4 ms·
Uhhh, 117.2 GB/s isn't hard at all on die, getting ~10X that amount is possible even through the HBM2 bus. Instead of downvoting me, can you actually present a
by deepnotderp 9y ago
Uhhh, 117.2 GB/s isn't hard at all on die, getting ~10X that amount is possible even through the HBM2 bus.
Instead of downvoting me, can you actually present a technical argument? Have you ever designed a chip or taped one out?
And yes, physically it definitely is indeed simple (what would you consider a "complex" processor and a "simple" processor?). Plus these processors aren't using caches, there's no need for them.
- 0xbear 9y agoThis _is_ technical argument, and it's not me who's downvoting you. You have to keep this thing fed, and once you factor that in, "5W" and "simple chip" become physical impossibilities.
- deepnotderp 9y agoSure, it's an argument, but it's wrong. What do you consider to be a "simple" chip and a "complex" chip? From a design perspective, it absolutely is a "simple" chip. And keeping the thing fed in terms of memory bandwidth is not that difficult if done on chip. Now, I have no idea what these guys are doing, but according to our simulations, (and people we've showed in the semi industry agree with the notion) it's quite possible to do that.
- 0xbear 9y agoIt can't be "on chip": data has to get in and out somehow. It's "quite possible" to do a lot of things if you don't intend to actually do any of them.
- deepnotderp 9y ago1. Errr what? You mean the input data? Sure, but that isn't usually considered the power consumption. Otherwise you can totally have most of the data (in the sense where data is the weights and activation maps) reside on chip. That's the only way to get high power efficiency. 2. Well, only time can tell there :)