embedded machine learning research engineer - georgist - urbanist - environmentalist

  • 0 Posts
  • 43 Comments
Joined 2 years ago
cake
Cake day: June 22nd, 2023

help-circle


  • The raison d’être for RISC-V is domain-specific architecture. Currently, computational demands are growing exponentially (especially with AI), but Moore’s Law is ending, which means we can no longer meet our computational demands by scaling single-core speed on general-purpose CPUs. Instead, we are needing to create custom architectures for handling particular computational loads to eke out more performance. Things like NPUs, TPUs, etc.

    The trouble is designing and producing these domain-specific architectures is expensive af, especially given the closed-source nature of computer hardware at the moment. And all that time, effort, and money just to produce a niche chip used for a niche application? The economics don’t economic.

    But with an open ISA like RISC-V, it’s both possible and legal to do things like create an open-source chip design and put it on GitHub. In fact, several of those exist already. This significantly lowers the costs of designing domain-specific architectures, as you can now just fork an existing chip and make some domain-specific modifications/additions. A great example of this is PERCIVAL: Open-Source Posit RISC-V Core with Quire Capability. You could clone their repo and spin up their custom RISC-V posit chip on an FPGA today if you wanted to.



  • NYC itself doesn’t have much (although it still has some! see image below) low-density zoning, but their suburbs sure do. The city itself also has a lot of other bureaucratic barriers to development that result in it having abysmal housing construction rates.

    As for vacancy, yes, the threat of not being able to sell is what stops builders from building too much. For example, it’s the reason no one’s even trying to build the Burj Khalifa in Bakersfield. But if you make it legal and reasonably easy to build, yes, people will build.

    Perhaps Tokyo is the best example. Biggest city in the world, and yet it’s actually relatively affordable, thanks largely to good land use policy:

    In the past half century, by investing in transit and allowing development, the city has added more housing units than the total number of units in New York City. It has remained affordable by becoming the world’s largest city. It has become the world’s largest city by remaining affordable.

    Two full-time workers earning Tokyo’s minimum wage can comfortably afford the average rent for a two-bedroom apartment in six of the city’s 23 wards. By contrast, two people working minimum-wage jobs cannot afford the average rent for a two-bedroom apartment in any of the 23 counties in the New York metropolitan area.

    In Tokyo, by contrast, there is little public or subsidized housing. Instead, the government has focused on making it easy for developers to build. A national zoning law, for example, sharply limits the ability of local governments to impede development. Instead of allowing the people who live in a neighborhood to prevent others from living there, Japan has shifted decision-making to the representatives of the entire population, allowing a better balance between the interests of current residents and of everyone who might live in that place. Small apartment buildings can be built almost anywhere, and larger structures are allowed on a vast majority of urban land. Even in areas designated for offices, homes are permitted. After Tokyo’s office market crashed in the 1990s, developers started building apartments on land they had purchased for office buildings.

    I think the key idea is to not have government bureaucrats or existing homeowners or landlords decide whether there’s “enough” housing, but rather let builders determine if there’s unmet demand. If there is unmet demand, they will build if you let them. If there truly is enough housing in a certain city, then you don’t need to tell builders not to build – they’ll simply stop building if they sense there’s not enough demand for it.


  • The “we have enough homes already” is a common (and unfortunately very harmful) myth.

    A couple good in-depth videos on the topic:

    The gist of it is that statistics on how many vacant homes exist are highly misleading, for two main reasons:

    1. Many of the homes are not where the demand is. A vacant home in St Louis does nothing to help with a housing shortage in NYC. People want to live in NYC because that’s where the jobs are. A house in St Louis isn’t worth much if you can’t find work there. And statistics consistently show that the most expensive cities have the lowest vacancy rates.
    2. A lot of the homes that are counted as “vacant” aren’t actually just free for the taking like “vacant” would have you believe. In these statistics, “vacant” can mean: 1) a unit that is between tenants, 2) a unit that just finished being built and is awaiting its tenant’s move-in, 3) a unit occupied by someone who doesn’t legally state it as their primary residence (e.g., student housing where the student still lists their parents’ home as their primary address), 4) a unit in horrible disrepair that is unfit for occupation, etc.

    Add to this the fact that high vacancy rates are GOOD for you, as it means landlords and sellers have a credible threat of vacancy, meaning they can’t demand ludicrous prices. Reducing vacancy rates is an incredibly anti-consumer, pro-landlord move.


  • Yeah, political opinions based on “regulations always good” or “regulations always bad” are lazy and unhelpful. For one, it ignores that many regulations are written for the express purpose of manufacturing or solidifying a monopoly.

    Regulatory capture

    And NIMBY land use policies really are just a textbook example of regulatory capture. Homeowners, who expect their homes to perpetually increase in value, lobby their local governments to manufacture an artificial scarcity of housing so as to drive their property values to the moon. All of this at the expense of renters and new home buyers.

    Imo, we should all be trying to form nuanced political opinions where we judge policy on whether it’s good policy or not.



