Публикация

Предыдущая публикация

AlphaFold has been a net-positive on society for protein-folding. Your CPUs' branch predictors have used neural networksfor years. When someone is against AI, they're probably upset about several things: 1. Reverse-centaurs 2. The lack of respect for consent and agency 3. Capitalism

opossumcaviar

26 июл., 02:22

Asking genuinely, are you sure about CPU branch predictors using NNs? It wouldn't surprise me, a small neural network can be teensy (esp if it's just an SLP), and I know there is research on this. But I've never read anything about these in a consumer CPU

Обсуждение

1 прямых ответов · 3 сообщений

Soatok

26 июл., 02:24

Actually, I'm not 100% sure. I could be misremembering!

Ответ для Soatok

Soatok

26 июл., 02:30

I had a conversation with infosec.exchange/@david_chisn... like 2 or 3 years ago where I heard this. Let me see if I can find the exact post when I'm back at a computer.

David Chisnall (*Now with 50% more sarcasm!*) (@david_chisnall@infosec.exchange)14.8K Posts, 128 Following, 5.34K Followers · I am Director of System Architecture at SCI Semiconductor and a Visiting Researcher at the University of Cambridge Computer Laboratory. I remain actively...infosec.exchange

Ответ для Soatok

opossumcaviar

26 июл., 03:55

Don't stress it at all if you can't find it! I went on a small rabithole (they seem to at least use one-layer NNs, which is not surprising or complicated? and something called TAGE but I simply ran out of time to find out what that entails)