>3 years old>Still a beast that runs everything at 4k120fps+Will the 7900XTX be considered a legendary GPU?
>>107741545>its legendary like the 3090What's so legendary about a 3090? Even the 4080 beat.
>>107741964Lying faggot. I run my 7900 at 8k with 12,000 FPS
>>107741903>my rtx 5080hahahahahaha you fucking faggot retard
>>107740063I think it pretty legendary, considering I bought my 7900xt for 530 dollars and this gay chart posting faggot forgets that 530 dollars<1000+ dollars for SLIGHTLY less performance.
>>107737417what's the current price for one? $3K
Purely from a business perspective, the way AI companies are moving to buy as much hardware as possible recently is really making me raise an eyebrow.We know that venture capital funding for these companies is basically gone, AI stocks flatlined over the past few months or, in the case of some like Meta and Oracle, went down quite a bit.All they can rely on right now is their circular investments and I think what they're trying to do is just turn that cash into real assets with the expectation that the models will just scale themselves, and investors will again be convinced to put money into the system again. This is of course augmented by the fact that companies like Nvidia, AMD, and Micron are the ones giving Open AI and Microsoft hundreds of billions to turn right back around and buy products from Nvidia, AMD, and Open AI.Throughout the cycle we just haven't been seeing the uncapped colossal buy orders we saw at the end of 2025. So what's the endpoint here? When the manufacturers are paying companies to buy their own products and investors aren't seeing the business turn profitable, what happens next? Where does the fire's fuel come from?
>>107742250Are you sure that's still true? Because YouTube has just as much licensed music and charges the same amount of money for ad-free at least in the USA. And YouTube is profitable despite having more staff, way more revenueless slop, and presumably higher adblock usage.
>>107742229No. What is much more likely to happen is that the resource bottleneck is what will crash the current iteration of AI from most frontier companies. Billions will be lost overnight, headlines will smugly rejoice, and the public will think that's the end of it. One or two companies might survive, whatever. But the employees will still have expertise, and they will be poached. They will especially be poached by people who want to replace graphics cards with a different resource. Right now, AI uses insanely inefficient matrix multiplications of floats on GPUs, because GPUs are what's most available due to existing games/rendering/bitcoin mining demand. But if we are instead building a processor for AI from the ground up, you wouldn't want to use matrix multiplication of floats, you would want to use ternary. The integration of the processors and AI training in one development pipeline will be what arises from the ashes of the crash. The NES replaces the Atari, and life goes on.
>>107742296I looked it up and they become profitable in 2024, good for them. If Open AI keeps it up they might someday be able to make one BILLION dollars every year
EMPANADA
>>107742310ThisOpenAI categorically can’t exist unless we just spontaneously start forgiving debtBut they have established the infrastructure for a national surveillance gridLLMs are retarded, you’re seeing about the apex of what they’re capable of, but ML algos have a lot of headroom to grow. They will never be available to the general public, though, because there must either be an agentic layer above them or they have to be baked into another tool. You MAY see 3rd gen ML in certain applications, or baked into your OS, but you will never be allowed to drive one off the lot.
Wtf happened to their reverse image search? It's even worse than Google's now.
why is yandex so janky nowadays?it doesn’t load anything if i change search settings
>>107739397When that happens you can just swap .com to .ru.
I imagine it has something to do with internet being flooded with ai slopWhy would you index images when 99% of it is garbage after all
>>107736822>Wtf happened to their reverse image search?Ukraine...
>>107736822they moved moved into the EU where tech goes to die
>120 page composition books are $6 each>a ball point pen is $5I used to be in the write your notes on paper camp but did Bidenflation finally make e-ink tablets worth the money?
>>107742278>>120 page composition books are $6 eachno they're not>>a ball point pen is $5no they're not
>>107742348Fuck you I just came back from Rite Aid, your lies won't work on me.
>>107742369If you pay more than 1$ for a pen you're being scammed
>puts bread on the table>makes inexperienced /g/ trannies seethe>makes C shart shills seetheYeah I'm thinking it's the best language
>>107739600good thing that he did not use SFINAE
>>107739605>(shittily) implements dispatch over a FIXED set of variant types>whole point of interfaces and generics is that it's extensible and can support new types without having to modify the original definition + a big faggy switch caseLOL. Sum types != generics. Thanks for outing yourself, retard.
>>107739395>i got filtered by OOP>that means OOP is bad
I actually like Java now. I'd still use C# before I use it for anything but it's not as bad as people rag on it for. My company recently upgraded from 8 to 21 though which made it a lot more tolerable.Now, picrel on the other hand, my god what an embarrassment.
>>107742208>8 to 21Same story. So much nicer with records, pattern matching, jshell, etc.>MavenJust curious- what are your issues with it? I've just started using Java more because of work and haven't yet had many issues with Maven (perhaps I haven't hit the major snags yet). I think I actually had more issues with Gradle.
>What phone has X and Y feature?Don't ask, use these!https://www.gsmarena.com/search.php3https://www.kimovil.com/en/compare-smartphoneshttps://phonedb.net/index.php?m=device&s=queryGood Resources:>Reviewshttps://www.gsmarena.comhttps://www.phonearena.comhttps://www.notebookcheck.net>Frequency Checkershttps://www.frequencycheck.comhttps://kimovil.com/en/frequency-checkerComment too long. Click here to view the full text.
YouTube went full retard so now scrolling speed is limited. What app can I replace it with? I'm not gonna pay for premium anymore
>>107742276newpipe / pipepipe
>>107742305Imma give you my pipe later and you'll like it. :DDD
>>107742305Thank you! I used to use newpipe a few years ago but pipepipe being a new fork I didn't know about. Retarded name but will work.
>>107739157Massive cope. I get it. You are invested in apple and have to live in fear. It's the worst putting everything in one basket. Just waiting for the basket to fall over. Been there. Never again!
Welcome to the Daily Programming Thread. What are you working on, /g/?Previous: >>107711909
>>107742225I'd like to see you do better, faggot.
>>107742198What I've wrote only applies when you're using one arena/stack allocator for everything. Python uses reference counting (or GC?) so it doesn't have this problem, which is the whole point. Arena/stack allocation is not enough unless you use several of them and this is going to be as painful as manual memory management. For scripting purposes, the kind of programs you'd write in a scripting language and the allocation patterns you would have, a GC is what you need.
>>107742309>What I've wrote only applies when you're using one arena/stack allocator for everything.and no moving of objects (allocate temporary objects, use them to construct the object you're going to return and before returning, copy the object you're returning on top of the temporary objects that are now dead)
>>107730163normalize turning the brightness all the way down on all your devices at night
>>107742360i want whoever touches my pc to be bathed in light
why can't we have nice things? are we supposed to just abandon the best app made for this site? how do you make the captcha appear?
>>107741299this is a thread about kuroba dev. kuroba ex is something else.
>>107739696No. Spambots using solvers is why the captcha changed. The new one is faster and easier than the old ones, anyway. Just use the arrow keys and spacebar to complete it
>>107742230New autosolvers popped up in no time. What now?
Users of all levels are welcome to ask questions about GNU/Linux and share experiences.*** Please be civil, notice the "Friendly" in every Friendly GNU/Linux Thread ***Before asking for help, please check our list of resources.If you would like to try out GNU/Linux you can do one of the following:0) Install a GNU/Linux distribution of your choice in a Virtual Machine.1) Install a GNU/Linux distribution of your choice on bare metal and run your previous OS in a Virtual Machine.2) Use a live image and to boot directly into the GNU/Linux distribution without installing anything.3) Go balls deep and replace everything with GNU/Linux.Resources: Please spend at least a minute to check a web search engine with your question.Many free software projects have active mailing lists.Comment too long. Click here to view the full text.
>>107741339>>107741644Try Kubuntu 25.10, even the live environment is functional enough to get an idea of how it'd work once installed.You can suspend/resume or attach extra monitors and see how stuff behaves on your hardware.
I still distro hop even after 15 years of using Linux. What do?
>>107742240You already know.
>>107742240Make your own distro.
Are distros like Gnome OS the future of Linux?
Do any of you use this? Is it usable in practice?
>>107741241>nta but I assume you could compromise QEMU, and QEMU has had a bunch of vulns over the years.there's difference between QEMU and KVM. KVM is the one doing the actual virtualization on Linux. Xen (type-1) is an alternative to KVM (type-2). If you use VirtIO devices (Xen uses Xenback drivers with optional VirtIO support) and keep the QEMU devices and features minimum (you could even compile it with Clang's CFI), it's not really bad. Look at where the QEMU vulns detected at. Then there's Crosvm that's more security-oriented and used in ChromeOS.>maybe you could also compromise some distro packages, or even qubes repos.it's not included in the threat model of Qubes OS. it's not a totally zero trust OS and you have to assume Debian or Fedora project didn't get compromised.
>>107741355you have ready whonix virtual machine file on their official website
>>107738449>>107741580>>107741355>>107740673>>107738483its XEN, not shitty QEMU or KVM. NOT LINUXYou have to compromise XEN from dom0 (VERY UNLIKELY)YOU HAVE WHONIX BUILT-IN in QUBES. IT's LITERALLY THERE, and you can choose to update the system through the whonix/tor network.whonix in qubes is the best and comfiest way of browsing the darknet
>>107739339how do you use mdns inside a qube? i selfhost shit and i cant rely on ip.
