MobiusHorizons

2,239 karmajoined 11 ปีที่แล้ว

comments

เมื่อวานซืนdiscuss

Why do you think it is not slow? As far as I know the only language that compiles slower is C++, and even then the compilation speeds between c++ and rust seem to be comparable. I believe c, Fortran, zig, C#, Java and golang are all faster compiling languages. That makes rust pretty slow in my book. I get that it doesn’t bother everyone, but that doesn’t change the facts.

3 วันที่ผ่านมาdiscuss

Small note of clarification. HF in a radio frequency (RF) context actually refers to what we would now consider fairly low frequencies between 3 and 30mhz where wavelengths are in the tens of meters. The black magic is mostly in UHF AND microwave regions where wavelengths start to measure closer to the size of circuit elements.

Although I have also heard it said “RF physics is black magic” to cover all the bases

3 วันที่ผ่านมาdiscuss

Spend used as a noun is a common finance term. I also hear “spend rate” used similarly.

It would be weird (but not necessarily wrong) to use that word as a noun outside of a business finance context.

5 วันที่ผ่านมาdiscuss

I had windows 2000 on a laptop in the 2008 2010 era. It was already old at the time, but I actually preferred it to XP and certainly vista. You get NT which is nice compared to dos based 95/98/me but you get those tried and true aesthetics, which work well. I’m not sure I’d go back to it now, but it definitely sticks out as the high point of windows in that time period.

6 วันที่ผ่านมาdiscuss

If they are using the typical human hair = one thou, then the 2500 is suspiciously close to the conversion between inches and millimeters times a power of ten. I get 10 microns

6 วันที่ผ่านมาdiscuss

> How is it eliminating market competition?

Verizon is the company powering the service for these watches, and is the company not sending 2fa to non-Verizon customers.

16 วันที่ผ่านมาdiscuss

Windows are harder to insulate than walls anyway, but the most important insulation is in the roof because heat rises, so skylights would leak even more heat than normal windows already do.

16 วันที่ผ่านมาdiscuss

> Moving beyond ARM to RISC-V

The reason to move away from arm has nothing to do with performance, but rather avoiding licensing snafus like happened with their laptop chips. So far no one has delivered a risc-v core with class leading performance outside of the really low end. Not saying it can’t be done, but it will likely be a step back at first.

16 วันที่ผ่านมาdiscuss

Roughly interchangeable with egg head I think, although more used and slightly more endearing.

17 วันที่ผ่านมาdiscuss

Yes heat pumps move heat that already existed from the cold side to the hot side, but they also consume some energy to fight entropy, meaning they pump more heat to the hot side than they remove from the cold side. This is a net heat gain, equivalent to the energy consumed in running the AC. The value may be considered negligible compared to other sources, but it can still be on the order of 500w per room, which adds up quickly if everyone is doing it.

Of course air conditioning is reasonably well suited to be a solar load during peak hours, but in most parts of the world if everyone just installed AC units like are common in many parts of the US it would mean a huge amount of extra fossil fuels burnt.

17 วันที่ผ่านมาdiscuss

I think we agree on that. Special interest groups proactively push for things, they don't just opt out and hope people notice. I am arguing for lobbying and special interest groups (that's why I suggested the EFF as an example, although certainly not the only one).

18 วันที่ผ่านมาdiscuss

When has this tactic (opting out to force change) ever worked on problems of any meaningful scale before? Please don’t say the civil rights movement, because those boycotts were not effective on their own. Pressure is required to force change, and opting out does not produce the necessary pressure unless it is genuinely a significant portion of users and lasts years.

These strategies have been tried time and time again and have proved not to be strong enough to overcome economic forces.

18 วันที่ผ่านมาdiscuss

Compared to “resist by opting out”, the EFF has been much more successful. So far I’m not aware of any meaningful change in tech enshitification from grassroots opting out (could certainly be missing some instances, though, happy to be told about any). The EFF has at least had some notable wins in crypto export, fighting copyright enforcement abuses, limiting certain types of surveillance (license plate readers in California for example), and right to repair.

I’m not saying they have been enough to prevent things from getting worse, but they have actually done things that have staying power, and affect policy to this day, which is better than the track record of opting out as far as I know.

19 วันที่ผ่านมาdiscuss

Are you really advocating for suppressing rational assessments for the likelihood of success because you think the analysis is too discouraging?

If you already agree the resistance will ultimately lead nowhere, why not focus that energy on something with a better chance of success? Best guess would be partnering with someone like the EFF for a solution through lobbying And the courts.

20 วันที่ผ่านมาdiscuss

In short modern CPUs can’t actually execute control flow very fast, so they usually cheat and memorize where the program is likely to go next. This works great with things like the loop branch, since most loops are pretty long the cpu can start on the next iteration of the loop while it waits for the condition to finish being calculated. You incur one branch misprediction at the end, but the rest of the loop is fast. But with a bytecode vm, the action of the code is entirely dependent on the bytecode instructions, so it’s not predictable unless those instructions themselves are predictable. That’s why the switch statement is slow. Explaining the optimization from computed gotos is more complicated.

As I understand it, the branch predictor stores its predictions using the address of the branch instruction as a key. In the switch implementation the unpredictable branch is the one that jumps to the specific case. The compiler emits an indirect branch for this just like the computed goto instruction. But it only emits one indirect branch instruction, so that branch is always unpredictable. In the computed goto version, there is no loop, each case statement ends in its own indirect branch, so each bytecode implementation is followed by a branch with a unique address. In practice this makes each of them slightly more predictable, because you only have to predict what comes after an inc instruction instead of what comes next after any instruction. Basically there are more branches to hang predictions on.

20 วันที่ผ่านมาdiscuss

The article and this thread are about shaheed style drones.

As far as I know, stopping fiber optic fpv drones leans more on physical barriers that catch the fiber (eg road nets) rather than trying to detect and destroy the drone. It’s usually too late by the time you can hear or see a drone that size.

22 วันที่ผ่านมาdiscuss

How does that help? Seems like mostly the end user suffers.

22 วันที่ผ่านมาdiscuss

Renewables have been incredibly effective at providing more capacity at good cost, but they still have drawbacks. Aside from the obvious variability everyone harps on, location does play a nontrivial role. Switzerland is not as good a candidate for solar or wind as many other countries due to constraints in usable land and availability (valleys have more shade than steppes for instance). I don’t think anyone needs to be saying no to renewables when they consider investing in nuclear. The two technologies serve very different purposes in an electric grid. Base load is still a problem even with battery storage, and it makes a lot of sense to have reliable power available that is not from fossil fuels to deal with that demand. There is plenty of fossil fuel based generation to replace, it doesn’t have to all be nuclear or renewables, in fact all one or the other doesn’t make much sense.

It’s true nuclear is expensive. It’s also true a lot of that expense was the regulation that has made it so incredibly safe. It’s unlikely a cheaper but still safe enough compromise will be reached in countries with existing regulatory structures quickly, but there is certainly room to improve over time.

23 วันที่ผ่านมาdiscuss

to be fair, if you just want to improve the UX for yourself you can totally have a fork, make your changes, and use it on your own machines when talking to github or whatever other git repository. Now getting someone else to accept your changes might be harder, but that's what happens when you try to change a tool so many people use, especially one with a lot of history. Maintaining a fork is literally one of the things git was designed to make feasible after all.

28 วันที่ผ่านมาdiscuss

If open source AI was better than what it is currently chasing, wouldn’t that take away the incentive for these companies to give it away for free? Training is expensive and companies will need to recoup those development costs once it stops being about jockeying for position.