Ai porn addicts are to blame for this
Just steal things
>AI porn addicts>look inside>jews and jeets
gpus are the new gold bullion
Actually its your shit economy and retarded tech companies and your incompetent government. Biden and Trump literally killed gaming. Its the greatest anime betrayal of all time from Trump's end. His biggest supporters, the ones who made it all possible, gamers, stabbed in the back. Never forghetti, rest in spaghetti
sorry
750 for a shit tier gpu is beyond the pale.
>>543673370that kind of money you should just buy a used 5070 ti
>>543673576it would be spics too if they were smart enough to learn how to do it
>>543673370buy a used 3080 it's lightyears better
>>543674452The "spics" are all suspiciously jewish.So no. Your addendum is invalid. Return to india.
uhhhhm sweaty, if they still had affordable tech for you, then china would win the ai race. you dont want china to win do you????
>>543674522yep im not jeet but since you feel targeted by what i said you are 100% a spic
>>543673370Damn I bought a 5070ti for that over a year ago LOL. Then again, if you don't need CUDA for genning, why the fuck would you even be looking at that card? You can get a 9070xt with the same amount of vram for the same price. You are probably just a salty poorfag that has to goon to disgusting dried up pornhub whores. I went over there just for old times sake a few days ago and my god, its nothing but peasant slop. I don't know how I ever lived without genning.
>>543673370video card fetish
>>543673370How the fuck do you guys even make AI porn, all the AI apps have rules against nsfw content.
You can rent a machine for like $5/hr with 128+GB of VRAM. Or under $100/month for a 32-64GB VRAM for a dedicated offsite rig. You can run basically every new game on 4GB VRAM no issue. Whats the fucking problem again?
>>543677811This is great until everyone has to do it. Then computing becomes a terminal and an access token to your government approved hardware and software. Its like you haven't read anything or had any second or third order thoughts.
>>543677957Haven't self hosted a website since the 90s. Don't care to go back to it. All trained weight networks can run on ANY sized ram or VRAM, only speed is effected, so there will never be a time when "computing" can be locked up with an access token....because to access that terminal line....you need a computer....which means you own some compute. This blip in price action doesn't matter at all. The chemistry factories being built like datacenters should worry people far far more.
>>543674486I bought a used pc of some nerd with a 3080 and i9 for $1300 before the AI apocalypse. Im very happy with it, plays everything and new games are overrated anyway. I'm playing dusklight the Legend of Zelda Twilight Princess PC port with 4k textures and its amazing.
>>543673370Autistic gooners are singlehandedly destroying the porn industry and you're blackpilling?
>>543673370I hate living in the 2020s
>>543678404>it can never be locked upExcept it can. If the components themselves are significantly more expensive than the ordinary person can reasonably buy then they will be forced to use someone elses equipment to do the work they need to do. Which comes in two forms; cyber cafes or cloud services.>terminal is worth some computeIt is, but not nearly enough to run anything of note or within a reasonable timeframe as you said.A completely compute locked system would be something like: a network card (which connects to one non configurable server), a screen and a graphics interface (not an actual processing graphics card) that takes XY sized buffers and places it on screen. There won't be programming the terminal, there wont be putting custom software on it, there won't be sufficient RAM to do anything but process network packets and there won't be an accessible terminal line. We're talking raspberry pi levels but without a functioning OS, little to no RAM and only 1 hardware level connectable network. Almost working like an FPGA. Now I'm not going to say that someone wouldn't be able to hack it but why even bother going down the avenue of all this when we could just not rely on third parties to provide all consumer compute and not balloon the prices on consumer grade equipment. Its a bit like everyone getting rid of their household tools because the handyman can do all the work, then the handyman goes out of town and nobody can fix their leaking pipes anymore, or the handyman decides to 5x the cost because he has everyone over a barrel. Why get rid of the tools in the first place?
