AMD is coming to the rescue
>AMD>Beating NvidiaI'll take things that will never happen for 100.
>>750533307AMD and NVIDIAS CEO"s are cousins btwNo collusion or insider trading whatsoever
>>750533397Intel would like to have a word.
>>750533307>AMD>I have no drivers and my fans must scream
I thought people don't use AMD because they dont have CUDA which is used for Ai? (forgive me if I'm wrong, this is just a general thing I've heard)
At no point has AMD ever competed with nvidia at all in history for the high end of the GPU market. Thats why you either pay the good goy tax or make do with AMDs "good enough" offering. Mid market is a different story with some good generations for each brand and some real fucking shitty ones
>>750533473CUDA isn't used for AI it's used for things like 3D modeling and other professional software. Which is true, AMD has no match for. Although in gaming it's a different story AMD just has to match things like DLSS.
>>750533451Intel had issues with 10nm and didn't fix their shit until comet lake.
>>750533473cuda isn't mandatory, amd has rocm/hip but cuda is much more mature, especially on windows.
>>750533307>we're going to be priced out out of quality GPU's thanks to AI and forced to use AMD shitchrist
>>750533307We could all be free of these fucking tech companies if devs choose to optimize their games for low end hardware
>>750533307can they fucking stop making new cards?
>>750533409Basically the same thing as WW1's leaders.
>>750533307Ok but how much will it cost with VRAM being so expensive now
>>750533669That would mean hiring competent white and yellow men instead of women and jeets.
>>750533603Only on super high end level. For normal video game player consumers who might do some 3d stuff as hobby there isn't much difference
>>750533307>bugman living off of internet engagement says something completely absurd
>>75053375810000 dollars minimum prescalped.
>>750533603>CUDA isn't used for AIIt absolutely is used for AI. In fact it's what powers the majority of AI in real world use.
>>750533603>dlssgood morning saar. graphic are in being still raster and amd raster very good very nice.
>>750533397you have to be dumb to believe thisquite literally the only thing that's stopping amd from beating nvidia next gen is their willingness to make a monster die that will rape the 6090nvidia is limited since it'll be monolithic while amd has the space for extra shaders from separating the memory out of the die
>>750533642how am i supposed to gen furry porn without cuda
>>750533603>CUDA isn't used for AI>AMD has no match for.you have zero clue what you're talking about
>>750533958>quite literally the only thing that's stopping amd from beating nvidia next gen is their willingness to make a monster die that will rape the 6090And it would slurp 800 watts and heat your home at the time.
>>750533307the price, officer?
>>750533960use rocm on linux or the vulkan backend of stablediffusion.cpp
>>750533603>AMD has no match for CUDAAMD has an equivalent. ROCm might not be as ubiquitous as CUDA but it's not like AMD is walking around empty-handed.https://www.amd.com/en/products/software/rocm.html
>>750534061you control the voltage and frequency your gpu usesat least they'd let their board partners use 8 pins instead of forcing them onto 12vhpwr
>>750533958Not him but I don't think amd wants to compete in high end levels. Best bang for your buck is what they're known for. Not for the most cutting edge and most expensive stuff
>>750533307I don't need "top" graphics card. I need affordable one that is "good enough".
>ConsumerFuck off. China will save consumer gaming.
>>750533307just give me faster 9070xt with 16gbs wtf
>>750534292saar just buy a 3060 12gb saar
>>750533463wasn't it nvidia drivers not that long ago that stopped fans from running and burned cards?
>>750534264why wouldn't they? the halo effect sells cardsthey were sandbagging rdna3 and rdna4 because they don't want to dedicate wafers to a gamer only architectureudna will be usable in both instincts and radeons
>>750534310The wumaos not being subtle at all today.
>>750533463nvidia has dogshit linux drivers and windows 10 drivers are EOL lmaonow you have to use the prime jeetware that is win11
>>750533307>AMD>Selling a card with thousands of dollars of VRAM soldered to it for consumersLmao. Not happening. Best you'll get is the new 9050.
>>750533831Dude, having AMD is like not having GPU at all in most 3D envoriomentsEven fucking Premiere works like you dont have GPUCUDA monopoly is what completely wrecked AMDIf you only game you can pickIf you work, its Nvidia or get fucked
>>750533307>top consumer card>36GBOkay but WHY? I know the answer is probably "why not?" but imagine the fucking cost of that thing even with AMD's more sensible pricing.Not a single game out there needs 36gigs of vram.
>>750533307>that'll be $6000>>still fails to beat rtx5090bless amd
>>750533451didn't intel scrap their gpu development again?
>>750534527FYI amd was the only manufacturer to offer HBM gpus to gamers. And they were pretty good, r9nano was a single slot card.
>>750533307Melty connector status?
>>750534534You need 36 gigs of VRAM for AI Loli's.
>>750534715amd doesn't force their board partners to use 12vhpwrsapphire and asrock chose to use that garbage on their high end cards
>>750534534AMD does market their GPUs to creators from time to time, not just for gaming.http://www.youtube.com/watch?v=BKzty6YdRNM
>>750534527but 4GB 9050 isn't even sold separately, it's only for IEMs to scam with their prebuilts
>>750534534No game needed 24GB VRAM back when the 7900 XTX and 4090 released either, but video editors and 3D artists were pretty happy.
>>750534739That's what all those cheap A100's are for. Can get 2173 for the price of just one 10090!
>>750534142then when i install any extension it assumes cuda and doesn’t work
Should I get a 9070 xt for the time being?
>>750533307>36GB VRAM$6000 RRP
>>750534963>A100>CheapHmm...
>>750534142Even a 5060 ti beats a 9070xt at image and video gen.
>>750534529Mostly use fusion 360. Works the same on my nvidia and radeon setup. Well radeon works better because it's newer and more powerful than the nvidia I've.Haven't been into video editing in ages but something like handbrake no problems. I don't see why premiere wouldn't work. Maybe in after effects if you go super crazy. But I'm talking about hobby stuff, not professional jobs.
>>750534998You can't wait a year?
>>750534998the base 9070 has 90% of the performance and is way more efficient out of the boxonly get the xt if you can get it for 10-15% more than the base 9070
>>750534998It’s the best high end price/performance card on the market. I specially if you don’t care that much about frame gen or heavy raytracing.
>>750533307AMD coming back to high-end GPUs? Absolutely based. I'm pleasantly surprised. Nvidia shouldn't be the only option in that range.
>>750533307Until they make CUDA it's no use.
>>750534994what extension, don't use comfyui it's cuda only unless you use a rocm fork last time I checked but they always worked half the time as far as I triedget the vulkan build of sd.cpp for linux or windows it works perfectly https://github.com/leejet/stable-diffusion.cpp/releases
Reminder that once upon a time only idiots and poorfags bought AMD cpus over Intel ones. Nvidia can absolutely lose if they abdicate the market.
>>750535164CUDA is proprietary and Nvidia will most likely never let them use it.
Good. The 7900 XTX needs a proper successor.
>>750533409Based.
>>750535164Doesn’t matter for consumer gaming cards.
>>750534479>linuxlmao
>>750535164ROCm is getting better every day.https://github.com/ROCm
>>750535028Yes, it's called a joke, it was about the state of the industries because of ram prices and the ever increasing demand for ram for AI.I thought the '2173' would be an indicator of its status as a jape. Clearly the number was too realistic and you misunderstood.
>>750535272>Not a single rothschild on the listIt's not the rich people you know about you should be afraid of, it's the ones you are barely aware of.
>>750533307>only 36 gigsmeh
>>750535287If you want a "consumer gaming card" why the fuck do you need 36gb and the best card in the world? You're still fine with practically anything made in the last 10 years.
>>750535178sdwebui/forge/reforge extensions don’t bother to support rocm
>>750533958>>750533451>>750533473>>750533527>>750533603No one will EVER beat Nvidia. They rich enough to buy the whole India.
