All boardsБлинчикиInternationalКрендельнаяЛес

#5 SciTech Thread


294 posts and 53 files omitted
See all
#1109090
1440×1440257.00Kb

>>1107957

audience is all thai even in sweden wonder if she ever puts her name in english somewhere would be easier to learn

#1109509
1440×810308.88Kb

updooted monitor firmware to see if i can fix the usb-c wakeup problem.

its possibly fixed maybe, but now theres a wakeup loop problem (like a boot loop but for waking up) a few times a week, and only remedy is taking out the power cable

why can't there be free lunches

#1109813

not sure if should buy ps5 with good discount on 9.9 now or wait for pro

#1110442

safari add to dock is perfect for youtube music

the dumb thing doesn't remember state after you close it but safari only updates during os update so almost never compared to chrome ("plese restart chrome i beg there was update" )

#1110614

>>1110504

yeah, I know, 3 things made me pull the trigger:

- I can't wait 2 more months facepalming myself

- 9.9 discount was pretty good and price of the pro is too high (I bought slim for $380, but pro would be at least $800 here, so 2x+ the price)

- I have only 1080p TV now


maybe I will stop playing it in 1-2 months as I did before, so I saved myself a bit of money that way too

this is mostly cope of course


but 4k60p on 65"+ OLED of course would have been much better, maybe one day

#1110620

>>1110614

>

but 4k60p on 65"+ OLED

probably should buy top PC for that though, because PS5 Pro is still not that good, compared to 4080/4090


i.e. PS5 < PS5 Pro < PC+4090


but PS5 Pro could be Pareto optimal tho

#1110995

nooo why are all the good component libraries react only i dont want to style things from the beginning

#1111336

>>1111335

if that genuinely works consistently i would be very impressed and willing to do some frontend stuff for once

#1114281

>>1114273

>

I don’t have to update rules all that often (usually a few rules/week)

that guy is delusional. or just don't have anything to do in his life except micromanaging computer software

#1114793

>>1114273

>

Since Homebrew is a well-established package manager, macOS recognizes it as a safe source

tried to download standalone yt-dlp but it gets warninged by macos meanwhile brew cli programs run just fine

#1114978

>>1114793

catgpt made this up, instead of this brew uses some magic to unquarantine installed binaries

#1115988
750×74779.78Kb

i wanna get bose qc ultra or sony xm5 how comfortable are noice canceling headphones tho, i dont really care about that feature but all the fancy stuff has it nowadays

#1116112

>>1115988

too much pressure oughta instead maybe get those around ear holders for airpod

#1118647

finally went to the concert but couldn’t get 100% possible experience:

- was pretty close to the scene, but not the closest place

- didn’t shake hand

- didn’t make selfie

- outfit wasn’t very sekshie


why can’t I ever get a perfect experience


at least made pretty good videos with 3x optical zoom, better than compressed youtube videos, I have my personal fancams now

but only few are 4k60, because wasn’t sure if I have enough space on iCloud for that

again, not perfect

#1118649

checked 1080p vs 4k, can see pretty big difference even on a phone screen (when zooming). and, even worse, the best moment I recorded is 1080p


maybe can record 4k30 as default now

#1118697

yeah 4k looks much nicer on a macbook screen when zoomed in, 1080p was a mistake...

not sure about 30fps vs 60fps tho

#1118777

but it's interesting question what's better: 3x optical zoom on 15 pro or 5x on pro max

sometimes you would want 3x, not 5x, because of the distance


on 16 pro and pro max it's always 5x tho

#1118791

>>1118649

>>1118697

>

For a wedding, the vast majority of viewing will be on 60-Hz refresh screens and (at least in North America) you'll have 60-Hz lighting. For these reasons along, you should shoot 30 (or 29.97) fps.


read a bunch of reddit discussions, it seems 4K is a must for a non-garbage video, because you can zoom

but 60fps has some issues like weird light flickering and (presumably) worse dynamic range


so 4k30 seems optimal choice unless you want to slowmo


but with fancams you probably would like to, so not sure

#1120343

https://www.youtube.com/watch?v=d35SlRgVxT8

probably should start using deno instead of the pnpm init && pnpm add @swc-node/register @types/node && cp ~/tsconfig.json ~/.prettierrc.yaml . boilerplate every time I need to work on custom TS script

on the other hand, not sure if deno can work with pnpm but deno + npm would be probably very slow

#1120892

>>1118647

england one? was there a lot of non-thai there, can't imagine its easy to discover her

#1120893

>>1118806

never thought about hz and lighting, i guess its some meme about how fast the current alternates

with leds getting more popular maybe it will go away since they are dc only

#1121750
3840×21601.10Mb

https://busy.bar/ we preordering or what like a hecking fancy programmable digital clock

#1126907

trying apple intelligence on the macbook

#1126910
1536×2048267.64Kb

are we ordering 32 or 64gb minis its a thousand buckeronis more

#1126979

>>1126910

>

64gb

do you want tons of VMs or big LLMs?

