Live data from Hacker News

Google offers free fabbing for 130nm open-source chips

fossi-foundation.org

391–400 of 404 posts

Re: Google offers free fabbing for 130nm open-source chips

#391
post #154

So, this doesn't appear to have been announced by google. It does seem to be real but the OP may be jumping the gun a bit. The authoritative source seems to be the slides of this guy at google: https://docs.google.com/presentation/d/e/2PACX-1vRtwZPc8ykkk... From the slides this is "current plans, subject to change". This is an 'open source shuttle process'. Shuttle processes are a relatively cheap way of making small…

- You get somewhere around ~100-400 chips, TBD later - Each shuttle run will accept 40 projects, if more than that apply a selection process will be used. A lottery was mentioned as a possibility. - First shuttle run this November Like you mention everything is just planned at this point, but the speaker said these were roughly their goals.

Thanks. That sounds about what I expected.

Re: Google offers free fabbing for 130nm open-source chips

#392

Earlier quoted context omitted.

> std:: and TODO as synonyms What does this mean? I've only been in the C++ game for 6 years and for me if I can get something from std rather than rolling my own or pulling in another library, I'm cheering.

Lots of people replace STL with a faster one. See EA

EA’s STL had more to do with custom allocators and the fact that a lot of the platforms we were developing for 10 years ago didn’t have mmap or paged memory management (or at least nothing that they wanted to expose to us lowly user-mode gamedevs).

Re: Google offers free fabbing for 130nm open-source chips

#393
post #330

Earlier quoted context omitted.

Wholeheartedly agree. In parts of society where our distorted ideas of "productivity" and "success" are absent, people share more freely. The island where my family is from comes to mind (Crete, Greece). There—in the rural areas—people were able to deal with the 2007-2008 crisis a lot more effectively than they did in the cities—esp. compared to Athens. What I observed is that if someone didn't have enough, the rest…

but does this only work when the number of people in the tribe is low, and everyone has some social accountability to each other? The problem of a free-rider is ever present, and in big cities, that problem is evident and thus you don't get the sort of community effort that you see in a small village.

Yeah - see “Dunbar’s Number”

Beyond Dunbar’s Number, it’s not clear that you can employ mutual-aid theories like tit-for-tat to restrain free-riding and cheating.

Re: Google offers free fabbing for 130nm open-source chips

#394
post #357
post #242

Earlier quoted context omitted.

Classical economics thinks of people as "rational utility-maximizing actors", which doesn't approximate reality in quite a lot of ways. There's been a move toward more sophisticated models - like that people minimize regret more than they seek the optimal reward ("rational actors with regret minimization.") This switch to more complex underlying models is similar to computational models used in computer science. Algo…

The problem with the "rational agent" model is that it's a tautology. Yes sure everyone wants more utility, great, but everyone's utility function is slightly different. As you say some are risk takers pursuing insane rewards/yields/profits with low probability, while others are super risk-averse conservative in their choices, etc. That said, economics doesn't rest on this. Macroeconomics doesn't care, nor micro/labo…

Yes, if people are rational actors, whatever they do is rational according to their private utility function. So, if anything they do is rational, then how is it "rational", which supposedly depends on some objective metric?

Re: Google offers free fabbing for 130nm open-source chips

#395
post #242

Earlier quoted context omitted.

Classical economics thinks of people as "rational utility-maximizing actors", which doesn't approximate reality in quite a lot of ways. There's been a move toward more sophisticated models - like that people minimize regret more than they seek the optimal reward ("rational actors with regret minimization.") This switch to more complex underlying models is similar to computational models used in computer science. Algo…

> "rational actors with regret minimization." I keep seeing something like this in workplaces people don't join forces they avoid friction. Until something (crisis) or someone (capable leader) flips the thing on its head.

Follow the incentives. Any individual employee is massively incentivized to keep their job. There is a slight chance that a certain leader may be incentivized to bring success to the company. That is the exception, not the norm. I learned this the hard way by caring about companies that didn’t care about me.

Re: Google offers free fabbing for 130nm open-source chips

#396
post #350
post #302

Earlier quoted context omitted.

Haven't thought very deeply about this, but FOSS doesn't seem like socialism, as the capital being competed for is user attention. FOSS projects without enough capital does not gain enough developers, and falls into disrepair and obscurity. Not sure what a socialist FOSS movement might look like, but maybe developers would be assigned to projects as opposed to them freely choosing the projects to work on?

> Haven't thought very deeply about this, but FOSS doesn't seem like socialism, as the capital being competed for is user attention. I think that's stretching the definition of 'competition' a bit too much. First off, many people work on FOSS just because they want to, or because they believe that it's a good thing. Not because they want 'user attention'. Furthermore, I'm pretty sure that even the people that are mot…

Hi Mercer, interesting points.

But I think in general, the FOSS projects that thrive, that gather users which report bugs, that has an active community, are going to be the ones which attract developers, not necessarily because they seek fame and fortune, but just by the simple aspect that they are more likely to run into those projects, and be able to contribute to those projects more easily due to better documentation and active devs who can answer questions. It is the projects which compete for attention for their livelihood, like an emergent behavior, even if the developers don't explicitly target it.

Besides, developers are humans too, and they can feel pride when their contributions are recognized, and they can consider the work to be helping to pave the way to future employment. Contributing to large active projects give higher chances for both prospects.

Regarding whether it's state socialism that I'm thinking of, you're probably right. I've never been very clear on the definition of socialism. On the most superficial level, I understand socialism to be that the community controlled the means of production. So when applied to FOSS, that seems to mean that the whole community decides what projects to work on, and in the case where the community decided that something needs to be worked on, but not enough devs volunteer, then it seems like the work will need to be assigned.

