I agree that this seems quite similar to Cerebras. Does anyone think Tesla might sell raw compute hardware?
An inside look at the custom CPUs in Tesla's Dojo Supercomputer
21–30 of 132 posts
Re: An inside look at the custom CPUs in Tesla's Dojo Supercomputer
#22Earlier quoted context omitted.
> Transistor density isn’t doubling. It pretty much continues to double: https://en.m.wikipedia.org/wiki/Transistor_count#/media/File...
Raw transistor count isn't the same as density.
Sure, but if you look at the raw transistor count of a series of similarly sized dies (take Apple's A series for example), it holds up. It only doesn't hold up if you look at Intel the last 5 years. If you look at series' made on TSMC or Samsung, it holds up.
Re: An inside look at the custom CPUs in Tesla's Dojo Supercomputer
#23Now assume that Moore's law stays healthy and in a few decades we're carrying around dojo-equivalent smart phones. Or implants. What do our apps do with all that compute? Detailed convincing AR, accurate real-time translation, identify almost any object in sight, turing-test passing answering machine. We'll be carrying around neural-net training hardware that is different than the stuff between our ears, often worse…
> What do our apps do with all that compute? javascript
This isn’t far from the truth. But embrace it rather than feel upset by it. The world of programmers can be unlocked by a good API. And the underlying components can be C{,++}, as V8 is.
Re: An inside look at the custom CPUs in Tesla's Dojo Supercomputer
#24Earlier quoted context omitted.
Moore’s law is quite dead. Transistor density isn’t doubling. We’re still getting speed increases due to architectural improvements, but transistor density is capping out because thermals are a problem. That’s why you see these massive wafer designs and they’re not shrinking.
Moore’s law was about count, not density.
I mean, strictly speaking, you're technically correct. Informally, though, it's a proxy for transistor density.
Re: An inside look at the custom CPUs in Tesla's Dojo Supercomputer
#25I agree that this seems quite similar to Cerebras. Does anyone think Tesla might sell raw compute hardware?
That's a business far easier to develop than a personal robot.
Re: An inside look at the custom CPUs in Tesla's Dojo Supercomputer
#26Now assume that Moore's law stays healthy and in a few decades we're carrying around dojo-equivalent smart phones. Or implants. What do our apps do with all that compute? Detailed convincing AR, accurate real-time translation, identify almost any object in sight, turing-test passing answering machine. We'll be carrying around neural-net training hardware that is different than the stuff between our ears, often worse…
Based on current computing resource usage: ads.
Re: An inside look at the custom CPUs in Tesla's Dojo Supercomputer
#27Earlier quoted context omitted.
Moore’s law is quite dead. Transistor density isn’t doubling. We’re still getting speed increases due to architectural improvements, but transistor density is capping out because thermals are a problem. That’s why you see these massive wafer designs and they’re not shrinking.
Moore’s law was about count, not density.
It was about cost. https://hasler.ece.gatech.edu/Published_papers/Technology_ov...
Re: An inside look at the custom CPUs in Tesla's Dojo Supercomputer
#28Is this power from the top method something other people do?
Looks similar to what Cerberus is doing.
Re: An inside look at the custom CPUs in Tesla's Dojo Supercomputer
#29Wow. All that packaging and power delivery must be incredibly expensive. Yet they claim massive TCO advantage over Nvidia. Amazing what you can do when not paying for Nvidia's 60% margins ;)
If this is as good as they say, they should spin this off and sell these. Probably won't.
Re: An inside look at the custom CPUs in Tesla's Dojo Supercomputer
#30I'm assuming this is a reference to the dojo scene in The Matrix.