Hacker News new | past | comments | ask | show | jobs | submit | NavinF's comments login

> HACL* is a formally verified library of modern cryptographic algorithms, where each primitive is verified for memory safety, functional correctness, and secret independence. HACL* provides efficient, readable, standalone C code for each algorithm that can be easily integrated into any C project.

> All the code in this repository is released under an Apache 2.0 license. The generated C code from HACL* is also released under an MIT license.

You have an unusual definition of "depends on Microsoft". Anyone worried about depending on Microsoft should be able to maintain 15k lines of C that are already formally verified. Python already vendored the code so who cares who wrote that code?


Maintaining a new service sucks. Not being able to do atomic commits to both postgres and the other db sucks.

This game is used to benchmark scrolling devices: https://www.youtube.com/watch?v=tzqJ1rJURgs

We have hot pluggable I2C at home. Every HDMI port (and pretty much every DP port with a passive adapter) has I2C on the DDC pins. The port also provides 5V 50mA so your MCU doesn't need external power. Example: https://www.reddit.com/r/raspberry_pi/comments/ws1ale/i2c_on...


Does HDMI use SMBus ARP?


Nope. Software on the PC side reads EDID data from a fixed address. The EDID can list other addresses

Last time I saw someone complaining about scrapers, they were talking about 100gib/month. That's 300kbps. Less than $1/month in IP transit and ~$0 in compute. Personally I've never noticed bots show up on a resource graph. As long as you don't block them, they won't bother using more than a few IPs and they'll backoff when they're throttled


For some sites, things are a lot worse. See, for example, Jonathan Corbet's report[0].

0 - https://social.kernel.org/notice/AqJkUigsjad3gQc664


He provides no info. req/s? 95%ile mbps? How does he know the requests come from an "AI-scraper" as opposed to a normal L7 DDoS? LWN is a pretty simple site, it should be easy to saturate 10G ports


Didn't rachelbytheebay post recently that her blog was being swamped? I've heard that from a few self-hosting bloggers now. And Wikipedia has recently said more than half of traffic is noe bots. ARe you claiming this isn't a real problem?


How exactly can a blog get swamped? It takes ~0 compute per request. Yes I'm claiming this is a fake problem


How can you say it's $0 in compute without knowing if the data returned required any computation?


Look at the sibling replies. All the kvetching comes from blogs and simple websites, not the ones that consume compute per request


Yikes. I was sure this was an April Fools article till I got to the end. Has any other vehicle lost 6DOF control with humans on board?



> Linux power management is pretty good

> defaults favor desktop and server performance

Desktops are in S3 half the day consuming ~0 power. During use, electricity costs are so much lower than hardware costs that approximately nobody cares about or even measures the former. Servers have background tasks running at idle priority all day so the power consumption is effectively constant. Laptop and phone are the only platforms where the concept of "Linux power management" makes any sense.


My Mac mini (M1) sips ~6W idle and is completely inaudible. It acts as a desktop whenever I need it to, and as a server 24/7. I only power up my NAS (WoL) for daily backups. The rest of the homelab is for fun and experiments, but mostly gone.

"Idle" x86-64 SOHO servers still eat ~30W with carefully selected parts and when correctly tuned, >60W if you just put together random junk. "Cloud" works because of economies of scale. If there's a future where people own their stuff, minimising power draw is a key step.


The low power draw is definitely not exclusive to Macs, a similar x86 mini PC with Linux will also draw around 5W idle.


Does the mini PC go from zero to eleven though? Can I play BG3, Factorio, or Minecraft on the same hardware? Can I saturate a TB3 port? Transcode video? Run an LLM or text2img? Any of that while remaining responsive, having a video call?

If I already need a powerful machine for a desktop, why would I need a second one just so it can stay up 24/7 to run Miniflux or Syncthing? Less is more.


Yes; for about $1000. Eg:

https://www.bee-link.com/products/beelink-ser9-ai-9-hx-370

I have the ser-8 model, and can confirm everything works under Linux. This one has an 80 TOPS AI thing, since you asked about llms.


Huh, pretty cool. Would you mind submitting your scores to Geekbench? Can you also test idle power? Genuinely interested.


Try their Deep Research or grok's DeepSearch. Both do many searches and read many articles over a couple of minutes


The "Deep" search features hallucinate like crazy, I've found.


That hasn't been an issue for me. Link to example?


This was announced yesterday so I'm kinda surprised it hasn't been submitted to HN yet: https://x.com/lmarena_ai/status/1896675400916566357


>You can't have it both ways. You can't say "I want to have nothing to do with Rust", and then in the very next sentence say "And that means that the Rust code that I will ignore cannot use the C interfaces I maintain"

Seems fair to me. I wish the title replaced "you get no say on it" with "you get no say on rust bindings"


Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: