technology

https://jalopnik.com/ai-models-are-eager-to-launch-nukes-in-war-simulations-1851235624

https://archive.is/XvWBK

Hi c/technology, I'm agitated about a specific topic and the 'normal' queers on discord are not really tolerating my rambles so I hope it's ok if I post em here. Prepare yourself for one embarrassing boomer-ass take

I have two sets of bluetooth headphones. One is the Sennheiser BT4.50 which I got on a firesale($40) when they were clearing stock, and one is the Anker Q35. They're both terrible!

There are the practicality concerns of course, like that it's yet another battery to charge, which also means most bluetooth headphones have limited lifespans and are guaranteed e-waste. It's yet another 2.4GHz signal on top of the pile of them coming from [old 2.4G wifi devices/2.4G wireless peripherals/etc], and yet another wireless thing that refuses to sync randomly. The Android apps that these things try to use are teeeeerrible shovelware that barely works (the Soundcore app took like three re-connects to pick up my Q35). It's a lot of trouble that could be fixed by just running a cable down my shirt.

I get that there's a niche for this stuff, though. If I'm watching Youtube or a movie or whatever on a computer box hooked up to a TV, nobody wants to run a 3.5mm cable ten feet or more from that box to the couch for headphones. I use little Bluetooth 3.5mm receiver/transciever hooked up to headphones (Fostex T50RP mk3) for this purpose. And the sound quality isn't even that bad; the difference between a flac over a cable and a flac through SBC codec is pretty small. AptX is hard to tell apart at all.

That's on a $250 (CAD!) set of planar magnetic headphones though. The BT4.50s and the Q35s? They are dogshit. I thought that the ability to run a cable to these things might remove the problem of their limited lifespan, but in the case of the Q35s they sound atrocious wired without the EQ built into their goofy app. The Sennheiser just sounds like trash period. Curiously both of them have pretty much no high end, and probably worse sound stage than fully closed-back headphones like the Audio Technica M40x I have.

It feels weird that the bottleneck in sound quality for wireless headphones is not the BT codec, but the actual drivers. That's the short and long of it though; I feed them stuff like the Killing Joke selftitled or Big Generator by Yes, basically any of my really high dynamic range CDs as well as some 24/48 vinyl rips (Face Dances by The Who, Dreamboat Annie by Heart) and then some more modern crushed stuff for fun, bits of Black Dresses or some math rock and whatnot. None of it sounds any good on the BT4.50 or the Q35 without EQ, and even with the Q35's EQ it can't really hold a candle to even a middling set of cans like the M40x. I guess this makes sense, because you're paying for the battery and the bluetooth stuff and the amplification and the mics/noise cancelling? But damn, your wireless options are the same price as wired for way worse sound, or maybe theoretically more costly for the same quality? Pls hmu if you know of any bluetooth headphones that would measure up to classic cans like the Sennheiser HD600, Fostex T50 etc. tyvm

For what it's worth, that noise cancelling is probably worth the money on its own - the ability to just block out the sound of the neighbour's kids shrieking or the annoying bass-pulse of an air conditioner is really rad. Both of these bluetooth headsets do a pretty good job of blocking out background sounds, and that's rad for autism gang & anyone else who has sensory processing troubles. But that is pretty much their only niche imo! If I wanna watch junk on Youtube and there's a jackhammer going across the hall, I grab a wireless set and hope the bluetooth will pair automatically.

But if I actually wanna listen to anything in non-vomit audio quality wirelessly, I'm strapping one of those little bluetooth re/trans-ceiver things to my real headphones and pairing that(which has a better hit rate for pairing than the Q35, lol lmao) to my PC or an android box.

To be real though, in 99% of the cases I use headphones, a cable is easier & more convenient & results in the best audio quality. Using headphones on a PC at a desk? Wired. Hooking headphones up to the phone for a Youtube clip? Wired. Fiio M7 for music? Wired. If distance isn't a factor, bluetooth is a huge hassle for a worse experience overall.

I just hate with a passion that this at-best niche product category that performs worse than the established counterpart (remember how badly stuff like the ~~Bluebuds X~~ Braggi Dash were for disconnects?) seems to be replacing wired options in the popular conscious?? Phones and laptops without 3.5mm jacks should be considered war crimes, trust me. No I do not care if it's about IP67 water protection, do better pls. No excuse. I despise this trend.

Granted, I'm a weird freak and willing to lay down a lot of convenience in pursuit of "shit sounds good". More than a lot of people, like I'm pretty sure the average person would look at you funny for suggesting that they carry a second device around for music. I didn't like MP3(or lossy codecs lol) as a premise, because don't decide what I can and can't hear(even if I can't hear a lot of difference lol), but I at least understand why MP3 was cool and rad. For one thing it predates flac, but also an album in MP3 is like less than a third the size of the flac equivalent, and big files would fill up your shitty 80GB IDE drive(or worse yet, your ipod) real fast. The tradeoff in sound quality which most people probably can't even hear for very small file sizes makes a ton of sense, even now with codecs like opus. I only fill big SD cards with flacs 'cause I'm a stubborn fucker.

But bluetooth audio in general seems like a niche technology with one or two cool uses(TV with headphones, and those embarrassing bluetooth earpieces executives used to have in the 2000s) that mainstream users are adopting with all its numerous faults, exclusively for the privilege of not having to take off the headphones before you get up from a desk. For the privilege of not having a cable running through your sweater? All this garbage just for that?

It's possible my myopic ass is missing some critical use case that normals have, and I know most nornals do not care at all for sound quality. I also know that people will have had both better and worse experiences as far as bluetooth goes, but maaan this shit sucks and it makes me buttmad that they wanna get rid of cables for pretty much no reason, grrrrr gonna go stare at the bits of vintage audio gear I have, rip

Most techies now dabbling in the media are arrogant amateurs who think that because they excel in one area, they are masters of all domains. What they really are is incompetent at giving any insight or illumination beyond their own narrow self-interests while decidedly cheapening discourse. Elon Musk is the patron saint of this practice, holding forth on everything from COVID to what Russia is doing to a recent series of disturbing declarations about immigrants, which are beginning to eerily echo the rants made by his grandfather in South Africa. Even mulling the implications of the head size of newborns delivered via cesarean has not escaped his twitchy fingers. Having had one of those for my eldest and pretty sure it had no impact on any of my four kids’ intelligence, my advice to Elon and his nearly all-male cronies: Take a seat, boys.

Unfortunately, rather than ceasing, they are now poised to take it all with an assist from the newest game in tech: artificial general intelligence, or AGI. In the short term, it’s already clear it could be devastating for media companies — Google will not just be providing algorithmic search results; it and others like OpenAI have been using the new tools to scrape content largely made by others and reformulate it for the masses. That’s a simple way to describe it, but what could happen is what happened before: a complete hijacking of the content universe. What do you need New York Magazine for if they can swallow it, digest it, and regurgitate it back up in ways both anodyne and dangerous like the careless Information Age turkey vultures they have always been?

cross-posted from: https://hexbear.net/post/1769620

1. I am directing most employees to work from home tomorrow, Wednesday, February 7, so everyone can be in a safe, comfortable environment on a stressful day. Most individuals will not be able to enter the Lab during this mandatory remote work day. A Lab access list has been created and those who will have access will be notified by email shortly. If you do not receive an email instructing you to be on Lab, please plan to work remotely, regardless of your telework agreement status. In addition, and to ensure we have everyone’s accurate contact information, I am also asking everyone to please review and update your personal email and phone number in Workday today.

I don't think I've ever seen a company or organization that had mandatory remote work day outside of really crazy weather during the peak of Covid. Perhaps it's to protect the equipment from distraught or disgruntled employees?

Once again techbros find a new way to make the world worse. Now your toothbrush can be part of a bot net. Nobody asked for this, and the suggestion that it should be possible is utterly deranged.

Banging my head against the wall anymore trying to search for simple shit and the results being unrelated crap from big name tubers who've SEO'd the system to get their shit up top first.

I warned you

need a new vpn. i don't care about functionality too much i just want something where my ISP won't chat shit at me for pirating movies.

Fun, exciting, riveting, when's Wayland ready, year of the linux dekstop

CPU-posting on main

MTI = MIPS Technologies (company that made MIPS (Microprocessor without Interlocked Pipeline Stages) processors, they make RISC-V processors now lmao)

At the time when the MIPS R10000, known as the "T5" while in development, was being designed, MTI had made a name for themselves as designers of high-performance computer microprocessors along the lines of the then-new philosophy of reduced instruction set computing (RISC). Actually, their R2000 design was the first commercially-available RISC microprocessor. By the time the T5 was being designed, they were no longer alone in the RISC microprocessor market. Several companies, including IBM and Motorola (joined together in the AIM alliance which produced PowerPC), DEC (who designed the Alpha line of RISC microprocessors after MTI owned them in the 80s when their radically simpler chips were performing better than VAXen), and Sun Microsystems (who were making the SPARC line of microprocessors) were now marketing RISC microprocessors. Not just even marketing but beating MTI in the market they had created. After trying and failing to develop their own complete computer systems alongside their chips, they were having financial difficulties until Silicon Graphics acquired MTI to secure availability of MIPS microprocessors for their famous ("it's a Unix system, I know this!") MIPS-based workstations and servers. Although their new (in 1993) R4000 and R4400 designs performed well compared to their contemporaries, they were quickly being made obsolete by MTI's competitor's new offerings and they were left with a problem:

The MIPS R4000 and the R4400, which is essentially an R4000 with bigger on-die caches, were more or less just an architectural evolution from the R2000. The R4000 made its performance in much the same way as the R2000 did, the classic RISC design process mantra: "let's make it simpler" and thus be able to run it faster. In particular, what this means for the R4000, and what is a key difference from its predecessors and its contemporaries, is a technique called superpipelining. In an instruction pipeline, the maximum speed at which your processor can issue instructions is set by the pipeline stage which takes the longest to complete. Superpipelining is one way of addressing this problem: you can subdivide each pipeline stage into 2 simpler pipeline stages that individually complete faster and thus be able to clock your chip faster without problems. However, this has its limits. Eventually, it becomes impossible to further "deepen" the pipeline like this or clock the processor faster in general without other problems. This is why MTI's competitors opted for the analogous superscalar approach: you can duplicate functional units of your processor and have multiple instructions "in flight" at the same time and usually this also involves multiple pipelines. At the time MTI thought this approach would result in more consistently higher performance (not to mention save die space) but were quickly proven wrong when their competitor's superscalar (and often with other architectural tricks) chips were outperforming the R4000 in spite of MTI's fabrication partners constantly improving their process and releasing chips that ran at higher and higher speeds.

Enter the MIPS R8000 (die not pictured here) in 1994, a weird and expensive 6-chip 4-way superscalar design meant for the high-end microprocessor market while the next-generation T5 (which would become the MIPS R10000, as mentioned earlier) was under development. It didn't sell well because of its high price and the fact that its integer performance, important for general-purpose computing applications, was lacking compared to the 200-MHz R4400 that was being sold by then. It did, however, have impressive floating-point performance, which landed many R8000-based systems in the TOP500 supercomputer list for a time. But this design could never be the high-performance and general-purpose processor MTI needed to compete with their competitor's offerings...

Introduced in 1996, the MIPS R10000 (die IS pictured here) was a significant departure from the architecture of the R4000 (which more or less was directly derived from the first research done at Stanford University where MIPS was initially created over a decade earlier). Dropping the superpipeline approach, the R10000 is a 4-way superscalar processor even capable of executing instructions out of order! Another big change is that it has a branch predictor and speculatively executes instructions after a branch as opposed to the R4000, which used the classic MIPS "branch delay slot" technique to schedule one more instruction in the pipeline after a branch and then stall lol (they should have added even more delay slots, caring about binary compatibility is liberalism). It's hard to find benchmarks for something this old but this design performed at least several times faster than an R4400 at about the same clock speed!

If you like my CPU posting and want me to post more in the future let me know

Also ask me any questions if you want too and I'll try to answer

Just straight refuses to load. Works fine when I turn uBlock off.

"If people think that an IPO means we're going to … push prices up, push the margins up, push down the feature sets, the only answer we can give is, watch us. Keep watching," he said. "Let's look at it in 15, 20 years' time."

I'm pretty sure those are the first things your new shareholders will be demanding but okay

Supposedly it's an error:

A person familiar with the kerfuffle who has visibility into the Windows giant, though who did not want to be identified, told us it appears that "if a user chose continuous import in the Edge first run experience on some other device, this state may be syncing incorrectly across their devices. This is not the intended feature experience." We're assured that Microsoft is addressing it for the next Edge Stable release.

https://www.businessinsider.com/instacart-deletes-ai-generated-food-images-2024-1

cross-posted from: https://hexbear.net/post/1729041

And with that, some questions.
Should I sign up for Twitter for news? Or should I say 'Fuck Elon' and not get a Twitter?
I get most of my news from Hexbear anyway.
Also, I have a Telegram. What news sources does Hexbear recommend on Telegram?

I found this script but I don't know if I should trust it, what is the best tool to use?

From the 'about us' section in the app:

I'm Ahmed Bashbash, a Palestinian developer, I got the idea to help the boycott movement by making an app that help people to know which products is in the boycott list, the app idea is so simple, just to scan the product you wanna buy and the app would tell you if that product is in the list ot not, I got this idea and I started to make it because I suffered myself from that brutal occupation, I lost my sister in 2020 after a month of suffering of medical problem and she couldn't get an authorization to go get the medical attentions she needed in Jerusalem at the right time and when she got it after a month from trying, she died in the ambulance that was going to the hospital in Jerusalem, and on October 31, I lost my brother from a random air strike to a 4 civilians houses which these random strikes were happening a lot, so l decided to finish this app and make it entirely free and unprofitably in behalf of my brother and sister, and any money I will get from the app I will send it to a Palestinians organizations that help the people in Gaza.

Place your bets on how many days they last until they literally jerk it till they rip there own dick off (you know it's a dude that agreed to get this).

More likely, he's probably lying.

Good video on the technical and technological challenges.