BG5USN
@BG5USN
Full-stack, MCU, Embedded System, Raspberry Pi, Javascript, c/c++. http://sourcekit.cc / http://stc.ninja / http://i2c.ninja
#ESP32 Backdoor? "they found 29 undocumented commands ... that could be used for Memory Manipulation (read/write RAM and Flash), MAC Address Spoofing (device impersonation), and LMP/LLCP Packet Injection" bleepingcomputer.com/news/security/…
#Rockchip RK182x co-processor for LLM. LLM on chip is on the way. #RKDC2025
#Rockchip developer conference 2025 at Fuzhou exhibition center. July 17-18
Just way better than Love,death,robot S04.
Prompt Theory (Made with Veo 3) What if AI-generated characters refused to believe they were AI-generated?
Remote 1993 2160p 4K AI Upscaled (Full Movie) youtu.be/duO4cAbcF0w?si… via @YouTube If we could not make new movies like this, we should upscale more from the 90s.
Linus Torvalds figured out Microsoft’s agenda way before the rest of us. 🙂
Test on CH583, a risc-v mcu without float process unit, 1/sqrtf() is faster than Fast inverse square root from quake3 with just 1 Newton iteration. (en.wikipedia.org/wiki/Fast_inve…). Did not expect that. but why?
所以4090 96GB版本已经搓出来了？？？ 第二个 NVIDIA-SMI 的数据是 reddit 用户 DirectAd1674 放出的 #AI #LLMs
“搬起石头砸自己的脚”的正确翻译：shoot myself in the foot.
"Not every post I make is a banger, and I do shoot myself in the foot, from time to time, but, at least, you know, it's genuine, it's not some sort of PR department, deciding things, and if you're constantly going through a filter, you're not being real.” Elon Musk