>>750535413oldgen. big tech mogs
>>750535530if you want advanced features like training you either go cuda or google collab. if all you want is functional inference go for sd.cpp
>>750535331dogshit source, you lose
>>750533307>10900XTMore like $10900 USD. This shit is going to cost a fucking fortune with that much VRAM
>>750535331I assume you can provide a trusworthy source for this, winjeet. Here's mine: https://flathub.org/en/statisticsBarely any indians using flatpaks from flathub if we take population into consideration.
>>750535352>claude, here's the CUDA binary you know what to do
>>750535652>Replying to obvious baitDon't give him more (You)s
>>750535623i just want orange button go brrrrrrrrrrrrr
>>750535703Who the fuck cares about physics. Nvidia won by invested in AI. Did you even see AMD on that chart ?
>>750535572The rothschilds have assets worth more money than all those people combined, buddy.The only thing big tech ceos mog is showing their fucking faces everywhere like retards.
>36GBAIfags will love it.
>>750533307>AMD will release their next high-end GPU before nV
>>750535561physics can beat nvidia, dumbassnvidia doesn't have the means to put the memory onto separate diesamd doesif amd tried to actually make as big a gpu as possible then nvidia legitimately could not compete with them
>>750535626>>750535652hmm
>>750535352At molasses speeds. They're forever stuck playing catch up. Dunno if it's because they don't care about the consumer segment or if it's simply a money issue, but it's embarrassing either way.t. 6800xt owner
>>750535812It doesnt matter retard. AMD card are shit for anything else except Gayming. People dont buy GPU solely for gaming. Nvidia card can do everything else and thats why they won and become the richest company EVER.
FIVE HUNDRED AND NINETY NINE THOUSAND DOLLARS
Well, let's see where this goes. The last time I had AMD cards in my main desktop was back in the day when I ran 2 R9 290Xs in Crossfire. I've been on NVIDIA ever since, but if AMD actually releases a good product and NVIDIA drops the ball in the gaming space then I certainly won't mind switching.
>>750534534Games will start needing that much by 2030.
>>750534534>WHY DO YOU NEED MORE >JUST BUY A GPU FROM 6 YEARS AGO FOR 3X THE LAUNCH PRICEno
>>750535910>>750535784yes, faggot, it mattersdon't talk about tech if you haven't got the faintest inkling of an idea as to what the fuck you're talking aboutnvidia literally CANNOT beat a card that has so many CUs that they can't put enough SMs on the card to match itand no, your slop isn't safe either, look up what the U in UDNA stands for
>>750533307How many times must they perform this act before you understand all this means is:95-105% Nvidia performance (but only in gaming)Nvidia price - $100
>>750536124All that bells and whistles didnt matter much if AMD dont have good software to back it up. Aka Nvidia CUDA
>>750533307I'd seriously consider that. Nvidia are becoming snobby elitists that simply don't care about individuals, they only are about enterprise customers.
>>750536263cuda is a proprietary pile of shit that will get extinguished as linux usage goes up
>>750535063youare not rendering in fusion thats the pointFor Solidworks/Fursion/Revvit drawing works its not a problem, CUDA is rendering thing not a workflow thingApart in premiere where 3050 eats 6900xt in performance
>>750535738then download sd.cpp vulkan
>>750536308Nvidia is 5 trillion company thanks to CUDA. Keep coping anon. AMD is NVIDIA little bitch so NVIDIA can avoid Monopoly
>>750536267If you had a banana that you could sell for $1 to an individual or you could sell it for $100 to a corporation, what would you do?
>>750536427you're a retarded jeet, kill yourselfnvidia is a 5 trillion company thanks to partnerships with AI slop manufacturers, they sell them shovels to bury themselves with
>>750533397if they focus just on compute and raster they will win, people who aren't retarded never turn on any upscaling unless it's DLAA
>>750533307>36GB of VRAMthat will be 10,000 dollaridoos please, gaming is saved
>>750534596i would be surprised if it didn't, we have advanced many nodes since the 5090 which already came out on an outdate process
>>750536516AI is working thanks to NVIDIA CUDA capabilities. Good luck doing AI with AMD GPU lol
>>750536548>never turn on any upscaling unless it's DLAAWhat's wrong with enjoying free frames without the overhead of overclocking?
