Okay, not as bad as I thought, there's a guy selling a 4090 for "only" 2k. There's nobody selling a 3090 in the local hardware forum. I'm not going to buy off ebay because if someone on ebay sells a dud, I can't go back to punch him in the face.
5090s start at about 5200, that's true. To be fair, the "4-5k" for a used 3090 was a bit hyperbolic, I'm just mad that they're so expensive that my last 2 cars cost less than a USED 4090.
Your newer car will also tell you EXACTLY which sensors are showing bad data and whether it's an electrical fault in the sensor's circuit or implausible input, etc. Unfortunately you need a scan tool for it, but depending on the make you might find a decent free or cheap app to use a generic ELM327 or better, STNN1110 based OBD2 tool, that lets you scan all the modules rather than just OBD2 standard stuff.
So rather than playing parts darts or verifying every sensor by multimeter using known resistance ranges or putting up with bad fuel economy and low power... You just scan the thing and it'll tell you.
I've owned cars from the '90s, '00s and '10s (2019 being the latest) and while my cars from '10s didn't have sensor issues, I'll tell you that I'll take the complex and super hated '00s German cars over the early '90s one ANY day of the week. Super easy to diagnose.
Why does what need root access? The person paying for the hosting? So they can install software cloudflare doesn't offer for example.
It exists but nobody wants to use it. We're a minority. Most want to open "the app" and not spend any effort finding things.
The tokens ARE language. They're just words or parts of words.
If you're using a western LLM, English IS its strongest language. Otherwise it might be Chinese or English.
There's no "LLM native language" to convert to for efficiency. Maybe pseudocode or actual code for things where you need to disambiguate.
A used 3090 alone is like what, 4-5k now?
Your setup isn't really achievable for most now.
I'm listening to the audiobook. This time I just remember the meanings. The important ones at least
After the last one they never really lowered prices back, we were still buying near crypto era priced GPUs before the AI boom.
Source on TSMC hiring ex-Apple employees lol
Apple uses TSMC's services because there's a huge difference between what Apple does (chip design) vs what TSMC does (chip manufacturing).
It wouldn’t be “years of failure and low output”. It would be years of nothing to build the infrastructure and then flip a switch.
No, the failure and low output come AFTER the switch is flipped, unless you get a TON of industry know-how from TSMC. First you have nothing, then you have low yields until you figure out the processes.
After the last crypto bust, prices stayed up though. They're still way higher than before the last crypto boom.
And nVidia, AMD, Intel all know better now than to overproduce consumer GPUs.
Nvidia already made tons of profit off it. They're the ones selling shovels.
What's the point in coming up with an entire new language for them if they can just write English and humans will also understand it?
Kinda pointless to have LLMs optimised to produce garbage humans can't understand because at the end, your goal is still to produce something humans will consume.