I’m not sure mac’s gpus can efficiently speedup popular openweight models though, because everyone uses cuda

#1127021

>>1126994

>

orbstack

doesn't it require subscription? or at least required

https://news.ycombinator.com/item?id=36189550

>

I use Virtualization.framework for OrbStack (https://orbstack.dev)

>

The popular belief is that Virtualization.framework is inherently faster/lighter than other VMMs like QEMU, but I haven't found evidence to back it up. I spent a day prototyping a custom VMM (with an open-source base) and was able to get dynamic memory allocation working [1], as well as faster file sharing in some cases. There's a lot of other things I could do better (faster, simpler, more reliable) with a custom, tightly-integrated replacement.

>

But unfortunately, that's not an option. Apple doesn't allow third-party VMMs to set the necessary CPU flags for Rosetta. There are too many users relying on Rosetta for fast x86 emulation for me to ditch it, despite all its bugs.

>

So yes, I still use Virtualization.framework, but only because I have no choice.


it's just Virtualization.framework tho, nothing magical


>>1126910

>

32 or 64gb minis

you mean m4 mini? but max is 24gb ram

#1127080

>>1126907

>

apple intelligence on the macbook

gosh it's so stupid and useless

the only good feature is smart edit (remove objects, etc) in photos app

#1127153
1920×10805.54Mb00:09

>>1127080

i can imagine

for siri people i can imagine if they made it actually understand natural language that might be cool

muh text editing though who cares

>>1127021

free for casual use but yeah sounds like it needs to phone home for license check anyway comments do say its beyond better than the other options though

#1127379

>>1087079

fone started to disconnect 4g or entire sim unless i take out the sim tray and put it back in

maybe ordering a 13 mini refurbished is the next step to take

#1127686

the new macbook lineup seems pretty good, finally they start with some reasonable base RAM

i might finally make the switch to macbook

#1127797

>>1127686

>

reasonable base RAM

16gb seems a bit small for dev purposes tbh

and why would non-dev by macbook tho. maybe except americans because apple devices are cheap there. but in Europe it's too expensive if you're just typical user of computer

#1127799
1396×476136.44Kb

>>1127080

https://www.reddit.com/r/MacOS/comments/1ged0sm/siri_still_kinda_dumb_on_macos_151_with_apple/

>

gosh it's so stupid and useless

oh, I got it, there's still no new Siri (LLM-based) in the 15.1


need to wait at least 15.2 and it's still not the fully finished version it seems


so all Apple did is just released new UI for the old dumb Siri and called that "Apple Intelligence"

couldn't they update the UI after rolling out the LLM Siri?

#1127801

>>1127021

>

but max is 24gb ram

oh it's actually 32gb max with m4 chip and 64 with m4 pro


10gbe looks like a cool upgrade option tho

also the Thunderbolt 5 120gbps with m4 pro

#1127802

>>1127801

120gbps means you can get full possible speed of the external SSD NVMe, like 8GB/s


stop, just googled, there's pcie gen 5 ssds now, with 14-16GB/s sequential read/write

I want that


120gbps should be able to fit at least 12GB/s with all overhead

#1127808
3072×40962.48Mb

>>1127797

thats for the mb pros with the base m4

the 24 ram starting point for the ones with m4 pro seem reasonable tbh

i wanted to get m3 pro at some point but i couldnt justify buying 18 because it seemed somewhat little, and i couldnt justify 36 because it seemed overpriced for a lot of ram i wouldnt really need i assume


for 2430 euros here the pro with 12 cores starts at 512gb (kinda funny...) but external ssds are a thing and 24 ram

for 2915 euros here the pro with 14 cores starts at 1tb / 24 ram


all apple shit is a ton more expensive in europe but it is what it is


#1127812

>>1127799

yeah they were just rushed into dumping some ai news and it shows, a bubble only lasts for so long

i dont expect any relevant AI features until at least 6 months from now

#1127890

>>1127808

>

the 24 ram starting point for the ones with m4 pro seem reasonable tbh

oh right, 24gb is probably ok too. can't go too crazy with VMs but should be enough for most other tasks


