12,396 karma · joined October 6, 2010
One-man Factorio is a lot different than big-team Factorio. As a single factory-developer, you need to be startup-like and remember that your time in the present is the most valuable resource.
So always automate everything the first time, because you WILL need more of that component. But do it in a quick and dirty and unscalable manner so you can move on to something else. Start with box-fed production lines; when you're sick and tired of refilling the boxes, it's time to re-do it better.
Build a system to support this throwaway development, where you can tear things down later and replace them and it keeps working. Start with a narrow bus, not eight lanes each for iron and copper — but keep it open for expansion downstream. When you run out of space near the start of the bus, don't just move things a few tiles to squeeze in an incremental upgrade: build a much bigger production line further downstream. You'll have more and better components to do it right.
And always, always, always, keep your interconnects separate from your production components. You don't need a single straight-line bus like some people think, right angles and parallels and grids are fine, but if your belts are snaking through something else, it's real hard to tear down that something-else, and it's real hard to scale up. Encapsulation is important.
And when you're finally scaling up like crazy, ready for the big leagues, use a high-throughput component someone else developed and have your robots install it.
For starters, there's this little thing called "commuting" you may have heard of. Ordinary people, without PhDs, do it plenty. The NYC Pepsi offices seem to be a medium-short walk from Christopher St. PATH; you would have no trouble living in Jersey City or Hoboken on the (computes) ~$2000/mo rent that would be 'affordable' at such pay. I bet you get tax-deductible transit benefits, too, and a monthly pass is around $100.
There are people in New York working full time jobs to make $15/hr, $30k a year. That's a lot more like the "poverty wages" you name.
It's a real system with various security and compliance concerns; Cloudflare and dev-focused services like Inlet or simple SSH forwarding are unfortunately not going to work.
The Jan 6 dudes ("terrorists" and/or otherwise) put their ugly mugs all over Facebook, Twitter, Parler, and the public Internet. I wouldn't exactly take opsec advice from that lot.
So the corner drugstore will have prescription drugs, and over-the-counter drugs, and bottled water, snacks, school supplies, greeting cards, basic climate-specific items (winter gloves, sunglasses and flip flops), shaving, cosmetics and makeup.
And to maximize these economies of scope, they'll also have services: an ATM, Western Union money transfers, film developing, digital-camera printing, passport photo printing, photocopying, package drop-off, maybe even a key duplicator machine... you name it.
You can Google around and find mail order processing for ~$5 a roll, and scan it yourself (flatbed with film-adapter, or for more money, dedicated film-scanning hardware that's usually much higher quality). You can develop black-and-white in a dark room yourself and probably get it cheaper. Not sure HN is the best venue for film-specific info, though; perhaps a film forum somewhere?
If so you will need a massive and bipartisan political effort for it to be so, at least in any US jurisdiction.
But the rule of thumb they tell me is: If you try to use CPUs to move data to all these GPUs, your performance is going to tank. If at all possible, you want tech like RDMA, something that throws bits directly from your filesystems to your GPUs. This will only become more important as GPUs get more capable and data hungry.
Can Cthulix actually control enough I/O to build the future of distributed computing? The approach described wants to explicitly minimize filesystem use in favor of message passing. I’m not sure how that’s supposed to be implemented exactly? but there are only a few ways to get the I/O performance you want for the big heavy GPU workloads, and this doesn’t naively sound like it’s one of them, but maybe that’s because it’s a big question mark.
Not a Bombadil fan, I see!
The reason people forget this is because McCarthyism itself was not actually a very good remedy. It was great for the self-aggrandizement of Senator McCarthy (until he went too far, and his target Mr. Welch called him out while being interrogated before the Senate for making a national show of a very weak question of Welch’s colleague, a question could have been raised far more timely without the spectacle). It was great for building an atmosphere of fear, too. But in the end it was a pretty bad spy eradication program.
(Of course if you like mixed use neighborhoods and don’t care for the overproduced generic pseudo-fanciness of that particular development, you can always just move to New York or something)
Declarative means “some system reads and parses the config, can do clever things to change what happens in different circumstances, show you a preview of what happens, track chances of what the config has been in a straightforward way, audit the config for consistency across several systems or for adherence to selected principles, and do other smart things”. These are possible because the config is a data structure now. You can do things with a data structure you can’t do with a shell script. The script is too opaque and our computers are not yet smart enough to analyze it in these ways.
There are great advantages to the simplicity of imperative configuration scripts, which is why shell scripts will not soon die. There is some peril (shell scripting in particular is full of tricksy traps; alternative languages may mitigate this.) There is great power and some peril in imperative config. It’s a choice.
With an indirect death penalty attached, yes. Can’t think of how this could go wrong at alllllllllllll
HPC is nice, but when you hear Jensen getting really excited, it’s not about dominating some niche like that, it’s about a vision of the shiny sci-fi high tech future, and actually delivering the tech to make it real.
So don’t just look at HPC to understand the NVIDIA ambition. Start at edge computing; imagine a world with ubiquitous autonomous robots (cars and drones and otherwise). Think of the onboard chips driving their vision and speech recognition models: That’s a great place for ARM and NVIDIA chips together, whether as one company or two. Watch a recent keynote and see how all the rest of the tech fits into place as part of that: 5G signal processing chips, for instance, something you might gloss over if you’re not in telecom. You don’t need a roadmap to see how it is all connected in support of this world of the future.
(I certainly don’t have the roadmap, either, I just watch the keynotes and help shuffle bits.)
i thought it worth mentioning
I don’t see the distinction? NVIDIA is fabless too??
(Heck, so is AMD I think)
Yes. One is investing, and the other is speculative grift. And if you're not utterly ruined then you may very well end up with substantially more money than I have.
In the event that this happens, I hope you find that the money is its own reward, because that will be the full extent of your reward. Certainly there will be no special plaudits from society in general.
i don't know if the crypto and meme-stock worlds realize this, but... out here in the real world we have tools to correct for that; you can compute things like a "real rate of return" on your holdings and see it come up positive, because... well, I know this will seem fantastical to crypto-land especially, but imagine having an asset that actually does something useful while you own it, making income, instead of just existing for speculation!!! it's much better than loaning it out through dodgy exchanges for the purposes of speculation.
we actually have nice things like this in the real world, and when we go through routine turbulence (like the last month) our losses are more like 10% of holdings, instead of, like, 50%.
The idea that this specifically matters surely arises from the same logic that thinks certain companies are worth more or more performant because their per-share price is higher.
Crops fail. Promising biotech turns out to be impossible; new drugs are shown to be ineffective. The housing market turns out not-to-go-up-for-ever. A drought in Taiwan stifles semiconductor manufacturing and good chips are hard to find for years; ASML's factory catches fire. Oil spills, and must be cleaned up; Union Carbide leaks gas all over Bhopal. COVID-19 kills people, makes others (rationally) fear for their health, and far less gets done.
Is that ridiculous? Hmm. Maybe try again. We can afford to be generous, after all! Let's say that it's 10% due to monetary policy, and 265% due to other factors.
... Nope, we're still not really there.
Either way, this is crazy money. $40k is about a year’s salary for the average American, and more than a year for the rest of the world. Yes, you’ll save a bit on gas, but still!