Linus Gabriel Sebastian & Linus Media Group / Linus Tech Tips - Narcissistic corporate shill YouTuber driving his media empire into the ground. KILL COUNT: 2

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
Also they didn't even mention that you can use an ide2sd adapter in older pc builds and its way easier to transfer files between retro rigs and modern rigs. You don't have to burn CD-Rs.
Or just have a... wired network. It's not like it's impossible to find an FTP server for modern OS's.
 
The kind of person to do that in the early days of a channel is also the kind of person with morals
I really don't think its that. Being the camera man for a youtube channel which is just a lot of fucking around and being a camera man for a youtube channel that are 80+ people deep are 2 very different things. And any reasonable person knows that you can only keep your fun fuck around job for so long.

The Bullshit to Money ratio got out of whack. It used to be fun and when you're young you tend to not care about the money so much. The money prob got better but not to where it should be and definitely not enough to make up for tard wrangling zoomers all day long. And if your job isn't fun, and doesn't pay, there is nothing stopping you from chasing the money.
 
  • Thunk-Provoking
Reactions: awoo
Or just have a... wired network. It's not like it's impossible to find an FTP server for modern OS's.
I'm pretty sure you don't even need FTP, I've seen videos of people using NEXTstep with perfectly standard NASes, SMB/CIFS is an 80s protocol after all. You may need to explicitly enable support for (unsafe) SMB versions less than 2, but you can easily whip up a share an old computer can access from any modern Linux distro, or presumably a Pro or better version of modern Windows.
I'm not particularly knowledgeable about Windows, so I looked it up, and it seems Microsoft added support for SMB in NT3.1 (1993) and Windows 3.1 Workgroup Edition (also 1993). Presumably 95 and up would also have this feature, but you might need some special Workgroup Edition there too. I do know it's included in Windows 2000 and works out of the box, I used my NAS to install pirated SMAC on a virtual machine.
 
AMD time to shine, I think, is in the next 4 to 6 years when they achieve 50% market share of the entire server market. They would have a lot more capital and money to burn to bolster their position in the GPU market. Intel and Nvidia seems to be plagued with useless middle managers that weren't smart enough to figure things out or respectable enough to be loved by the actual engineers, they probably could use some more job cuts.

Thread tax: while I dislike Linus, I don't hate him enough that I wish bad things upon his family, fuck his employee though, they're disgusting. I think he should learn a little more from Gamers Nexus or Hardware Unboxed; they don't seems to try to market Intel, AMD or Nvidia and understands that these companies are here for your money. Also, they should stop trying to make videos specifically made for kids and upload less. That's the least they could do to bring in young adults, which is the focus of every technology company, to spend more on their shit.
I've always used AMD, their processors kick ass, and Intel has always struggled to keep up. Nvidia has stayed top dog, but the question is for how long? You don't get that big without incuring bloat. Look at Linus. Making these 5000$ rigs while his employees slave away, all in the most ineffective way possible. That's what happens when you get big. On the youtube side, it's nice when I find a smaller channel that puts in a shit ton of effort, that only posts when there ready. That's art right there.
 
Nvidia has stayed top dog, but the question is for how long?
The biggest risk point is probably ten or so years after the current generation of senior boomer engineers retire. I was going to say "or until the RTX relevant patents expire" but with the rate graphics technology moves, by the time its just accessible, it'll also be irrelevant. The fact is that regardless of their potential bloat, Nvidias senior designers, RnD and such are still proving highly effective and leading a lot of major industry initiatives with raytracing, AI acceleration, AI upscaling, etc. The most likely point for them to stumble on this will be once a lot of that senior talent steps out, and they reach the point in the RnD lifecycle where the projects from their successors are coming to market. The wider boomer retirement is almost certain to leave novel skillset or talent gaps in any organization, so if they're going to stumble, it'll be from that.

It is also entirely possible that they effectively fill the gaps, or that the retirement coincides with such a significant industry shift in some aspect design that they inadvertently end up with the leading experts in the new thing just by virtue of being one of the few orgs big enough to pursue it with new talent. A lot of its just too far out to really predict with much accuracy.
 
-Brandon left this week, one of the original employees and longtime cameraman. He had an intel upgrade if you don't remember who he was. He was competent in what he did but it seems odd he couldn't just offer more projects or better pay for a guy who was so loyal he was using his own camera equipment when they started.
I liked Brandon. He always seemed like he had a stick up his ass but he also seemed very dedicated to his craft. With him gone who is left from the old house?
 
  • Thunk-Provoking
Reactions: WelperHelper99
The biggest risk point is probably ten or so years after the current generation of senior boomer engineers retire. I was going to say "or until the RTX relevant patents expire" but with the rate graphics technology moves, by the time its just accessible, it'll also be irrelevant. The fact is that regardless of their potential bloat, Nvidias senior designers, RnD and such are still proving highly effective and leading a lot of major industry initiatives with raytracing, AI acceleration, AI upscaling, etc. The most likely point for them to stumble on this will be once a lot of that senior talent steps out, and they reach the point in the RnD lifecycle where the projects from their successors are coming to market. The wider boomer retirement is almost certain to leave novel skillset or talent gaps in any organization, so if they're going to stumble, it'll be from that.

It is also entirely possible that they effectively fill the gaps, or that the retirement coincides with such a significant industry shift in some aspect design that they inadvertently end up with the leading experts in the new thing just by virtue of being one of the few orgs big enough to pursue it with new talent. A lot of its just too far out to really predict with much accuracy.
Well until then I'll keep buying their cards; AMD makes great cores, but only passable cards. I can only hope that the bloat doesn't get to the engineering teams once the Boomers/ Gen Xer's leave.
 
It is also entirely possible that they effectively fill the gaps, or that the retirement coincides with such a significant industry shift in some aspect design that they inadvertently end up with the leading experts in the new thing just by virtue of being one of the few orgs big enough to pursue it with new talent. A lot of its just too far out to really predict with much accuracy.
We are also in a period of rapid progress in machine learning, and Nvidia are at the forefront of that.
It's entirely possible that the last thing Nvidia has those engineers do is train a machine to be their replacement. Thinking machines designing the hardware running other thinking machines. At that point we'll be looking at some fairly rapid progress that conceivably could even be called evolution.
 
I liked Brandon. He always seemed like he had a stick up his ass but he also seemed very dedicated to his craft. With him gone who is left from the old house?
I think it's Nick (the business guy), Edzel the main editor turned producer, and Luke who technically isn't part of LTT. Dennis arrived midway IIRC so he could count. That's about it.

Just skimming through the old videos for verification and the old episodes are so much better. Quieter, no annoying way too loud intro, less ADD editing, and I think Linus looked better without the cuck beard.
 
We are also in a period of rapid progress in machine learning, and Nvidia are at the forefront of that.
It's entirely possible that the last thing Nvidia has those engineers do is train a machine to be their replacement. Thinking machines designing the hardware running other thinking machines. At that point we'll be looking at some fairly rapid progress that conceivably could even be called evolution.
Without another breakthrough, unlikely. Current Machine learning is almost all focused on understanding, not inventiveness - Machine learning is already used to extreme degrees with specialized software to help iterate and optimize board and chip designs, but machine learning is nowhere close to being able to design a new paradigm of render chip and associated machine code. There's been extremely little success on AI that create something concretely new, from a functional sense.

If there is a breakthrough and AI manages to start handling multi-level goals to create something completely new, then maybe. But as it stands, AI will help optimize CUDA and RTX hardware, not create new spirit-imbued AI chips or whatever. But that AI breakthrough has been the same story as Fusion, totally ten years away bros, totally.
 
Oh, I'm not saying the machine would do the whole thing on its own. They'll still need engineers. But what machine learning could do, and has proven to be very adept at, is "fill in gaps". If one team trains it to understand a new process node, and another team does part of the work translating an architecture to that node, the machine could assist in that translation process, it could look for flaws the engineers may not have noticed, and with sufficient compute power it could eventually extrapolate on work the engineers have already done and offer suggestions for improvements.

A few years ago we would have thought it impossible that machines could ever program, but before chatGPT was lobotomised I used it to speed up writing some python scripts for work. It wasn't perfect, but what it managed to do with function headers and some basic descriptions was astonishing to me, and getting custom code I could basically copy paste into my file was much quicker and helped me find some errors I hadn't been aware I was making. I'm not a skilled programmer, so I'm not saying it could replace anyone less incompetent than me, but that it was able to do this at all surprised me. I might be stupidly optimistic about a field I don't fully understand, but I honestly don't think we're that far off from machine learning speeding up engineering thousandfold the way it already has obsoleted artists and "freelance journalists" or whatever you'd call those people..
 
Oh, I'm not saying the machine would do the whole thing on its own. They'll still need engineers. But what machine learning could do, and has proven to be very adept at, is "fill in gaps". If one team trains it to understand a new process node, and another team does part of the work translating an architecture to that node, the machine could assist in that translation process, it could look for flaws the engineers may not have noticed, and with sufficient compute power it could eventually extrapolate on work the engineers have already done and offer suggestions for improvements.

A few years ago we would have thought it impossible that machines could ever program, but before chatGPT was lobotomised I used it to speed up writing some python scripts for work. It wasn't perfect, but what it managed to do with function headers and some basic descriptions was astonishing to me, and getting custom code I could basically copy paste into my file was much quicker and helped me find some errors I hadn't been aware I was making. I'm not a skilled programmer, so I'm not saying it could replace anyone less incompetent than me, but that it was able to do this at all surprised me. I might be stupidly optimistic about a field I don't fully understand, but I honestly don't think we're that far off from machine learning speeding up engineering thousandfold the way it already has obsoleted artists and "freelance journalists" or whatever you'd call those people..
I may be wrong, but from what i have seen of AI training, it is on the level of monkey see monkey do. The fact that they can't get them to automatically stop saying Hitler quotes after being let loose on the net for a few hours is telling. It is good tech that has its uses. But from the example you gave, i could compare it to how the pocket calculator is used in the math's world. It did not replace math's teachers or scientists. The artist AIs would never really be used for personal artwork. People would still pay millions for the content of a trash bag thrown around a white room with a shit name and appeals to how "deep" it is. Same with personalized pictures of Fursona's. Its main use would be corpo logos and flash art.
 
