@soft_fox_lad well, i guess *really* really, they do different things; you could imagine doing both fat binaries (which are a developer convenience for easily distributing programs) and Arm64X (which enables the plugin compatibility/optimized OS libaries/incremental port scenarios)
@Colonthreee JIT knows how to make `&&` branchless, but doesn't, because it thinks `!three` is much more expensive than it is:
> Skipping if-conversion that will evaluate RHS unconditionally at costs 11,3
Though, this pass would result in `cmov`, which is strictly worse than the `&` I think
@Colonthreee `&&` is short-circuiting, so Roslyn does not emit branchless IL (unless the operand is a simple variable read, which is why the `!` breaks it). Using `&` instead is a simple workaround.
@xXshaurizardXx@lauriewired as well as new APIs to request that Windows audio engine uses smaller buffer sizes (e.g. 2 ms instead of 10 ms). the article claims that ~10ms roundtrip latency ( audio pulse generated, played, detected by microphone) is possible in WASAPI shared mode. I think that's pretty good!
@xXshaurizardXx@lauriewired There was substantial investment in Windows 10 https://t.co/8bsxTh6eX2
including free performance improvements:
> All applications that use audio will see a 4.5-16 ms reduction in round-trip latency ... without any code changes or driver updates
@xXshaurizardXx I'm interested in making our power efficiency better. I did some rudimentary profiling; comparing Zed and Sublime Text w/ hw accel, our CPU time is ~double or triple, and GPU time is the same (fill-rate bound? I don't have access to a great profiler). I think we stack up OK :-)
super excited to be joining the @zeddotdev team! i'll be focusing on fixing our top-upvoted community issues, rounding out our Windows support, and more!
give Zed a spin and let us know what you think!
@soft_fox_lad social media trains you to be incurious because uninformed, unjustified rageposting gets more engagement than levelheaded discussions that acknowledge tradeoffs. extremely frustrating!
Something that's always upset me is how almost nobody who cares a lot about a thing (whether it be college admissions, skincare, or something else) will actually bothering trying to understand how that thing actually works.
CourtListener & Google Scholar/Patents are right there.
You can understand the world better than 95% of people automatically if you're willing to spend more than 15 seconds looking into a subject you encounter before forming an opinion
am i holding it wrong or is π¦ just less fun than twitter? my feed there is almost entirely neurotic handwringing and dooming about how bad things are. and, like, they are, but the discourse about it on here feels way fresher and more vital. the other site just makes me feel ill
@rieszspieces@numfuksteinstan well, in fairness. this is just the first question you ask so you know whether to waste your time on the hard question
but yea it filters out a ton of people
this guy has obviously been duped into thinking that concerns about conservative extremism are overrated by not realizing that even the vilest people in the world won't likely be rude in day-to-day interactions with an old, sympathetic white man
disqualifying for a "sociologist"
it should be embarrassing to be a sociology professor & tell your students
"trump's election has empowered his base, but this is good since it'll incentivize them to be *less* extreme & will open more opportunities for dialogue. of course, that's only if the left will cooperate"
this guy fancies himself an expert in empathy & someone who, unlike intolerant extremists, listens with an open mind to people on both sides
understanding opposing viewpoints is a good thing. developing the skill to do this *rigorously* is hard & being bad at it makes you dumber