Stating compute scale in terms of power consumption is exactly how one looks at data centers or capacity planning right now though. It's the major constraint.
Reminds me of those couple of guys who made a bicycle with backwards controls. They learned how to use it, which then rendered then unable to use normal bicycles lol.
Yea 90% is a more realistic fail rate of my 2am ideas which seem great at 2am, but then terrible a few days later with good sleep. If GP is batting almost .400 for insomnia fever ideas, that sounds pretty stellar to me.
Reminds me of playing Tribes. First start playing, can't hit anything ever because things move too fast and your brain doesn't get the physics. But then you're hitting stuff zipping around at crazy speeds by just kinda intuiting.
Guess we're evolved to throw spears, so we're good at that kind of thing.
I guess that depends on the country? In the US, motorist fatalities from crashes outnumber pedestrian/bicycle/etc fatalities like 4:1, I think? I guess that includes both motorist killing self or occupants and also motorist killing other motorists.
In the US there are probably 1000x miles traveled in cars vs on bicycles or as pedestrians everywhere except probably the top 10 metro areas. That the casualty rates are only 4:1 shows the danger that cars pose to non-car road users.
The metric you would need for this is "fatalities and crashes caused because someone was struggling to deal with their bloody touchscreen", which can be both motorists and non-motorists. I don't think anyone is tracking that.
Ah very cool. I thought this was going to be a tongue-in-cheek article about the proliferation of random corporate-mandated CPU hogs like croudstrike or w/e.
Pleasantly surprised by the technical article :). Reminds me of a lot of precise-number-of-idle-cycles functions from embedded world in a prior life
Likewise. Wasting user space CPU cycles is actually a moderately tricky thing to do when we have optimising compilers. Its not something that comes up a lot but it is occasionally something you do need to do in order to test other things and it is different to making a lot of system calls.
I think volatile is usually fairly well contained I can't immediately think of an obvious case where it could adjust the performance of the calling context indirectly but its pushing the problem onto that cache coherence part of the CPU so its not quite the same as an algorithm that consumes all the memory bandwidth or execution ports. Lots of little interesting trade offs in the weeds of this and other potential solutions depending on if any of this matters.
Making games in rust in retirement is my plan. Not, you know, good games (zero game dev experience), or games that anyone would realistically want to play, but seems like it would be interesting to learn.
It's really fulfilling to be able to show people your work and have them play with it. It's so different than like.. spec'ing out a new database schema and then building some APIs over it. They're both coding, but one's a little harder to have a convo about at the dinner table.
Rust is such a mature language to use coming from a JavaScript background. I don't think it makes the best language for writing good games because it's too challenging to write bad prototypes you intend to throw away. You have to refactor frequently and code-compile-run loop is so slow. The lack of quick prototyping discourages me from playing around with ideas that might not work out and that makes for a worse game. However, as a programmer, Rust is an incredibly satisfying language to write in. Everything you do always feels very technically correct. The Rust quip that "if it compiles then it probably works" is very accurate and is a continuous source of pleasure.
I haven't looked into it seriously, no. Scripting support in Bevy (the ECS framework I was exploring) is pretty limited. There's an add-on that can provide unofficial support for LUA (https://github.com/makspll/bevy_mod_scripting) but I doubt that would be compatible with a WASM build target.
>I doubt that would be compatible with a WASM build target.
In principle, I don't see why not. In practice, yeah I can see it. In both cases, can't know without checking, and I haven't. Bevy is pretty cool though!
Maybe one of those would be minimal enough for the rest of the engine to not manage to get your way, as frameworks are known to do. As always, happy hacking!
As someone coming on 30 months of retirement, I'm consciously shifting away from toy projects to projects where I commit to a more polished deliverable.
It helps with commitment and pursuing a deeper learning of the activity instead of doing quick and dirty stuff in my experience. Just don't expect it (or aim for it) to be a steam top-seller, my aim is usually to have at least one other stranger get some amount of value out of what I produce.
Not to say there isn't a place for quick and dirty projects, of course. Bespoke 3D models to fix things around the house are my current favourite category for that.
I’d like to invest time into the Linux gaming ecosystem. Though it’s a little daunting that so much of the work would need to be done by Nvidia and major game studios that don’t really care.
Windows just feels irredeemably mediocre at this point. Maybe Windows 12 will improve things, but I’ve been pretty down on 11.
I'll see how I really feel when it happens but when I retire I don't want to look at a computer for a very long time. I suppose after some time I might get drawn back but I plan to de-tech my life as much as I can.
"irreparable harm" in this case is a legal term of art, which sort of translates to "cannot be fixed by *any* amount of money later". If you lose a job for 5 years, work a minimum wage burger flipper job, and then win a $100 million judgement, you're more or less made whole in the eyes of the court.
So losing a grant is probably more along those lines, in the context of "irreparable harm" for an injunction.
You could make the argument (and I'm guessing it was?) that for scientific grants specifically, if the goal isn't money in the first place, and the lack of grant makes a scientific career impossible to fix later even with any amount of money (say grants 10x as large), then maybe you meet "irreparable harm"? I don't know if the courts would buy that.
Spot on. And the inverse - forcing the govt to dispense the grant is not reversible. Once that money is spent in buying equipment, salaries etc. it's not coming back.
> You could make the argument (and I'm guessing it was?) that for scientific grants specifically, if the goal isn't money in the first place, and the lack of grant makes a scientific career impossible to fix later even with any amount of money (say grants 10x as large), then maybe you meet "irreparable harm"?
I think I'd agree with that, yes. I'd even go as far as to argue that you've caused irreparable harm to the public insofar as the grants specifically would have funded open-access publications.
If destroy a department, money does not bring it back to life. It disposes of its materials, disperses its personnel, and loses its laboratory. Unless somebody volunteers to keep your data, it gets lost.
And if you're working in biology, samples and research animals will be literally dead.
Perhaps you could explain one more thing for me. How did the filing lawyer make such a (seemingly) elementary mistake? Why wasn't it caught earlier (by the judge, by their colleagues, by anybody with an interest in this case)?
Um, this seems ass-backwards to me. If some grant is denied when it should've been approved, the only harm that should be considered is "was this money allocated to the research?"... not "was this researcher's career harmed?". We do not allocate grant money to further the career goals of academics. The public and the government that represents it have zero interest in furthering anyone's careers.
In the US legal system in general, the interests of the government are not trump cards. If the law says that the government should pay you $1000, it doesn't matter whether the public or the government feel they have an interest in paying you, nor whether some executive official thinks the original purpose of that law is served by paying you.
In this case, the law does give the executive some discretion to decide based on their priorities whether they really owe you $1000, but the plaintiffs argue the NSF has exceeded that discretion.
> Um, this seems ass-backwards to me. If some grant is denied when it should've been approved, the only harm that should be considered is "was this money allocated to the research?"... not "was this researcher's career harmed?".
Right, exactly. And that seems to be exactly how the court ruled on this injunction. Since "just money" typically can't reach the bar of "irreparable harm", then an injunction that requires meeting the bar of "irreparable harm" is not granted.
The case itself still proceeds, there's just not an immediate injunction granted while the (slow as hell) court proceedings continue.
It's just a different abstraction level.
reply