gateorade

I'm not sure this is obvious. For the consumer market, maybe. But at the end of the day, there are physical limits on how much compute you can squeeze into any given area. If your job requires more parallel compute than can reasonably be squeezed into a package for power/thermal reasons, then distributing the workload off-package is a requirement.

Maybe hardware will morph into a more heterogenous mix than we have now, with many single-package cpu-gpu nodes working in parallel, instead of a few cpus orchestrating a gigantic sea of GPUs, but maybe not.

I'm more convinced that this is just one stage of the cycle. GPUs in consumer hardware weren't really a thing before the 90s. From the 90s to like 2013 graphics software (and it's ability to load hardware) improved rapidly. Since then it's kind of stagnated and a lot more of the focus has shifted to doing the same amount of rendering with less power/less heat/less space. Even if we do see a shift toward SoCs/APUs/whatever in consumer hardware over the next few years, I'd bet on some sort of paradigm shift to come along (fully raytraced rendering?) and swing the pendulum pack toward big discrete GPUs at some point again.

This is a nice anecdote, but it doesn’t mean anything.

In the US at least, being raise by a single parent is a massive statistical indicator for basically every negative social outcome. Poverty, criminality, future generational single parenthood, etc.

This is basically the case for any digitally controlled effector/sensor. Need a serial bus to get data from some ADCs? An FPGA can do N buses in parallel. Need a PWM to control a motor? An FPGA can do N PWMs in parallel.

The problem of course is that the effector/sensor is usually still a physical object that has a cost to replicate, even though the control circuitry in the FPGA is free to replicate. As is the case with ultrasonic transducers.

Just spitballing here but it might have to do with the fact that materials that absorb light are more heat conductive than materials that are reflective.

The SR71 was famously painted black as a mitigation for surface heating issues because the black paint conducted heat away from the areas the were really susceptible to ram-heating.

Perhaps reflective surfaces reflect some percentage of the incoming energy away, but thermally conductive surfaces conduct a larger percentage of that energy away and are able to safely sink it into some thermal mass.

I’m generally skeptical of pharmaceutical weight loss solutions as a bandaid over the underlying problem, but for this statement to mean anything you’d have to compare it to traditional methods of weight loss.

If you simply eat at a calorie deficit and aren’t taking specific steps to preserve muscle (resistance training and high protein), a lot of the weight you lose will be muscle mass.

This is kind of weird because there’s a lot more data to suggest now that dietary fats are almost never the problem when it comes to obesity, and in fact dietary fats are really important for a lot of critical processes (brain function, neuroprotection, hormone synthesis). I get it for caloric restriction though since fat is denser than protein or carbs.

It's complicated. Reducing calorie intake is absolutely a path to weight loss, but we tend to ignore the adaptability of the body. If you suddenly reduce your caloric intake, the body thinks it's entering into a time of scarcity and will lower its metabolic rate through hormone modulation and by shedding muscle.

This is one reason why people struggle so much with dieting. They lose lot's of weight at first, but then plateau when the body adapts. When they get discouraged and fall off the diet, they put the weight back on quickly because they now are burning less calories than when they started.

The most effective way of losing weight long term is likely some combination of a slight caloric deficit paired with resistance training (to discourage the body from shedding metabolically expensive tissue) and short bouts of reverse dieting . (ie. slight deficit for 3 weeks, then 1 week of a slight surplus, etc.)

This is largely incorrect, and also not what the person you're replying too is saying. Calorie numbers in food are just a weighted sum of the grams of carbs, fats, and proteins in the food. Each of those groups are assigned a fixed weight based on an estimate of how many usable calories are in a gram of each.

The commenter above is talking about how we often ignore the adaptability of the body, and that if you suddenly cut caloric intake by say 300 a day, the body will surprisingly quickly adapt by lowering metabolic rate to compensate. This can be done through the simple modulation of hormones and may also involve the shedding of metabolically expensive tissue, like muscle.

compute in memory

I agree with this and would even go as far as to say color E-INK screens are basically ready to push out greyscale ones right now. The problem is that no one is making a device with them that's any good. I owned one of the boox color e-ink devices for like a week and the experience was awful because it's a full android os shoehorned into something that kind of works on an e-ink display.

Please stop. Doing this the first week ChatGPT was released was cute. Now it's so so tired.

Yeah that’s basically my point. The hype on HN/Twitter/etc. forget this.

This has been my experience. I’m really impressed by how well GPT-4 seems to be able to interpolate between problems heavily represented in the training data to create what feels like novelty, eg. Creating a combination of pong and conway’s game of life, but it doesn’t seem to be good at extrapolation.

The type of work I do is highly niche. I’ve recently been working on a specific problem for which there are probably only a hundred at most implementations running on production systems, all of them highly proprietary. I would be surprised if there were any implementations in GPTs training set. With that said, this problem is not actually that complicated. A rudimentary implementation can be done in ~100 lines of code.

I asked GPT-4 to write me an implementation. It knew a decent amount about the problem (probably from Wikipedia). If it was actually capable of something close to reasoning it should have been able to write an implementation, but when it actually started writing code it was reluctant to write more than a skeleton. When I pushed it to implement specific details it completely fell apart and started hallucinating. When I gave it specific information about what it was doing wrong it acknowledged that it made a mistake and simply gave me a new equally wrong hallucination.

The experience calmed my existential fears about my job being taken by AI.

Why would you start a company that is entirely dependent on a third party’s closed product, over which you have zero control? This just feels immensely risky.

Weird question, but if you encourage bees to come through these methods, how prevalent are they? I like the idea of doing this, but I also have a young child and am somewhat allergic to bee stings myself and wouldn't want my yard to become overrun with bees.