Finally, as you said, it's not about FOSS being like socialism, but that they devs are not directly competing for big cash prizes that's the interesting aspect. I agree. There are not that many industries that are so fundamental to our society which has a large contingent of people doing critical work for free. Perhaps education comes close. Politics, with political power taking a similar role as attention for FOSS projects, but that's straying pretty far.

Re: Google offers free fabbing for 130nm open-source chips

#397
post #357

Earlier quoted context omitted.

The problem with the "rational agent" model is that it's a tautology. Yes sure everyone wants more utility, great, but everyone's utility function is slightly different. As you say some are risk takers pursuing insane rewards/yields/profits with low probability, while others are super risk-averse conservative in their choices, etc. That said, economics doesn't rest on this. Macroeconomics doesn't care, nor micro/labo…

> The problem with the "rational agent" model is that it's a tautology. Yes sure everyone wants more utility, great, but everyone's utility function is slightly different. As you say some are risk takers pursuing insane rewards/yields/profits with low probability, while others are super risk-averse conservative in their choices, etc. There's nothing in the rational agent model that assumes that everyone has the same…

Yes, of course, but that's what I mean by problem. Saying people are (bonded rational) utility function maximizers doesn't give us a predictive theory, it just means in a fancy way people do what they do for "reasons", and everyone usually have different set of reasons.

Of course, it's fine as a very-very general fundamental theory, if you then want to study how people's revealed and non-revealed (old names for implicit and explicit) preferences aggregate into a utility function. (There's a whole bunch of math about pairwise comparison matrices. Lately there's some movement in that space about using perturbation to model inconsistencies in preferences, etc.)

Re: Google offers free fabbing for 130nm open-source chips

#398
post #394
post #357

Earlier quoted context omitted.

The problem with the "rational agent" model is that it's a tautology. Yes sure everyone wants more utility, great, but everyone's utility function is slightly different. As you say some are risk takers pursuing insane rewards/yields/profits with low probability, while others are super risk-averse conservative in their choices, etc. That said, economics doesn't rest on this. Macroeconomics doesn't care, nor micro/labo…

Yes, if people are rational actors, whatever they do is rational according to their private utility function. So, if anything they do is rational, then how is it "rational", which supposedly depends on some objective metric?

It's rational because objectively the agent does what is best for it. Basically that's the definition of being rational. But without any qualifications, constraints, or additional context this description of decision theory is not really useful. (That was the point I tried to convey in my original comment.)

What even the general theory is useful for, is to quickly reduce the question of the very general behavior/decision problem in to "show me your utility function and I'll tell you what you will do". Of course then that becomes a problem. How to model utility, how to formalize expected utility over all possible actions. What about priors (as in path-dependence, so do we want to encode that into the utility function - and constantly dynamically update the function, like a Bayesian agent would update its beliefs - or somehow manage it separately)?

...

Basically I was trying to persuade people to don't say things like "economists are dumb, because people are not rational utility maximizers", because that's a false implication. Of course they are, but - just like you said - this doesn't help us better predict people's (agents') behavior, just gives us a new task to model their utility functions. And that's where regret comes in. (Which is basically risk-aversion, which is what behavioral economics studies. People are not symmetric about positive and negative expectations. They value avoiding negative-utility events more than they value encountering positive-utility events. And of course with this insight now we can build better economic models - for example we can use this to "explain" why wages are more sticky - especially downward - than we would expect, also with some handwaving we can explain why in times of crisis people are let go instead of decreasing their hourly compensation, why some people are so gullible [peer pressure, confirmation bias ~ cognitive dissonance], and so on.)

Re: Google offers free fabbing for 130nm open-source chips

#399

Earlier quoted context omitted.

> "rational actors with regret minimization." I keep seeing something like this in workplaces people don't join forces they avoid friction. Until something (crisis) or someone (capable leader) flips the thing on its head.

Follow the incentives. Any individual employee is massively incentivized to keep their job. There is a slight chance that a certain leader may be incentivized to bring success to the company. That is the exception, not the norm. I learned this the hard way by caring about companies that didn’t care about me.

'keeping your job' implying submission to social pressure to be accepted and nothing more ? I'm not sure I got your whole argument, so if you have more details, I'd be happy to read :)

Re: Google offers free fabbing for 130nm open-source chips

#400

Earlier quoted context omitted.

Check out IceStorm (+Yosys, nextpnr...) for a pretty complete open source toolchain for ice40 FPGAs. It's really amazing to go from the infuriating to use vendor tools with all their quirks and bugs to a single 'make' that just generates the bitfile without using the broken IDE or complaining about licenses... The FPGAs are also relatively inexpensive ( The tools should also support the more powerful ECP5 FPGAs, but…

Interesting, thank you for the pointer. I've been playing around with FPGAs a bit and this sounds like it might just be what I was looking for. Is there any starter board that you'd recommend for this workflow?

I've only used the bare chips on custom boards, so I have no experience with the eval boards.

A popular board a few years ago was the iCEstick for about 20€, but I can't find it anywhere for that price now. The ICE40 breakout boards from Lattice are also more expensive now for whatever reason. Olimex has a board with the HX8K that is more reasonably priced, but unfortunately doesn't have an onboard programmer...

For the ECP5, the ULX3S board looks interesting. It's not exactly cheap, but also has more features than just buttons, LEDs, and pin headers like on the other boards.

Post reply on HN