thomasjb

It's not a great UI, but for messing with ideas in a VM, the free tier is very hard to beat, otherwise I'd be on Claude Code, but still on a VM, because I won't let them near anything I actually care about.

Easiest to use when sshing into a VM.

I was inspired by that post also, got a Qwen Coder 1.5B up to 27tok/s prompt eval and 13tok/s decode on an e5-2650v2 inside a GNOME box

I aspire to someday move up to an AM5 system, but for now, the Dell T3610 has to do everything. Might get a second GPU for it though.

Or watch and wait as models get denser

I've been watching and waiting for this, interested to see how smart it is, as it fits with my interest of getting the smartest possible model running in 10GB of VRAM (RTX3060 that has to drive 2 monitors and run an llm)

I can see the appeal, not having to deal with much rust or bolts breaking (the 2 things which cause the most trouble for me in working on vehicles). 800 miles across the desert is some way to run it in though!

Opencode's free models have been fine for me, they're what I tried after Gemma 4 8B proved hard to persuade into usefulness (I want to revisit with 12B and messing with harnesses, but I'm happy for now).

Unfortunately there's no gguf quants of the assistant model yet: https://huggingface.co/models?other=base_model:quantized:goo...

That's what I want to know too. A smarter E4B that's happy in opencode would be a good selfhosted model for me

There's probably still going to be a box of hard drives in a datacenter somewhere, it does make sense to have a layer to manage the agent interface, rather than letting agents completely loose on all your storage.

I second this recommendation!

One of them was returned to service after 40 years in the boneyard in Arizona, back in 2011, I would expect they'll look at the other airframes there to see if they're suitable sources for a rebuild. Wouldn't be surprised if this is the end of this one though, it was already doing pretty well for a design that first flew in 1949 (the English Electric Canberra design that was then built by Martin)

It's snappy, seems to work well

Working in silicon design, Verilog and SystemVerilog are the bulk of the work, but a lot of scripting of EDA tools too. I enjoy it, I think it's very different working on something which has to be complete when you're done versus something that can be updated easily.

It's been succeeded by Lemontron: https://lemontron.com/

Possibly. Realistically this is replacing the expensive category of FPGA (Zynqs or similar with strong hardware CPU cores), this means they get all the peripherals they desire in hardware, and they can pick the core variant in order to optimise for their workloads (all the different vector extensions for example). There's an interesting market for that kind of thing, either full FPGA to ASIC replacement, or drop in replacement FPGAs of lower cost (The Rigol MHO98 replaced the Xilinx FPGA of the previous generation with a substitute from Fudan). If you're shipping a lot of hardware, that sort of thing becomes worthwhile.

C, because it makes every problem into a memory management problem, which is good for you in an 'eat your vegetables' sort of way. It's also the starting point for a lot of other programming languages and related things like HDLs, which is helpful to me.

I'm plodding my way through the 2015 challenge here: https://git.thomasballantine.com/thomasballantine/Advent_of_... , it's really sharpened me up on a number of points.

Burning off black spray paint for etch resist worked for me. An extra step but easy enough

I think its another symptom of the problem of there not being a clear way to get from your smart brainbox which you run a proper operating system on and do heavy computation on to driving lots of motors or similar. There are options, but there's not one that everyone defaults to and you can get good information on.

What we need is a web made in a similar way to the wicker-bodied cars of yesteryear

I wonder what aspect ratio we'll settle on for little corner of our eye viewing on AR gear?

Inline Python where all you have to do is put in <python></python>

I'd encourage you to get his email newsletter as well! He puts out really interesting and thoughtful content in my experience.

Do everything in serverside cgi programs, at least until WebAssembly disposes of the currently necessary JS wrappers to access the DOM.

Note for anyone trying to receive lower frequency signals (below 10MHz): The supplied coaxial cable is really really lossy at those frequencies, a change of cable ought to make a big difference in what you receive.

This is a really beautiful bit of kit, really like the case and the thought given to signing, but the analog side is really impressive too. Would it be viable to use the same hardware as a base for an NTP server?

Wayne Green [0] is responsible for that. Responsible for some publishing innovation in the electronics world, he started with writing about radioteletypes for hams in the 50s, and moved up to Peterborough after starting 73 magazine

[0]https://en.wikipedia.org/wiki/Wayne_Green

Love this! Part of why I went into electronics is reading archives of Byte and Popular Electronics

I think an easy win for tram/tramtrain systems would be to have capacity to move cargo, either in rollcages like grocery shops use, or on pallets with dollies from a peripheral location, to city center locations and eliminate at least some light goods vehicles. The problem is, it would probably be a very inefficient use of labour to have humans do all the last few hundred yards pushing.