>>107742007I wouldn't particulary call KVM shit. it's used at server space and fuzzed by google so that has to mean something. also Xen isn't exactly a good one on its own lane. it's still a relatively big monolithic kernel that isn't written with significant formalisms. though they're trying to improve it at least, by bringing some MISRA-C rules to the codebase for example.Still though in the end qubes is the only practical OS with relatively good security all around. You don't have to fuck around and worry if you left out something. Like having to passthrough network / usb to VMs like Xen's dom0 disaggregation and I'm not even sure there's a written code to proxy USB events safely from VM to host right now.
Forget about the AI slop, now everyone and their mother uploads their videos in 60 fpsLike that retard mental outlaw his videos are mostly just images and he still makes them 60 fps
>>107741958i must have missed the part where that's a bad thing. 60fps should be the bare minimum for all video
Ai has ruined breast feeding tutorials on Youtubenow its the exact same instructions with random Ai generating women giving them.and whats worse is indians sometimes splice in videos of them jerking off into their Ai breast feeding videos
>>107742301anon
>>107729974Was the rest supposed to change what happened? What did happen, anyway?
/biz/ herre.How is AI not another dot-com bubble?What are its practical applications, aside from porn generation and helping high schoolers with their homework? How does it make money?
>>107741602>Haha, can you imagine what it will do for us in just a couple of years?Total surveillance, that's the greatest part
>>107741602destroy the usa economy when colapsing, i just hope the rest of the west doesnt follow, bc it will be a shitshow to recover
>>107737777>How is AI not another dot-com bubble?It is but the dotcom bubble still lead to web 2.0 boom. Same is bound to happen with Ai
>>107737777>practical applications/biz/ wouldn't be asking dumb ass questions like this. Practical applications are not what makes lines go up
>>107740874NASA is just a gay military front, and has been almost since it's inception. Retards braying about NASA funding cuts never seem to realize that most shuttle missions had some sort of glownigger payload, and fully exclude the possibility that it's simply the elimination of redundant programs running independently and concurrently across military and "civilian" space programs.>Inb4 muh moon landing wasn't military For sure it was. If we can launch what was essentially an ICBM with a detachable bus and land that detachable bus on the fucking moon, we could 100% fill that detachable bus full of nuclear warheads and drop it on Moscow. Look up when we started to MIRV ICBMs and compare that to the timeline for the Apollo Program.
>If you're interested in the specifics as to what Windows collects from your machine, you can actually download Diagnostic Data Viewer from Windows Store, which provides full insight into telemetry data, unencrypted.>Anyone who has actually looked into diagnostic data & telemetry will tell you that it's absolutely nothing to be worried about. The ones who say Windows is spying on everything you do spread unwarranted FUD that has been proven incorrect again and again.
>No you don't understand we have to spy on your for legal reasonsmkay don't think I'll be using your product then
>>107742174Yes, you can simply not use Microsoft's cocreator.
Don't care still using Linux
>>107742218or any other product from a company who does that
>>107742009>They spy on the images you generate using their AI yesNot sure about this, might just be the prompts and not the images. Either way I don't care, this is a total tangent.
are YOU learning how to re-ball GPUs anon?You will have to make that 4090 last another decade. Basic electrical skills are going to be required once the jews price you out of new hardware, until the markets realign.
reball vram is easy, reballing core? fuck that just buy a new gpu.
Everyone doing this DIY will be dead in 20 years from various chemical exposure Reminder: these were built in China
>>107737018Instead of buying an entire soldering station, desoldering station, a digital microscope, a soldering station for tiny chips, an oscillator for thousamds of bucks I'll just buy a new card
I HATE BGA
Just stop being poor and just buy a replacement. It's not hard.
How is RedoxOS coming along? It'd be nice to have a usable microkernel operating system.
>>107741408I don't see any reason why performance should be worse than on a monolithic kernel other than the fact that they tend to receive less development and support.
>>107740569Windows will be fully Rust before Redox is complete.
>>107741514Okay but my interest is more in the microkernel and security features than Rust. I'd extend it to GNU Hurd if I had any faith in it ever being useable.
>>107741499That's because monolithic kernels have never been tried. No, not even Linux. It's a hybrid LARPing as a monokernel, but it's not. The Finnish retard copied the API from a self-admitted microkernel (Minix), whose explicit purpose was to keep complicated shit out of kernelspace.A true monokernel wouldn't block on open and reads. A true monokernel would allow you to create sockets and FDs in batches. A true monokernel would have compound functions (open + mmap + close) for the most common patterns.
>>107742231Also, a true monokernel would've probably used something like shared mappings between user and kernel to properly utilize memory layout and out-of-order execution, and write-lock the plot upon submission to make sure that the thread cannot modify the data in mid-flight. *And our current hardware doesn't support that very well* because the page table is a process-wide resource, not a thread-local one - meaning that, if you write-lock pages, you end up with per-core TLB shootdowns that can easily take much longer to complete than just copying all parameters from userspace to kernelspace, even if it's an in-order copy.Oh, and speaking of TLB shootdowns:>batched mmap
