­BRAIN THATDRIVESIT HOME

NVIDIA's Jet­son TX1 to Power New Gen­er­a­tion of Au­ton­o­mous Ma­chines

HWM (Malaysia) - - TELEPORT -

132

HWM

|

DE­CEM­BER 2015 NVIDIA's en­deavor into the realm of ma­chine learn­ing and the ad­vance­ment of Deep Neu­ral Net­work (a.k.a. Deep Learn­ing) has so­lid­i­fied its place in the realm of sci­ence and ar­ti­fi­cial in­tel­li­gence. That's a huge feat, es­pe­cially when you con­sider the com­pany's main­stream rep­u­ta­tion as a provider of high-end, gam­ing-grade graph­ics cards and work­sta­tion-grade graph­ics so­lu­tions.

In ad­di­tion to that, the com­pany also has a stake in the de­vel­op­ment of self-learn­ing au­ton­o­mous ve­hi­cle, and has been do­ing so since the in­tro­duc­tion of the Te­gra K1 mo­bile su­per­com­puter back in 2013. Fast for­ward to to­day, and the com­pany re­cently an­nounced the suc­ces­sor to the Te­gra K1, the Jet­son TX1 mod­ule.

The Jet­son TX1 is the com­pany's first em­bed­ded com­puter that is de­signed to process deep neu­ral net­works and ma­chine learn­ing with the com­pany's pro­pri­etary cuDNN (short of CUDA Deep Neu­ral Net­work). Specif­i­cally, we're talk­ing about a mo­bile su­per­com­puter mod­ule that is able to per­form com­plex tasks like im­age recog­ni­tion, con­ver­sa­tional speech pro­cess­ing, and real-time nav­i­ga­tion via a crowded room.

Spec­i­fi­ca­tion-wise, the Jet­son TX1 is based on NVIDIA's Maxwell GPU ar­chi­tec­ture, which is made up of 256 cores that has a per­for­mance of 1 Ter­aflops. Sup­port­ing is a 64-bit ARM A57 CPU and 4GB of LPDDR4 RAM with speeds of up to 25.6 gi­ga­bits per se­cond.

The Jet­son TX1 De­vel­oper Kit is al­ready avail­able and can be pur­chased for US$599 (ap­prox. RM2,616), while the mod­ule it­self will be avail­able to con­sumers some­time in early 2016 at a price point of US$299 (ap­prox. RM1,306).

Meet the Jet­son TX1 mod­ule, NVIDIA's new mo­bile su­per­com­puter.

Newspapers in English

Newspapers from Malaysia

© PressReader. All rights reserved.