abbadon420@sh.itjust.works to Funny@sh.itjust.works · 1 month ago"LLMs can do anything" except make positive cashflow.i.programmerhumor.ioimagemessage-square135linkfedilinkarrow-up11.28Karrow-down119
arrow-up11.26Karrow-down1image"LLMs can do anything" except make positive cashflow.i.programmerhumor.ioabbadon420@sh.itjust.works to Funny@sh.itjust.works · 1 month agomessage-square135linkfedilink
minus-squarepiccolo@sh.itjust.workslinkfedilinkarrow-up3·1 month agoNPUs are already a thing, specialized cpus for nerual networks. The problem is the models are just databases, and you need lots of fast memory in order to feed the NPUs data, and thats the current bottleneck.
minus-squarewholookshere@lemmy.blahaj.zonelinkfedilinkarrow-up1·1 month agoI’m not denying they still gobble those up, but those components are not what’s drawing all the power.
NPUs are already a thing, specialized cpus for nerual networks. The problem is the models are just databases, and you need lots of fast memory in order to feed the NPUs data, and thats the current bottleneck.
I’m not denying they still gobble those up, but those components are not what’s drawing all the power.