281–290 of 377 posts
Earlier quoted context omitted.
If they were cost competitive for ... anything at all really...
I picked up several Rpi 4 2GB for $20 each just before covid-19. At that price point they've been quite competitive for small homelab workloads. The current RPi 5 makes no sense to me in any configuration, given its pricing.
For anyone interested in playing with distributed systems, I'd really recommend getting a single machine with latest 16-core CPU from AMD and just running 8 virtual machines on it. 8 virtual machines, with 4 hyper threads pinned per machine, and 1/8 of total RAM per machine. Create a network between them virtually within your virtualization software of choice (such as Proxmox). And suddenly you can start playing with…
Earlier quoted context omitted.
> the raspberry pi is still an amazingly cost effective choice It's really not though. I've been a Pi user and fan since it was first announced, and I have dozens of them, so I'm not hating on RPi here; we did the maths some time back here on HN when something else Pi related came up. If you go for a Pi5 with say 8GB RAM, by the time you factor in an SSD + HAT + PSU + Case + Cooler (+ maybe a uSD), you're actually al…
> SSD + HAT + PSU + Case + Cooler (+ maybe a uSD) The only 100% required thing on there is some sort of power supply, and an SD card, and I suspect a lot of people have a spare USB-C cable and brick lying around. A cooler is only recommended if you're going to be putting it under sustained CPU load, and they're like $10 on Amazon.
Particularly with Pi 5, any old brick that might be hanging around has a fair chance at not being able to supply sufficient power.
Earlier quoted context omitted.
I did some calculations on this. Procuring a Mac Studio with the latest Mx Ultra processor and maxing out the memory seems to be the most cost effective way to break into 100b+ parameter model space.
You could try getting a DGX Thor devkit with 128GB unified memory. Cheaper than the 96GB mac studio and more FLOPs.
There is a reason all the big supercomputers have started using GPUs in the last decade. They are much more efficient. If you want 32bit parallel performance just buy some consumer GPUs and hook them up. If you need 64bit buy some prosumer GPUs like the RTX 6000 Pro and you are done. Nobody is really building CPU clusters these days.
The best option for DP throughput for hobbyists interested in HPC might be old AMD cards from before they, too, realized that scientific folks would pay up the nose for higher precision.
Earlier quoted context omitted.
Like the joke about the economists not picking up the $20 bill on the ground? Faith in the perfect efficiency of the free market only works out over the long term. In the short term we have a lot of habits that serve as heuristics for doing a good job most of the time.
> Like the joke about the economists not picking up the $20 bill on the ground? For those like me that don't know the joke: Two economists are walking down the street. One of them says “Look, there’s a twenty-dollar bill on the sidewalk!” The other economist says “No there’s not. If there was, someone would have picked it up already.”
Earlier quoted context omitted.
What if you need ARM? What is the best sub $100 sbc that I am missing? Orange Pi hardware always looks good but I hear a lot of negativity about the software that I don't really experience with Raspbian.
Then you should take a good hard look at older, much cheaper Raspberry Pis. Then look at Apple’s ARM offerings, and AWS Graviton if you need ARM with raw power. If you need embedded/GPIO you should consider an Arduino, or clone. If you need GPIOs and Internet connectivity, look at an ESP32. GPIOs, ARM and wired ethernet? Consdier the the STM32H. Robotics/machine vision applications, needing IO and lots of compute pow…
Earlier quoted context omitted.
Not completely related, but do you know if hardware kickstarters typically have any IP protection? I'm surprised there haven't been any cases of large companies creating patents for ideas from kickstarter at least that I've seen
One cannot (or at least, one is not supposed to be able to) patent someone else's invention.
Not so good, and this is the sort of title. you need to bring the punters in for YouTube.
I don't mean to sound too cynical, I appreciate Jeff's videos, just wanted to point out that if you've spent money and time on content you can either ditch it or make a regret video.
Just so long as the thumbnails don't have an arrow on them I'm happy.
If Pi Clusters were actually cost competitive for performance there would be data centres full of them.
If they were cost competitive for ... anything at all really...
If one just wants a cheap desktop box to do desktop things with, then they're a terrible option, price-wise, compared to things like used corpo mini-PCs.
But they're reasonably cost-competitive with other new (not used!) small computers that are tinkerer-friendly, and unlike many similar constructs there's a plethora of community-driven support for doing useful things with the unusual interfaces they expose.