>

but external ssds are a thing

unfortunately apple doesn't want to support anything except things you can buy at apple store. you will suffer a lot unless you go an apple way. e.g.:

- monitors: can't use <4K monitors (e.g. 1440p) because apple removed subpixel antialiasing support

- mouses: you have to install 3rd party apps to fix scrolling because you're supposed to use magic moue

with ssds you're supposed to upgrade internal storage because macos would automatically unmount/eject external SSD when go into sleep and that sucks a lot because you would have to reconnect it

#1127897

>>1127890

i dont see a problem with reconnecting, tbh i never even keep my external ssd plugged in at all times on windows at the moment

#1127900

>>1127897

for me it was a problem because I wanted to do backup every day automatically, of SSD too, and because of auto-eject it didn't always work, and I didn't find how to stop this auto-eject during sleep

#1128014

>>1127797

probably an even better deal for non-devs

a baseline mac will be beyond better than any pc they can buy


my 2000 eur work thinkpaderino (from year of our lord 2023) takes 20 seconds to wake up from "sleep". really its hibernation because the battery would take a massive hit overnight if it really used sleep so it goes to hibernation after 30 minutes of sleep

#1128606

>>1128014

i think that might just be a windows issue, it seems like there can never be laptop windows that can have a proper real sleep mode that wakes up instantly and doesnt just kill the battery

mac is really good at that somehow, i think i shut down/restart my work macbook maybe once every 2 months

#1129213

https://deno.com/blog/jsr_open_beta

do you remember when Ryan copied the package design system from Go and thought it's a good idea (packages are imported via absolute URL from some HTTP host)

"bruh you don't need a package registry because Pike thinks it's 'complicated'"


now Go has go.mod and Deno has deno.json with almost exactly same functionality as all other languages


same as when Pike thought that generics aren't needed but then added them to the langauge anyway, in a shitty form


gosh, Pike is one of the worst language designers I've ever seen, why did dumb Google even let him to design Go, how is he good

now we live in a shitty world with one more shitty popular language because Google can push it

#1129215

why does every language have its own package registry anyway, can't we abstract the language and support multiple languages in the same registry?


or it's just because every language designer wants to build one from scratch because NIH

#1129259

>>1129213

idk about the original version but the current go package management thing is super nice just download the source from the repository, instead of having some middle man registry with mystery files that may or may not be the original source

#1129331

>>1129259

>

having some middle man registry

how is github not a middle man?

npm is even owned by github

#1129334

>>1129261

isn't it a known fact that the base model of most Apple products is the best value?

same as gateway drug basically

#1129521

>>1129331

middle mane between dependency source code and the user


really does make me wonder whats the advantage of crates.io its just extra fluff. at least javascript being a total mess kinda justifies npm what with requiring a bunch of different artifacts to support the various js and ts module types, tho even those would be better to commit to the repo

#1129785

>>1129261

everyone really praising that base model tho and to be honest it is probably some very good value

#1129821

>>1127801

looking into 10gb since im considering a mini but no matter how much you pay for turris or unifi a 10gb router doesnt even exist

#1130279

bought 2nd dualsense controller, official charger for controllers and big TV

PS5 Slim now has much bigger discount compared to what I had 2 months ago tho


also TV isn't the best (good price/value tho) and I don't have the PS5 Pro

maybe can upgrade everything to the max next year...


can upgrade to Netflix 4K plan now tho

#1130579

>>1129261

can't choose

if you run out of memory in something liek duckdb performance becomes dog so 64 would be better than 32 but youll run out of memory at some point anyway and what then


maybe ill wait until delivery dates are 1 day so i can impulse either and stick with it

#1132391

>>1130579

>

maybe ill wait until delivery dates are 1 day so i can impulse either and stick with it


nice tactic

#1132608

so many new tech videos on youtube, don't have enough time to watch everything

not even talking about kpop, didn't listen most comebacks for past few months

#1133797
1720×388159.83Kb

>>1133561

there's no free lunch, 2x machines is extra abstraction which will never work as good as one with 2x specs


not even talking about the fact that 1x ram <- network bridge -> 1x ram would never ever be as good as 2x ram in any kind of multicore task, it will be much worse. same for 2x cores