The artist AIs would never really be used for personal artwork. People would still pay millions for the content of a trash bag thrown around a white room with a shit name and appeals to how "deep" it is. Same with personalized pictures of Fursona's. Its main use would be corpo logos and flash art.
The people paying millions for literally garbage aren’t doing it because art, but because it’s a convenient way to launder money or pay for services without being taxed for the transaction.

Furfaggotry and other commission artists absolutely are being replaced, Stable Diffusion has tonnes of models that can do just about any kind of art, including weeb and furry stuff. Why would you pay someone on deviantart 60$ to get one piece of art next month, when you could spend an hour spamming prompts at Stable Diffusion until it gives you what you want for free? You’d spend more time than that just explaining to a real artist what it is you want. Especially if you know how to sketch yourself, since img2img can really speed up getting a good result.

You just need to look at reddit to see what a panic these people are in. Actual human drawn art is summarily banned because it looks like it might be AI-drawn.
Activision-Blizzard was hiring concept artists just a few months ago, specifically asking after experience with AI art, and having used Stable Diffusion to create illustrations of characters for D&D, I can see why. It comes up with some truly amazing outfits sometimes. Instead of a room full of concept artists sketching all day, you could have one concept artist sketch up a figure, run Stable Diffusion on it to generate a hundred pictures, and then just filter out the bad ones. Train an adversarial model or a hypernetwork and it will filter out the bad stuff on its own.
 
The people paying millions for literally garbage aren’t doing it because art, but because it’s a convenient way to launder money or pay for services without being taxed for the transaction.

Furfaggotry and other commission artists absolutely are being replaced, Stable Diffusion has tonnes of models that can do just about any kind of art, including weeb and furry stuff. Why would you pay someone on deviantart 60$ to get one piece of art next month, when you could spend an hour spamming prompts at Stable Diffusion until it gives you what you want for free? You’d spend more time than that just explaining to a real artist what it is you want. Especially if you know how to sketch yourself, since img2img can really speed up getting a good result.

You just need to look at reddit to see what a panic these people are in. Actual human drawn art is summarily banned because it looks like it might be AI-drawn.
Activision-Blizzard was hiring concept artists just a few months ago, specifically asking after experience with AI art, and having used Stable Diffusion to create illustrations of characters for D&D, I can see why. It comes up with some truly amazing outfits sometimes. Instead of a room full of concept artists sketching all day, you could have one concept artist sketch up a figure, run Stable Diffusion on it to generate a hundred pictures, and then just filter out the bad ones. Train an adversarial model or a hypernetwork and it will filter out the bad stuff on its own.
Going by the hysterics of reddit is not a measure of real world implications of AI. Do you have any hard data that says artists have suddenly been thrown in to the trash bin of history?
You claim stable diffusion is significant. Yet since it leaked i have not seen any evidence of this supposed artist genocide. Just the spastic sperg outs from people with unwarranted self importance. Normal people will still commission art done by human hands. The point i was making is that the up their own ass art fags, will still continue to be art fags. Perhaps invent some soul/souless meme based on "you can never replicate the emotion of human hands with a brush" and use that line to continue to be pretentious fags. Just setting up Stable Diffusion and learning how to use the software is a barrier of entry.
I just feel the whole thing is a storm in a teacup. I have SD set up on my machine. It does produce some good looking stuff. Yet a dumb shit ass furry wont get exactly that he wants without the human touch.
 
FR FR, no cap AMD was SHIT in the early 2010s until Ryzen. No excuse to go there except retardation, and maybe cost, I don't remember if they were actually a better value or not? I don't think they were, meaningfully. Maybe around 2014-15 when they were selling old stock. Herp a durr I have eight "real" cores!!! Why my multithread no scale??? What is a shared pipeline bros? I have more GHZ why am I still slower than shilltel bros please help??? Bulldozer, what a shit release. Those were real bad times.
 
Furfaggotry and other commission artists absolutely are being replaced, Stable Diffusion has tonnes of models that can do just about any kind of art, including weeb and furry stuff. Why would you pay someone on deviantart 60$ to get one piece of art next month, when you could spend an hour spamming prompts at Stable Diffusion until it gives you what you want for free? You’d spend more time than that just explaining to a real artist what it is you want. Especially if you know how to sketch yourself, since img2img can really speed up getting a good result.
I can actually see this happen.
 
Back