>>750536449grow more bananas until i have enough excess to sell some to individuals
>>750536627ai is working thanks to nvidia being there and being bigger than amd who fucked themselves by having an architectural separation between gaming and computenvidia doesn't deserve any credit, they were there and just existed
>>750536646if you don't care about how a game looks and just how it feels then fine, but most people don't like fine lines looking staggered and random things being grainy
>>750533307It's interesting that suddenly, now that it got extremely expensive, they are putting a fuckton of vram into gpus. Just few years back you had everyone shilling with "oh vram isn't that much important, clock speed is needed more".
>>750533931>>750534019Is it not the tesnor cores that matter more than the CUDA? >>750533941I'm white and own a 5080. >>750534156I thought HIP was the amd version?
>>750534479I don't care about tranny shit. Pro stuff only.
>>750536747>I thought HIP was the amd version?HIP is a subset of ROCm.https://rocm.docs.amd.com/projects/HIP/en/latest/what_is_hip.html
>>750533958AMD GPU goes back to being monolithic for this gen.Lightning doesn't strike twice.
>>750536705Because of Nvidia CUDA. You gonna think AI gonna use OPENCL ? Lmao
>>750536619I will believe it when I see it and still keep my rtx5090 because I got it before the prices went up
>>750536870no, it doesn't rdna4 is monolithic only because it's a stopgap generation
>>750536747Yes it is, these people are stupid.
Niiiiice. Just need a bit more over my 5070ti and I'll probably be set at 4k for a very long time. I also swapped to linux so.
>>750536882ai will use whatever it's programmed forNvidia was there to swoop in with cuda jewry
Literally zero games are worth upgrading for. What's the point? Playing something I already can play at higher res and higher fps. Yeah, that's not worth 3 months wages.
>>750533307Nvidia doesn't care about gayming anymore so AMD releasing anything doesn't really count if the other side has literally all but quit the gayming space.It's different with Intlel because Intlel tried to compete against Ryzen and then lost bigly. Nvidia is beyond gayming now and they don't give a fuck even if AMD has a 5080/5090 killer.Since its AMD it'll still be expensive and unattainable due to the same RAM shortages. Nvidia, RAM makers and AI techbros have essentially killed PC gaming.
>>750533409Yeah theyre the same team of pilfered post grads from SGI
>>750536747>Is it not the tesnor cores that matter more than the CUDA?yes but you program them through cuda
>>750537075CUDA was good even before AI. They used for /3/ and /tv/ stuff.AMD is shit because theyre incompetent bro. Just accept the truth already. No one can beat Nvidia unless AMD makes their own CUDA
>>750533307>starting price: $10,000 USD
>>750533307I hope it happens, i also hope the AI bubble eventually pops and we have Nvidia crawling back to us after this happens AND i want someone to use the AI to make the "it´s just good business" meme when it happens. I can wait.
>>750536747>I thought HIP was the amd version?it is, it's part of the rocm which is often used as synonymous for hip
>>750537427Can you wait until 2040?
>>750537253anyone could make their own cuda if not for nvidia's proprietary and monopolistic implementationyou are simply a jeet way out of your depth here, shut the fuck up
>>750537679>anyone could make their own cudaThey cant. Despite Nvidia making trillions, they still continue to develop proper software unlike AyyMD. Cope
>>750537864they don't develop shitthey just shut down competition by using proprietary garbage exclusive to their platform and bribe corpos into using itand you're a jeet shill that supports this
>>750537942*shut down competition by using better software compared to their competitorFixed that for you. Cope2X.
>>750538131nope, just briberynvidia is the same type of jeets as microsoft, embrace, extend, extinguish
>>750537075Wrong. Nvidia has been investing in AI related research and scooping researchers and buying up patents since at last 2010. They have correctly predicted the trends.
