Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Apple Is Working on a Dedicated Chip to Power AI on Devices

bloomberg.com

Post reply on HN
Print view View original

111–120 of 123 posts

Re: Apple Is Working on a Dedicated Chip to Power AI on Devices

#111

Post by jacquesm »

bitmapbrother wrote: post #82
jacquesm wrote: post #41

Earlier quoted context omitted.

I'd bet on this being an ASIC, doing this on an FPGA with any serious size matrix would require a very expensive FPGA, whereas an ASIC would allow more gates in a smaller volume and would consume less power to boot. From the manufacturers point of view phones not being future proof is a feature, not a bug, that way you'll upgrade to that new shiny item which will keep the profits rolling in.

I don't think so. I believe Apple is basically going to do what Google and Qualcomm do and use a DSP's.

'A DSP's'? (sic)

DSPs are not nearly as good at matrix math as a GPU and the phone already has one of those. DSPs are typically good at signal processing in a fairly limited domain, they would not be calling it an AI chip if it was just a DSP.

Re: Apple Is Working on a Dedicated Chip to Power AI on Devices

#112

Post by fauigerzigerk »

amirhirsch wrote: post #96
djrogers wrote: post #52

Earlier quoted context omitted.

You don't put an FPGA in a device you're going to sell 200M+ of. The cost per unit would be way higher than an asic, and your just going to come out with a better version next year anyway so why bother? I foresee it as similar to their M series co-processors - the first one was pretty basic, and more sensors and jobs have been given to the newer ones each year.

I work on FPGA embedded vision. I think a lot of people in this thread are making incorrect assumptions about FPGAs implementations of neural network applications. (1) forward networks are constant multiplications, i.e. Fixed shift and add. FPGAs are very nearly optimal architecture for programmable constant-shift-and-add (2) individual neurons in a network can be bitwidth optimized and huffman encoded for bit-serial…

But what about unit cost?

Re: Apple Is Working on a Dedicated Chip to Power AI on Devices

#113

Post by petra »

highd wrote: post #14

This is probably going to be a hyper parallel fixed point / integer engine like TPU gen1. Doing fast matrix multiply over really small fields is very subpar on CPUs and GPUs. That was the initial reasoning behind TPU gen1 - improving runtime performance. One question is if it will architecturally be closer to a GPU or an FPGA. The field moves so fast that it might make sense to "future-proof" a bit with a live-reconf…

Mythic is working on analog/digital neural network chips. Some others also. Much better power-consumption.

Maybe Apple aims on those ? it will certainly achieve a much higher WOW factor, which is key here.

Re: Apple Is Working on a Dedicated Chip to Power AI on Devices

#114

Post by ferbivore »

moonforeshot wrote: post #104
LrnByTeach wrote: post #89

Earlier quoted context omitted.

You nailed the Apple plan, they did exact same thing staring from Introduction of Siri (4S?), ... then Force Touch etc. With each iteration of phone, they add a shiny new feature that is ONLY available on the newer model.

i don't see how force touch can be added to older hardwares...

It can't be. Night Shift, on the other hand...

Re: Apple Is Working on a Dedicated Chip to Power AI on Devices

#115

Post by mercer »

DannyBee wrote: post #78
djrogers wrote: post #77

Earlier quoted context omitted.

That's exactly why Apple does things like analyze your photo library locally on the phone - for privacy.

Yes, surely this has no relation to their inability to build reliable cloud products ....

It could be both. Perhaps Apple concluded that 1) they're subpar with cloud services and will have difficulty competing, 2) there's a growing need/demand for more privacy and less 'cloud', and 3) Apple's products are already, on the whole, recommended when it comes to privacy.

And based on that they figured privacy was a good thing to aim for. Play to their strengths and differentiate based on that.

Re: Apple Is Working on a Dedicated Chip to Power AI on Devices

#117

Post by msh »

bonoetmalo wrote: post #116

I'm not too familiar with the concept of ML specific chip designs, but isn't most AI done on servers and the results returned to the device? What kind of applications involve local ML code execution?

If you want to protect user privacy it's a very relevant selling point to be able to say that the data never leaves the device.

Re: Apple Is Working on a Dedicated Chip to Power AI on Devices

#118

Post by gigatexal »

npgatech wrote: post #91
bitmapbrother wrote: post #83

Earlier quoted context omitted.

Google announced it at I/O. They're using DSP's, specialized for neural network processing, on the SoC with TensorFlow Lite on the device. https://youtu.be/Y2VF8tmLFHw?t=1h22m8s

Oh wow, thanks. I feel like Apple is a step behind in pretty much all things AI.

Yeah SIRI has been lame compared to Google's hey Google for a long time.

Re: Apple Is Working on a Dedicated Chip to Power AI on Devices

#119

Post by Grazester »

mathieuh wrote: post #102
midhunsezhi wrote: post #98

Earlier quoted context omitted.

Siri vs Google assistant. Google assistant being way better in understanding voice and performing tasks. I've used both iPhone and Pixel and I can confirm that Google assistant is way more smart and intelligent than Siri at this point.

Don't know why you're being downvoted, this has been my experience too. Siri feels more like a toy with its infuriating jokes and stunted capabilities, Google Assistant feels much more polished. I have confidence that attempting a command I've never tried before with Google Assistant will work, with Siri it's potluck.

Yeah it's not even up for debate. Google is way ahead of Apple in Ai and just about nearly another we could name.

Re: Apple Is Working on a Dedicated Chip to Power AI on Devices

#120

Post by nickpsecurity »

jacquesm wrote: post #41
highd wrote: post #14

This is probably going to be a hyper parallel fixed point / integer engine like TPU gen1. Doing fast matrix multiply over really small fields is very subpar on CPUs and GPUs. That was the initial reasoning behind TPU gen1 - improving runtime performance. One question is if it will architecturally be closer to a GPU or an FPGA. The field moves so fast that it might make sense to "future-proof" a bit with a live-reconf…

I'd bet on this being an ASIC, doing this on an FPGA with any serious size matrix would require a very expensive FPGA, whereas an ASIC would allow more gates in a smaller volume and would consume less power to boot. From the manufacturers point of view phones not being future proof is a feature, not a bug, that way you'll upgrade to that new shiny item which will keep the profits rolling in.

"I'd bet on this being an ASIC"

Aren't all the iPhone chips ASIC's with the main one custom by those hardware people they acquired? Seems to be default expectation for whatever they add next. They sure as hell have the money, too. :)

Post reply on HN