>>543674144you can play stardew valley with integrated graphics tho
>he doesn't gen his own porn nowSucks to suck
>>543679079You think this is bad? Wait until the 2030's
>>543680038nice hand retard
>>543677811>You can run basically every new game on 4GB VRAM no issue.Lmao no, especially with all the UE5 goyslop. 6 at least, for lowest 1080p of course.
>>543679467>within a reasonable timeframeThis is the sticking point. Any code a normal person, or small team of people, would write right now (or in the next few decades) will likely be single threaded. A $40 cpu will compute basically just as fast as a $2k one. Lets say you multithread it. Great. You are still going to be cache locked. Lets ignore that though for the sake of argument. So youre super duper program runs 10x slower than an "approved computing device"....lets say 100x or 1000x. What super duper important computation requires a turn around in seconds rather than hours or days? I've never run into one. I've never cared if a simulation takes a few days to complete. What is the problem?>FPGAand? You are coding in an abstracted language. You don't care about chip level instructions. Will the FPGA have a compiler? of course it will. It MUST. So its just a FLOPS machine again. The CPU isn't special, it adds, multiplies, compares and holds registers....all things a FPGA will let you do. >But muh open-weight networksAgain. Why would a AI assistant need to be in the seconds for response time? If cursor took overnight instead of a few minutes, or if my AI email assisant/scheduler needed a dedicated raspberry pi....what changes? I understand your feeling that something is being taken...because thats what the news is saying, and these prices are annoying. No disagreement there. But besides gaming on nice settings, compute for normal coders is stupid cheap. And if you are coding for the GPU, you are not normal, so this wouldn't matter either. The compute crunch is good. Hopefully it'll push companies and products to actually care about resource useage for the first time in 2 decades. Why is a chromium tab taking GBs of RAM to begin with?!?!?! Thats the real fucking problem here, the software bloat. Not the hardware crunch. Single Threaded Applications are still the goto paradigm, and single threaded performance has stalled
>>543673370I haven't found a need for a new graphics card. mine is still the one I've had from 2014
>>543680194You wouldn't be focused on the hand in the full image, retard.
>>543673370>Implying they can't make a purpose built card for AI gen and are forced with guns to their head to inflate the prices on the consumer market and get rich from it.Also, AI porn is rubbish. Any professional gooner knows that 80's/90's porn is S tier.
>>543673370i hope it gets even more expensive. poorfags need to be culled from the gene pool.
>>543675117>You can get a 9070xt with the same amount of vram for the same priceYeah the nVidiot tax is real.
>>543673576Putting jeets in the same intelligence bracket as kikes is a little ambitious dont you think rajeesh. Tone it down a notch
>>543680552nah, late 70's early 80's is S tier
>>543673370I don't blame the gooners. They aren't using most of it. It's dumb third worlders and vibe coders mainly.But I'm still going to blame the companies. Most notably the dram producers.Samsung, SK Hynix, and Micron need their entire C-suite publicly executed.
>>543680676They're practically the same.>zero future planning>delusional gloating vs begging for acceptance>let me tell you why you must include me...>nigger dick obsessionThe hypothesis that Jews are a dalit offshoot is quite compelling.
>>543675117I'll argue for even using AMD for genning. It takes tinkering, but at this point ROCm is pretty much good for anything that isn't actual training/finetuning, which most don't do.
>>543680546i think you're just lazy and using a bad image model
>>543673370>WAAAAAAH!! I CAN'T PLAY THE NEWEST VIDYA GAME THAT SWEET BABY INC SAID WAS BASED!"
>>543677204You run local models like H3 Minimax, Qwen, etc. through comfyUI on your desktop. The ones like Grok and ChatGPT don’t let you make porn. That’s why AI gooners all have 5090s. You need a lot of computing power to run those local models.
when crypto mining was still on gpus like 6 years ago, i sold my 2080 super for 900 bucks. 3000 series came out within the year and i bought it for basically free. been on the 3080 ever since. feels good man. thank you for reading my story
