Cyfrit

6 ms·

1. 1.▲
   Show HN: Offline 180.9M-parameter LLM and Agent inference on ESP32-P4(github.com)

   2 points by Cyfrit 1mo ago|0 comments
2. 2.▲
   by Cyfrit 2mo ago

   I actually got a free exam attempt once (I've forgotten which cert it was now), but because their proctoring software didn't work, I eventually just couldn't be bothered to take it lol. As for this, it really depends on your
3. 3.▲
   by Cyfrit 2mo ago

   While I'm not sure about the exact e-paper panel model used in this project, in my experience with these small e-paper screens, a full refresh is pretty much mandatory every 10 or so partial refreshes—which is significantly worse than
4. 4.▲
   by Cyfrit 4mo ago

   Unbelievably, even DDR3 isn't spared. Our suppliers have been quoting us much higher prices for DDR3 for months now, and we're experiencing persistent stock shortages. Fabs are abandoning legacy nodes to prioritize AI capacity, ma
5. 5.▲
   by Cyfrit 4mo ago

   The main downside is shifting from inline validation to out-of-band state syncing. For handshakes to stay small, browsers must constantly cache fresh "landmarks." If a device has been offline and hits a flaky hotel captive portal,