#1133928
>

Top Justice Department antitrust officials have decided to ask a judge to force Alphabet Inc.’s Google to sell off its Chrome browser

imagine verizon chrome or whatever

i almost went all in on chromeos a few months ago, after ordering a chromebox i read the news google had removed multiple browser profiles on the very same day and cancelled my order

#1135324

>>1118647

went to one more concert, this time I was better prepared: recorded all clips in 4k30p (but lightning was way way worse tho ) and shook hand


but still there were lots of people much closer to the stage than I am

I can't ever get the perfect experience, it's always ruined by the people who get the better spot


theoretically it should be possible to eventually achieve perfect score in one small local niche, but I'm not enough insane for that...

there're so many things in the world, why limit yourself to tuning one small irrelevant thing

isn't that cope tho


anyway, I think people who get the best result (e.g. best places at kpop concerts and visiting fanmeets with shaking hands) must be "cheating" in some way: e.g. have local friends or lots of money from parents

I'm not sure what is the percent of the people who perfect some niche from the low start, but it must be small

#1135329

>>1135324

>

there're so many things in the world, why limit yourself to tuning one small irrelevant thing

on the other hand, if you think about speerunner community, they do exactly that and quite fine with it...


probably can think about getting best experience in real-life scenario in the same way


speedrunners seem much more motivated tho, if I put just 1% of their effort into the real-life isolated case, it seems I will perfect it with ease

#1138365

>>1135324

>

shook hand

hopefully you passed her some bills to make her happy


best place in kpop concerts probably requires fanclub membership for early access tickets

#1138698

>>1138365

>

hopefully you passed her some bills to make her happy

yeah

but on the other hand pun not intended, lots of people give her bills and most of the time she will hold hand much longer than in my case


first you want something, then you achieve it, then you want something else

how to escape that stupid circle

#1140326

>>1140325

so if you have N number of friends, you can say M number and the one who happened to put hand mth time will lose


so if e.g. 4 friends and you say 5, the one who puts first will lose

it's basically to calculate M mod N, and never put your hand for that case, isn't that too easy


unless the order of people always changes, but still it's just mod operation, quite fast to do every round

unless you're too drunk and stupid at that time

#1140679

>>1140326

you are severely underestimating korean math drinking geims. they are quick af

#1141490
600×60055.63Kb

best consoomable of 2023 was pic related

this year it was open back headphones. so nice no squeeze head or get hot

#1142176

that gemini 2.0 release makes me scary

things are moving way too fast

#1143592

so many people asking/telling what are they gonna do on New Year holidays

even if I do something too, now I have to make sure my way of celebrating isn't worse than 50% of the people, otherwise I will have massive FOMO


why can't people just be boring and stop competing over every little thing in life

or maybe it's just me...

#1145947

>>1143592

it's just you, you could instead be a chill guy and do whatever you were gonna do whether that was something special or nothing special at all

#1147300
1206×1948521.62Kb

why doesnt applel add webm and av1 to quicktime for free thumbnails hardware decoders are in

#1150339

trying cursor now

also tried cline + deepseek, but that shit just isn't working

#1152455

>>1150339

>

but that shit just isn't working

Roo Cline is actually working with DeepSeek

but too lazy to test it much, Cursor's Composer is easier to use...

on the other hand, the actual task/bug solving works very badly, I would rather do all the changes by myself

except the quick fuzzy refactoring, simple functions/code generation and Copilot-like autocompletion on steroids (Cursor Tab) are very nice


but I don't even use chat much, because they often hallucinate and give incorrect answers about simple topics google + stackoverflow still feels more solid in terms of knowledge I get

#1152457

>>1152428

>

for another year

more like for another month


I've stopped hoping, it seems inevitable now

should have invested in Nvidia stocks/real estate so I wouldn't have needed to work hard to earn money every month

but it seems impossible to earn much money now, because you would be replaced by AI soon and be in slavery for people who earned lots of money before, who would become your lords chae_dead:


and what did I do, knowing that might happen? nothing RIP gg

#1153458
1206×2144359.42Kb

>>1020245 ➡

>

question: blah blah c++ how to learn

>

i want to keep my distance

it's over for dumb low level languages, python won

i switched to circuitpython myself

#1153459
640×36016.83Kb

>>1152457

>

more like for another month

i don't work in a tech firm so maybe noone will notice my kind is obsolete