Suck at what aspect of (analog) electronics specifically? Not contradicting the claim, just want to understand it.
I have not tested yet, but I suspect that LLMs with a harness that can execute code can do SPICE simulations rather ok these days? I have seen MCPs for measurement equipment also, maybe they can even close the physical loop?
spice is a very bad simulation. it's not like a unit test or a VM. it works very differently to the real thing, only approximating it in some highly controlled situations.
If you have accurate device models, and you know how to use simulations, you can predict quite well the behavior of the analog circuits that you design.
Simulations that work like the real thing are an absolute necessity in the industry. It is frequent to design the analog parts of some integrated circuit, when you might have to wait months to get prototype samples that you can measure in the laboratory.
If the prototypes do not work exactly as expected, you will probably have the chance to make once a correction of the schematics and/or layout, which will be incorporated in the final product.
But when not even the second try does not work as designed, that is likely to be a failure, because any other redesign would add months and many millions of $ until the product can reach the market.
So "spice" simulations that can be trusted are a necessity. When designing integrated circuits you cannot use empirical methods that work when you make manually a single-use board, where you can replace components or adjust resistors, capacitors or inductors until you get what you want. For integrated circuits, you must know in advance how the circuit will work, because you cannot tweak it.
It is true however that it is easy to configure simulations that will give non-sense results, due to lack of convergence or other numerical errors, so to avoid this, one typically needs both experience with laboratory measurements and experience in using the simulator.
I doubt that an LLM would successfully run SPICE simulations, because there are few examples of written instructions of how to do this in the right way (which varies depending on the domain of applications, and each SPICE-like simulator may have its own quirks).
> I have seen MCPs for measurement equipment also, maybe they can even close the physical loop?
I'm working on this but for various reasons can't have my physical lab up at the moment. But yes there are many options for connected test equipment that could rather trivially be interacted with via LLM or pretty easy to write libraries.
The cards are also coming... Some of the very first are already out - but it will take a couple of years before they hit their stride in full. And cards will likely be primarily for Chinese companies initially, and then go for exports later. So it will go under the radar for many western techies for a while. Seeing what China has managed with PCB production, low to mid-end semiconductors, phone manufacturing and electric I would hesitate to bet against them.
Nvidia is now lobbying to remove the export ban, to try to reduce the incentives for developing their own high-end LLM training/serving (and cash in on current demand).
Nvidia will fail in that respect. Sanctions and restrictions are too popular with the American government. Where the United States will lose, is that almost every country in the world that can will use other solutions not from the United States.
And unfortunately, we are losing credibility on all fronts in the last year and a half. I wonder why?
Even in a fine tuning scenario, the preferred form for making modifications is the weights plus the training pipelines, including evaluation protocols, tooling, etc.
Not comparable today. But in 10 or 20 years the situation might be considerably different. Hopefully closer to a everyone-can-build-using-hardware-they-already-have.
Does anyone have a good article on the topic? This one seems rather LLM generated. Not sure the specifics can be trusted... Overall thesis is maybe right, but would like to understand this in a bit more detail.
Capability per GB and per watt has also been going up lot. This will continue in the future as well (not necessary as the same rate as last years). But enough that I think Opus 4.8 level is reachable on consumer PCs within 10 years from its release. Say at the price point of 2000 USD in 2025 dollars.
I have not tested yet, but I suspect that LLMs with a harness that can execute code can do SPICE simulations rather ok these days? I have seen MCPs for measurement equipment also, maybe they can even close the physical loop?
reply