>>750538186>BriberyZero proof. Cope3X. Keep crying lol waaah waaah waaah lol
>>750538195>Wrong, it was... patent jewrylol
>>750538229nvidia has a rich history of bribery, my dear little jeet
>>750538195He will keep coping no matter what you say to him anon. His fault for get scammed by buying AMD GPU. Let him cry what he wants i want to drink his tears
>instantly sold out due to scalpers>relisted for 6k+>all future releases will be 5k+ for the base model cardIf you didn't build a PC when you had the chance you'll never build one in the future.
>>750533831this, the average joe just wants Counterstrike to run at 240fps and dont give a shit beyond that.
>>750538292Like ??????Cope4X
AMD could beat Nvidia by releasing a card that has 0 raytracing features. Just design all the metal in the card to render triangles and shaders as efficiently as possible.
>>750533831CUDA is used for AI. Im using CUDA for ComfyUI video gen
>>750538348gameworks, geforce partner program just off the top of my head, you brain damaged little jeet
>>750538410>can't play Indiana Jones and the Great CircleYikes.
>>750538441Thats not bribery, thats >>750538131Cope5X
>>750538468>playing any indy game that isnt the lucasarts originalsZoinks Scoob!
>>750538502>that's not bribery because... i don't even know what you're talking about!find elsewhere to shit in the street, nvidiajeet
>>750538410Like a GTX 1660? Fine, as long as they also release a card *with* RT so I can use it for Blender.
>>750538551Done crying ? Cope6XNext time dont listen on /v/ for GPU advice and buy NVIDIA
>>750537227Yeah but the heavy lifting is the tensor cores. Without tensor cores we don't have LLMs as we know them right now. They do all the matrix calculations. That's the difference of pre 1000 series to now and why shit went crazy it started in 2017.
>>750533397Just to be clear, Nvidia and AMD build their cards out of the same factory in Taiwan. AMD as a brand has always targeted a more budget-conscious market because it's more economical to target an underserved niche that to try and and compete with an established brand in its established niche with what would ostensibly be a near-identical product.It's not that AMD is incapable of producing a top-end card, it just never made sense for them to focus on competing for the top end.Nvidia's naked contempt for its own consumers has left its niche wide open and its overwrought focus on AI performance over conventional metrics means it doesn't even really have products to offer right now.
>>750538701sorry, i don't endorse brownoid tech, which is why i bought a card that appeals to white sensibilities, the 9070xtjeets like you don't mind the melting connector or the dogshit drivers where you can only use windows 11but white men want linux support
>>750538779Lol. Mindbroken.
>>750538834you lost and you can't handle itnvidia cards are for the brown and the retarded
>>750538872>Nvidia card literally more valued than Gold >So valued people literally make a GTA Online tier heist for it>LostLMAOOOOOOOOOOOOOOOOO
>>750538975>nvidia rapes their customers so much that they've completely stopped producing consumer cards beyond entry level shit>this retard believes it's a good thing
>>750539067Datacenters are more valued. Its impossible to do both at same time.Nvidia literally the only company who profited from AI.This is why its on you for waiting lol.
>>750533307>36 GB VRAMso is the 10900 supposed to be the MSRP?
>>750533307>but what about the ram shortage>ram shortage?its like it was all fake all along.
>>750539265ok? you, as a nvidia shill, lose, since you're not a datacenterdo you think that worshipping nvidia will mean that their wealth is going to rub off on you? is this the level of izzat jeetery you're actually on?
>>750533307I saw a similar article like a month ago stating the R10700 will have 48GB of VRAM. But pretty sure that one was AI slop. Regardless, 36GB or 48GB, these cares are going to be unobtainable for gamers because VRAM capacity is king right now.Software support has matured immensely for AMD and you can get similar text inference speeds. No clue about video and image generation but I have a 5090, so I'm good there. AMD powers my text/TTS/STT server.
