Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Copperhead – Cursor for circuit boards

copperhead.sh

Post reply on HN
Print view View original

101–110 of 121 posts

Re: Show HN: Copperhead – Cursor for circuit boards

#101

Post by delfinom »

utopiah wrote: post #72
anirudhojhaanil wrote: post #15

With AI enabled consumer hardware usecases going to explode this is something super relevant. At least 5 startups have pitched their wearables to me and they could definitely use this

Will it though? Do you think people will actually be more creative with this or will we get yet another e-ink device with a slightly different form factor, or a "less than a phone" with different constraints? In the end is the PCB design the bottleneck to create new devices or is it mostly about available components and their price?

PCB design is no bottleneck. These services are solutions in search of a problem.

There are already millions of vendors all over the world making infinite variations of PCB designs for the same/similar product. You can hire a design shop in China for peanuts to make a variation as well.

Ultimately the bottleneck is the consumer does not care for infinite variations. The vast majority just want a "top rated product" at the lowest price. Which is whatever Amazon sorts their results as or whatever the Jones are buying that they need to keep up with.

The biggest driver of "unique product" (not just the 9 millionth clone product) commercial success is marketing, it is branding, it is finding a unique niche, it is timing, it is existing social & business relationships to drive sales.

Re: Show HN: Copperhead – Cursor for circuit boards

#102

Post by Joel_Mckay »

murderfs wrote: post #91
Joel_Mckay wrote: post #64

Earlier quoted context omitted.

Only decades of standards compliance, workmanship training, and experience. Human beings are very good at spotting patterns in noisy jungles, and making rational design tradeoffs. Auto-routers have been around far longer than even smartphones. Somewhat functional in the trivial problem domains, and always useless where design choices mattered. Best of luck solving a named problem. =3

> Somewhat functional in the trivial problem domains, and always useless where design choices mattered. This is laughable when you consider that semiconductor design is the least "trivial" problem domain, and it happens entirely in RTL. There's basically zero human intervention when it comes to layout: even floorplanning is being automated nowadays.

Fab standard cell libraries are already the practical optimization of their current capabilities, and even FPGA try to minimize routing proximity with various levels of success. However, there are again named problems that constrain what naive people can get away with....

https://en.wikipedia.org/wiki/Clock_domain_crossing

> floorplanning is being automated nowadays.

Procedural generation or parametric design is not the same, as the behavior and constraints were rationally engineered by people.

LLM would simply look at the outputs during training, abstract it as looking similar to a potato-chip factory, and generate a nonsense answer some fool assumes is appropriate.

Humans are wired to see meaningful patterns in chaotic systems even when they aren't really there. Neuromorphic computing may create real "AI" someday, but it almost certainly won't be from LLM cults. =3

https://en.wikipedia.org/wiki/Pareidolia

Re: Show HN: Copperhead – Cursor for circuit boards

#103

Post by Joel_Mckay »

CamperBob2 wrote: post #90
Joel_Mckay wrote: post #75

Earlier quoted context omitted.

LLM do not think because they are not real "AI", but it does copy the linear patterns people exhibit if statistically salient within the granularity of the higher dimensional vector search space proximity. Watermarking does skew the compaction slightly, but not far from resulting output patterns. =3 https://en.wikipedia.org/wiki/The_Subservient_Chicken

What were the chicken's thoughts on the Jacobian conjecture?

Finding a consistent pattern by conflating most papers, resolving a similar isomorphic vector search proximity result from an entire field of research.

LLM are very good at brute force context search spaces, and people see patterns in nonsense even when it is nebulous. Similar to the Newton's Apple story people are fond of telling each other. =3

https://en.wikipedia.org/wiki/Pareidolia

Re: Show HN: Copperhead – Cursor for circuit boards

#104

Post by Joel_Mckay »

adrianvi wrote: post #82
Joel_Mckay wrote: post #75

Earlier quoted context omitted.

LLM do not think because they are not real "AI", but it does copy the linear patterns people exhibit if statistically salient within the granularity of the higher dimensional vector search space proximity. Watermarking does skew the compaction slightly, but not far from resulting output patterns. =3 https://en.wikipedia.org/wiki/The_Subservient_Chicken

I do think that slightly smarter LLMs and a good (although not perfect) autorouting algorithm could solve 99% of hobbyist's projects and simple industry boards, due to the repetition of those patterns in real life. The LLM would learn the heiristics (example: data lines first, power lines later, etc.) and would request the autorouter to do that routing, then take the image output and request a different part (depth-f…

Every trace on a PCB is a traveling salesman problem with ballooning complexity.

https://en.wikipedia.org/wiki/Travelling_salesman_problem

While physics informed models do exist, they are still going to burn a lot of compute to generate failure modes people didn't know were possible. =3

https://www.youtube.com/watch?v=T4Upf_B9RLQ

Re: Show HN: Copperhead – Cursor for circuit boards

#105

Post by Aurornis »

Joel_Mckay wrote: post #51
Aurornis wrote: post #48

Earlier quoted context omitted.

If the magic LLM pixie sprinkles can brute force their way to a solution using viable heuristics and their training data, I will use them.