  • Sounds similar to some of the research my sister has done in her PhD so far. As I understand, she had a bunch of snapshots of proteins from a cryo electron microscope, but these snapshots are 2D. She used ML to construct 3D shapes of different types of proteins. And finding the shape of a protein is important because the shape defines the function. It’s crazy stuff that would be ludicrously difficult and time-consuming to try to do manually.






  • I moved from California to Montreal a few years back to study, and now I’m staying for good. I tried duolingo on and off for far too long, but I found it super uninteresting and hard to remain committed to.

    Best strategy I’ve found is called comprehensible input. The idea is to find books or other reading material that you can get the basic gist of when reading, despite not understanding every single word and phrase and grammatical construction. The more you read, the more you’ll find yourself able to understand, which is also very motivsting!

    Also, make sure it’s material that actually interests you. The idea is it’s better to read extensively, reading things that actually interest you to some degree and keep you mentally engaged, than to just really intensively study a much smaller amount of (less interesting) material.

    This actually mirrors how we acquire languge. The idea is to intuitively understand French by having seen a lot of it rather than to basically memorize French. You ultimately want to be able to glance at a sign, for instance, and just know what it means without having to translate in your head.

    Some resources I found useful were these French illustrated books in Dollarama, but even better is a series of books designed to be comprehensible input by Olly Richards. He’s a native English speaker and polyglot who has written a bunch of graded readers that gradually increase in vocabulary and difficulty. He has several books for French, including beginner short stories, intermediate short stories, beginner conversations, intermediate conversations, climate change, WW2, and philosophy. The nice thing is he actually does a good job of making the stories and content interesting to an adult learner, unlike the children’s books at Dollarama.

    Even his beginner books might be a little too advanced for your level so far, though, from what you say. If they are, it’d be best to find some material at a lower level that you can understand a little better. After all, if it’s too hard for you, it will make the process much slower and less enjoyable, which will make it much more likely that you quit. You could even simply try googling “french comprehensible input” to try to find material suitable for your level.

    One last resource is the government of Quebec offers free in-person courses for immigrants and many French learners. They are part-time, and they offer multiple options for hours per week, so you could pick what works best for you. It would be worth checking to see if you might qualify for those courses once you move here.


  • It’s so clear that the GOP is flailing. They’ve only won a single presidential popular vote since 1988. Old white people are dying and brown kids are reaching voting age. Young people at large are overwhelmingly fed up with the GOP. They are losing the demographics game big time, and I think they know it.

    And I think the GOP base knows it deep down, too, and I think Obama was really emblematic of it. Suddenly, they had to wake up to a country that was rapidly shifting, a black man was president, gay people were getting rights and becoming broadly accepted in society, people were starting to talk about racism as an actual problem again, movies and advertising were getting more diversity, trans people were getting rights… And I think this deeply unsettled a significant chunk of the population who felt like their control over America was waning, hence the Tea Party, hence the MAGA movement.

    This fascistic movement we’re seeing from the GOP isn’t logical at all. It’s losing them elections left and right, but if they don’t pursue this, they lose all their primaries, because this movement is about the Republican base lashing out over them losing demographically.



  • Yeah, I’m working in embedded ML, and it’s an insanely exciting time. We’re getting more and more microcontrollers and single-board computers with special AI accelerators, many of them RISC-V, by the day it seems. One of the next steps (in my opinion) is finding a good way to program them that doesn’t involve C/C++ (very fast but also so painful to do AI with) or Python (slow unless it’s wrapping underlying C code, and unsuitable for microcontrollers). In fact, that’s exactly what I’m working on right now as a side project.

    What’s also cool is RISC-V promises to be the one instruction set architecture to rule them all. So instead of having PCs as x86, phones and microcontrollers as ARM, then all sorts of other custom architectures like DSPs (digital signal processors), NPUs, etc., we could just have RISC-V with a bunch of open standard extensions. Want vector instructions? Well, here’s a ratified open standard for vector instructions. Want SIMD instructions? Congrats, here’s another ratified open standard.

    And all these standards mean it will make it so much easier for the compiler people to provide support for new chips. A day not too long from now, I imagine it will become almost trivial to compile programs that can accelerate tons of scientific, numerical, and AI workloads onto RISC-V vector instructions. Currently, we’re stuck using GPUs for everything that needs parallelization, even though they’re far from the easiest or most optimal devices for many of our computational needs.

    As computing advances, we can just create and ratify new open standards. Tired of floating point numbers? You could create a proposal for a standard posit extension today if you wanted to, then fork LLVM or GCC or something to provide the software support as well. In fact, someone already has implemented an open-source RISC-V chip with posit arithmetic and made a fork of LLVM to support it. You could fire it up on an FPGA right now if you wanted.



  • It’s especially dumb because RISC-V is – dare I say it – inevitably the future. Trying to crack down on RISC-V is like trying to crack down on Linux or solar photovoltaics or wind turbines. That is, you can try to crack down, but the fundamental value proposition is simply too good. All you’ll achieve in cracking down is hurting yourself while everyone else gets ahead.