alt Hacker NewsThis is his opinion, nothing more, nothing less. He currently has a partially implemented piece of software that hasn't seen a release since November and isn't performant at all.
Take the free offer, prove everyone wrong and then start to tell us how great you are. https://x.com/HotAisle/status/1880507210217750550
Replies
AMD is so behind NVidia that it's not even funny. If AMD board had any sense, they'd be carpet-bombing every researcher, AI startup, and random Joes with the latest engineering samples of unreleased top-of-the line products. And giving them a direct line to the engineering team.
This would end up costing maybe tens of millions at most, but the potential return is indeed measured in billions.
And yep, lots of people like geohot are (to put it mildly) eccentric. So deal with it. They are not merely your customers, they are your freaking sales people.
As it is, I work in a startup that does a bit of AI vision-related stuff. I'm not going to even touch AMD because I don't want to deal with divas on the AMD board in future. NVidia is more expensive right now, but they're far more predictable.
To be fair, having seen his software evolve, and having seen ROCm evolve, I'm more optimistic for his software in a year than yours.
He picked his problem better. The whole reason that tinygrad is, well, tiny, is that it limits the amount of overhead to onboard people and perform maintenance and rewrites. My strong impression is that the ROCm codebase is simply much too large for AMD's dev resources. You're trying to race NVidia on their turf with less resources. It's brave, but foolish.
I can see how Tinygrad could succeed. The story makes sense. AMD's doesn't, neither logically nor empirically. NVidia would have to seriously fumble.