>>750539372I won because i bought the card before price gouge happen. Technically i can sell my 5090 at 10 times the price i bought before.Also your last sentence is pure seethe lmao. I just laught at that lmao
>>750539498lol, rightyou won because the price of your card will either crater because of next gen silicon or stay inflated as nvidia quits the consumer marketwhat a victory, it's free money, like selling a kidney
>>750539696Yes, i won. Now cry
>>750533307""consumer graphics card""
>>750539763nah, winning the paralympics still means you're a retarded cripple lmao
>>750539357The shortage is that they're diverting all their memory into workstation cards so the gaming gpus have none. The shortage is for the average consumer
>>750539372>that their wealth is going to rub off on youNo need to worship, you just had to buy stock a decade ago.
>>750539835>crying while saying thatlol
Fanboying over a company that doesn't even release consumer graphics cards anymore.Reaching new depths of cuckoldry.
>>750533931>Ai responcesasuga
>>750540263Install ComfyUI and see what happens
>>750533307>late 2027so gamers have to suffer for another 400 days?
>>750536548UE5 games have some level of innate upscaling or frame gen or DLAA or all 3 that is near impossible to turn off.Take ARK's remaster for example. Even if you use console commands, launch arguments, scripts, etc. to turn them off and set your resolution scale to 100 and use native everything you will still see smearing shit on particle and lighting effects.I know it's a beaten horse, but UE5 is just an unbelievably dogshit engine, that enables the absolute biggest retards to push out broken games. Simple as.
>>750534061Imagine saying this when nvidia cards have had 1000w+ transients for nearly 5 years now and are normalizing worse, unsafe power connectors that WILL start a house fire if they aren't installed perfectly flush lmfao
>>750540709Bending the wires too much can also cause issues
>>750540768Noooo but muh 2x6 revision that totally removed user error! Please ignore the fact that you need software to monitor each pin in realtime 24/7/365, to make sure your $20k GPU isn't going to immolate you and your family ;^)Fundamentally its an inferior power delivery method, but fanboys don't give a shit lol. jensen could come out and say the next line of consumer GPUs have a 20% chance of being loaded with plastic explosives that will detonate at a random time in the next year after purchase, show them literally loading 1/5 units with it and shipping them out, and they would still buy them lmfao
I remember when they said there was a holocaust in Auschwitz, too.
>>750533307>rescueTheres no rescue, we're talking about a potential $9999 card. If anyone here didn't make a new PC last year before shit hit the fan, then fuck it, look for another hobby or get into 2010's games, pretend its 2006, heat up some hot pockets, grab a mntn dew and play Oblivion and CoD MW.
>>750533682The 7900 XTX is overdue for a successor.
>>750534529>Dude, having AMD is like not having GPU at all in most 3D envoriomentsYou could model in Blender and then send to Unreal Engine for rendering.https://www.youtube.com/watch?v=iadFWjfTdyohttps://www.youtube.com/watch?v=dMLe-LX3vRE
>>750541948lol retardthe only reason the 5090 is $10k is because it's out of production
>>750533307>juts like they did with IntelIntel killed themself by doubling down on their faults and blaming (You).
>>750533307And it will only cost $7000
>>750533307lmao I would rather keep my 3080ti than downgrade to gaymd.
>>750542563if the 7900XTX was a battlemech, he would be the marauder
The 5080s are 3k now lmaoThe gpu market is cooked
>>750534998if you think you can afford a 5090 aka whatever the fuck amd is releasing in 2027 then wait if not then get a 9070xt
>>750544717I can get a 5080 for under $2k right now lol.
>>750533397For the longest time, Intel were the only CPU manufacturers people cared about. AMD equaled the playing field with Ryzen.It's very possible for a similar situation to occur if Nvidia indeed starts moving away from gaming and more into AI.The issue with AMD is, they kind of never miss an opportunity to miss an opportunity. Every time Nvidia fucks something up, AMD can't stay too far behind.So it's hard to predict what will happen. I'm content with my AMD card because it gets the job done and I haven't had any issues with it. But I wouldn't say either of those companies is perfect.
