adgjlsfhk1

That doesn't help memory latency and most of the time a cpu is waiting it's because of latency, not bandwidth.

The other really complicated part is that it is very common to have intermediary ints that have very little relation to the ptr that they come from/turn into (e.g. XOR linked lists are probably legal even though they have ints that mix provenance of 2 completely separate pointers).

C++ memory provenience is very complicated (and possibly unsound) because it lets you do all sorts of complicated things like this.

C is definitely not the most weakly typed (e.g. Javascript/PHP are both weaker), but IMO C is weaker than python in that python will give you a runtime type error if you do something invalid, while in C you cast something and just get UB. Most higher level strongly typed languages are able to make their casts checked because they have generics to remove the need for 99% of them, but C is missing those features and thus has to rely much more heavily on unchecked and unsafe runtime casts.

not in a way that forces the user to actually handle them properly. C is a weakly typed language and as such is pretty badly suited to systems programming.

it's a mixture of experts so it's somewhat viable to stream weights from nvme

Intel's newest gen is a pretty clear favorite for laptops (especially with their better wifi/bluetooth/USB chipsets). Outside of laptop though, Intel has nothing compelling.

with ai their future seems very uncertain. ports are now very easy, so languages have to stand in their own merits

Not really. The main purpose of this is to run CI that never was going to be a personal machine anyway. This is just making it difficult for devs to have automated testing of their software on macs.

If they can sell them for higher margin than a mac mini/macbook it seems sensible enough.

I feel like the ideal would be a 2U, 8 mac mini with dual 10Gb ethernet, shared dual power supplies, actual datacenter heatsink layout would be such a great CI platform.

A lot of the complication is reduced by the fact that modern air to air missiles have 100+ mile ranges (so you don't need a traditional intercept). You also don't need 100% coverage. Rather than covering all of Ukraine 24/7, you only need to cover an unpredictable part of it such that flying is risky everywhere. Getting basing in Poland (or Keiv) would obviously help reduce the flight time and awacs amount needed.

To be clear, I'm not saying it would be politically easy or low risk. It would almost certainly result in a couple fighter planes being shot down.

Presumably a combination of SAMs and fighter planes. The US couldn't put a no fly zone over Russia, but a no fly zone over Ukraine would be pretty straightforward (even just flying out of Germany). A no fly zone doesn't need to be perfect to work. If you shoot down a couple planes (or even have fighters in the area), you make every future flight by Russia a massive risk.

Most of that is likely retirement savings. That's 3.5k per year if you have employee match.

28*13=364. Months are different sizes so month*12 would be an over-estimate

H.264 (if you want to support 20 year old devices) or av1. Don't use h.265. The patent situation around it is a total mess.

Bluetooth 5.3 with LE audio would also be very nice.

> What if they get blocked from the network and repeatedly change accounts, IPs and then MACs repeatedly to get around progressively escalating blocks from the network admins?

If you think anyone who makes a new account in a videogame for getting an old account banned should be charged with felony hacking, that's certainly a position you can take...

> What if they run from the police who try to stop them?

Running from police officers isn't generally considered hacking.

no. The DFT result as published uses unlimited precision, but the comparison with the int multiplication shows that the unlimited precision just makes the paper shorter/easier rather than being an exploit in the computation model.

> but either it's wrong or the previous results were wrong, no?

Not quite. This result uses polylog(n) precision which isn't bounded, so it escapes the previous bounds. This doesn't noticably affect the runtime because not all of the math needs this extra precision (in this case you need a couple computations in higher precision but less than n operations), so the extra precision can be payed for even in a TM/circuit model.

The DFT result isn't RAM specific. It holds up in multi-tape turing machines. (it is, however, galactic).

Well optimized radix sorts are actually pretty fast (assuming you have the proper fallbacks to quicksort/insertion sort to deal with medium/small sizes)

For long term storage, the right batteries don't exist yet (commercially). The good options are going to be one of sodium ion, flow batteries, or thermal batteries.

The ~grid scale design of this for reference is somewhere around 200mx200mx50m for above ground, or ~100m cube (or bigger) for quarry infill. Such a design is going to get crushed by battery/hydro over the hours/days, but wins on cost for seasonal usage.

Realistic designs for this can achieve 450C delta t. Your energy store can store at ~1000C (slightly below rock melting point), and run steam turbines with a hot side of 600C and a cool side of 150C.

Over the season, your sand drops from ~1000 to ~600 (which is fine since your power needs drop in the spring)

right, but anything linearly reducable to kSAT is trivially sub-exponential, and a discovery like this means that linear reductions are a whole lot more important than they used to be.

Also for NP problems that can't linear reduce to kSAT just having the knowledge that SAT is sub-exponential (and the tools), makes it a lot more likely that we see a whole lot more problems become exponential.

That's not really a problem. A spin up over a couple hours matches nicely with the slow sunset over a continent. The sun takes ~4 hours to go from a bit above the horizon in the east coast to a bit below the horizon on the west coast (solar panels make some power in twilight).

Wind also helps since wind makes more power at night, as does demand which is dramatically lower at night.

For coal, Great. Making coal economical is excellent. We should have banned it worldwide 20 years ago.

the presumption is that these results are a side-effect of reinforcement learning to make the models smarter.

OTOH, it would be surprising if this sort of technique didn't apply to other NP problems.

no. It's huge.