NP-Hard is NP-hard... "heuristics" in this case is just a euphemism for a fantasy cost-function to scam shareholders. It is a named problem, and can't be misappropriated like the $9Tn worth of FOSS code "AI" scraped and compacted for isomorphic plagiarism tokens. =3

> NP-Hard is NP-hard... "heuristics" in this case is just a euphemism for a fantasy cost-function to scam shareholders.

I don’t know why you’re trying to make this so hard.

The problem is NP hard whether I do it or a machine does it.

I can still do a good job despite being NP hard. I believe a machine with enough compute will be able to, too.

Re: Show HN: Copperhead – Cursor for circuit boards

#106

Post by Joel_Mckay »

Aurornis wrote: post #105
Joel_Mckay wrote: post #51

Earlier quoted context omitted.

NP-Hard is NP-hard... "heuristics" in this case is just a euphemism for a fantasy cost-function to scam shareholders. It is a named problem, and can't be misappropriated like the $9Tn worth of FOSS code "AI" scraped and compacted for isomorphic plagiarism tokens. =3

> NP-Hard is NP-hard... "heuristics" in this case is just a euphemism for a fantasy cost-function to scam shareholders. I don’t know why you’re trying to make this so hard. The problem is NP hard whether I do it or a machine does it. I can still do a good job despite being NP hard. I believe a machine with enough compute will be able to, too.

Copying millions of other peoples work for a similar looking result excludes the design compromises every engineer must make during the process.

LLM don't think because it is not real "AI", and simply conflates every isomorphic vector search proximity output.

Auto-routers have existed for decades, and they do work in the trivial design cases... but are almost always a liability where design choices matter.

Best of luck =3

Re: Show HN: Copperhead – Cursor for circuit boards

#107

Post by Aurornis »

Joel_Mckay wrote: post #106
Aurornis wrote: post #105

Earlier quoted context omitted.

> NP-Hard is NP-hard... "heuristics" in this case is just a euphemism for a fantasy cost-function to scam shareholders. I don’t know why you’re trying to make this so hard. The problem is NP hard whether I do it or a machine does it. I can still do a good job despite being NP hard. I believe a machine with enough compute will be able to, too.

Copying millions of other peoples work for a similar looking result excludes the design compromises every engineer must make during the process. LLM don't think because it is not real "AI", and simply conflates every isomorphic vector search proximity output. Auto-routers have existed for decades, and they do work in the trivial design cases... but are almost always a liability where design choices matter. Best of lu…

LLMs are not the same as auto routers that have existed for decades. They function completely differently and are capable of taking more context into account, all the way to fetching and reading data sheets and considering design compromises.

Re: Show HN: Copperhead – Cursor for circuit boards

#108

Post by adrianvi »

Joel_Mckay wrote: post #104
adrianvi wrote: post #82

Earlier quoted context omitted.

I do think that slightly smarter LLMs and a good (although not perfect) autorouting algorithm could solve 99% of hobbyist's projects and simple industry boards, due to the repetition of those patterns in real life. The LLM would learn the heiristics (example: data lines first, power lines later, etc.) and would request the autorouter to do that routing, then take the image output and request a different part (depth-f…

Every trace on a PCB is a traveling salesman problem with ballooning complexity. https://en.wikipedia.org/wiki/Travelling_salesman_problem While physics informed models do exist, they are still going to burn a lot of compute to generate failure modes people didn't know were possible. =3 https://www.youtube.com/watch?v=T4Upf_B9RLQ

Yes, the optimal solution won't be achievable, it will burn lots of compute and we need better simulations so that less errors are made (and will be made).

Still, if LLMs are capable of writing working code, art and solving math problems, they qre definitely capable of doing suboptimal routing (with some algorithmic aid)

Re: Show HN: Copperhead – Cursor for circuit boards

#109

Post by Joel_Mckay »

Aurornis wrote: post #107
Joel_Mckay wrote: post #106

Earlier quoted context omitted.

Copying millions of other peoples work for a similar looking result excludes the design compromises every engineer must make during the process. LLM don't think because it is not real "AI", and simply conflates every isomorphic vector search proximity output. Auto-routers have existed for decades, and they do work in the trivial design cases... but are almost always a liability where design choices matter. Best of lu…

LLMs are not the same as auto routers that have existed for decades. They function completely differently and are capable of taking more context into account, all the way to fetching and reading data sheets and considering design compromises.

Sure, one is a conflated chaotic nonsense generator, and the other is a chaotic random walk local "fit" evaluator set by an now arbitrary cost function.

NP-hard is still NP-hard... and named problems are rarely solvable with naive approaches. Best of luck =3

Re: Show HN: Copperhead – Cursor for circuit boards

#110

Post by tylergetsay »

dghlsakjg wrote: post #70
tylergetsay wrote: post #13

Can I take the output of this and get a fully assembled board mailed to me? Thats my dream...

Yes. Look at jlcpcb or other Chinese assembly houses. If you stay within their component library it is dead easy, just upload your gerber and BOM, and you can have assembled boards within a week or so. Slightly more difficult if they have to order parts for your BOM, but still not really that hard.

if I needed something on the board that was outside the component library of the assembly house, can I ship it to them? or have them arrange for it somehow? How constrained is the component library?
Post reply on HN