Try having it output the circuit in SPICE. It actually works surprisingly well and does a good job picking out components values for parts and can describe the connectivity well. It falls apart when it writes the SPICE (professionally, there isn’t really one well accepted syntax really)and making the wires to connect your components, like you say missing the minds eye. But I can imagine adding a ton spice schematics with detailed descriptions with maybe an LLM optimized SPICE syntax to the training data set… it’ll be designing and simulating circuits in no time.
Yeah, how to you thing that schematic is represented internally? How do you think the netlist is modeled? It's SPICE and HDL all the way down!
There are good reasons not to vibecode Verilog, but a lot of test cases are already being written by LLMs and the big EDA vendors (Cadence, Synopsys, Siemens) all tout their new AI capabilities.
It's like saying it can't read handwritten mathematical formulas, when it solves most math problems in markup (and if you aren't using it you're asking for trouble).
I brainfarted a bit and mixed up my attempts with making LTSPICE asc schematics (which are the text representations of the GUI sch, with wires) with the normal node based SPICE syntax. I just tried this specifically asking for spice to run with ngspice to run in a CLI. Seemed to run great! Going to play around with this for a bit now…
preemptively adding for us AMD users - it’s pretty seamless to get Ollama working with rocm, and if you have a card that’s a bit below the waterline (lowest supported is a 6800xt, i bought a 6750xt), you can use a community patch that will enable it for your card anyway:
I specifically recommend the method where you grab the patched rocblas.dll for your card model, and replace the one that Ollama is using, as someone who is technical but isn’t proficient with building from source (yet!)
What's the benefit of the container over installing as a tool with uv? It seems like extra work to get it up and running with a GPU, and if you're using a Mac, the container slows down your models.
Another reason to not use a DSLR is that many (all?) have timeouts (<30min) in their video mode due to some import tax reason, even when hooked up to a computer.
Atleast this is what I found when I tried a canon DSLR with canons webcam software.
That's only if you actually hit the record button and are actively recording to the memory card, but if you're using the DSLR as a passthrough, it works all day. I have done it w/ my Sony a6500 and it works really well.
Sony cameras work (as mentioned below) but nearly every mirrorless or dslr I’ve used does this by default or can be set up to by switching to continuous AF modes. This is best done while connected to a power source, as the motors drain battery faster.
This changed in 2019 IIRC. The EU changed its regulation and the 30 minute record limit no longer applies. Furthermore, it was always possible to install custom firmware on many cameras that bypassed this limit. Record limits due to temperature and overheating, though, is a different story.
"Digital cameras that are only capable of recording still images remain classified in this subheading. Cameras of this subheading may also have video-capture capability to record continuous periods of video. However, when such apparatus are capable, using the maximum storage capacity, of capturing video in a quality of 800 x 600 pixels (or higher) at 23 frames per second (or higher) for a continuous period of at least 30 minutes (regardless of the fact that the captured video images may be recorded in separate files of a duration of less than 30 minutes) they are always to be classified in subheadings 8525 80 91 or 8525 80 99."
Is it such a big issue? My Canon DSLR turns off every 30min, but that's only for a couple of seconds, it then turns back on. On a positive side, it's now easy to notice when 30min or 1hr meeting is running over, it's a nice reminder :)
I looked at the kweld, came very close to buying one, but ended up getting the Sequre version instead. About the same price, well reviewed, and looks to ship ready to go rather than you having to source pieces. It arrived earlier in the week, I haven't had time to open the box yet. kweld looks solid though. I thought about trying the DIY car battery and starter solenoid hacks, but wanted something a little more accurate.I have around $700 worth of battery packs to repair/rebuild (lawn mower and discontinued electric bike), so I hope to get my money's worth out of it.
>Low-orbit satellites spend half their day in darkness too. To hit a receiver on the night side of earth a satellite would have to be at a very high orbit, reducing beam efficiency and increasing launch costs.
Pedantic, but for LEO there are dawn-dusk SSO orbits that ride the terminator [1] so they get continuous sunlight, you could power some peoples evenings depending on how far the grid spans into the dark side. Not a solution for getting power at 2am though.
Might be able to compete if the cost if a rectenna is cheaper/kwh than solar panels (and free as sunlight). Then you can use your same battery solution for night time.
Yep, and why every supercar's 0-60 time is limited to about ~2.6s these days. Also why tesla is claiming to be able to achieve <2.6s only with the cold gas thruster "SpaceX" package, they're no longer solely relying on the tire traction.
The exact number wasn't really the point + elon has a history of releasing high performance spec's before the actual technology is ready. Maybe they have special tires in the works, but the rubber that's on the market today can't do that