>>750534527>4gbBeyond fucking GRIM
>>750534534I think AMD wants to expand their consumerbase so they want to make their cards better for productivity reasons, not just for gaming.
>>750545368Link? This morning orders were getting canceled.
>>750534156>>750533642Consider the following: Apple Metal caught up with Nvidia and SURPASSED (in some cases) software-wise within just a few years on ARM chips.It's been decades, and AMD drivers and software is still fucking ass
>>750545529I mean I'm not gonna link you to my local microcenter because thats pointless but they have 2 models under $2k and a bunch under $2.5. In fact I can buy a prebuilt PC with a 5080 for exactly $3k right now with 32gb of ram a 9800x3d and a 2tb ssd.
>>750545567do applecattle genuinely believe this lmao
>>750545567And yet they arent you mongoloid nvidia employee, also apple sucks ass.
>>750533409This, whatever happens it has already been decided behind closed doors
>>750535063>gpu video encodingno one at all does this because the quality is utter dogshit
>>750545457Radeon will always find a way to fuck up.
>>750533307kekthese fags said the same thing about previous generation, nvidia -50 bucks. not worth it.
>>750546201Now it's nVidia -500 bucks.
>>750533409
>>750533307I DON'T NEED 36GB OF VRAM FOR FUCKING GAMINGAITARDS FUCK OFF REEEEEE
>36GB VRAMHow many Skyrim mods is that?
im guessing the price will be only $7999,99 + tip ofc
>>750538410I don't care about ray tracing or fake frame gen shit like fsr and dlss.
>>750546420Games are being made with these features in mind now. You'll be looking at shit playing without it.
>>750538410No one wants a feature that kills FPS...>>750538468You don't need it if your drivers have RT emulation
>>750533409The more you look into things, and often times doesn't even take that much effort, the more you realize how fake and gay everything is. It's a shame so much of it affects regular people, beyond just gaming/technology.
>>750545567The M5 Ultra's GPU beats out the 4090 and the 5080 in the Blender benchmarks and it's getting dangerously close to the 5090.https://opendata.blender.org/benchmarks/query/?compute_type=OPTIX&compute_type=CUDA&compute_type=HIP&compute_type=METAL&compute_type=ONEAPI&group_by=device_name&blender_version=5.2.0
>>750533307>36 GB of VRAMyeah this is just not going to happen lolMAYBE on their flagship to compete with 5090 which will be like $2500 minimum but no way on everything else
>>750546324>>750546379You boys are so funny and silly!
>>750534998it's probably the second best deal on a gpu today, after the regular 9070, unless you NEED nvidia's ai upscaling
>>750546626don't these things cost 20k?
>>750546883the cheapest one is like 5k and you get the whole fucking computer not just the gpu
>>750546883I thinks it's $11k for the 256GB unified memory version over here. That's a better deal than Nvidia >>750546626Yup, they're also MUCH better per WThe biggest limitation is the faggy form-factor they went with instead of a proper tower
>>750533463AMD has the best drivers unless unless you're a winshitter
>>750535903It's because your hardware can't handle modern number formats dumbass
>>750533758>>750533881Yeah AMD ain’t getting nothing without cheaper prices. Are they going to?
>>75053494924GB will be upper-mid once PS6 comes out. People buying top end GPUs for games often want something that lasts 7-10 years.
>>750533409never shoulda trust the goddamn chinese
>>750534998Don't buy a 16GB GPU
>>750547087They're also backordered into December. Uncle Sam really fucked the market and blamed it on AI
>>750546253You will once next gen consoles bump the floor up.
>>750550130You can get one with the right channelsI got one 17% off
>>750549842RTX 3000 series is roughly from the same time and it's getting plenty support from jewvidia.
>>750550991That's a different chip with different capabilities you fucking moron. There's no point trying to optimize AI on a chip with 8GB VRAM that can only process FP16.
>>750533307Nvidia literally stopped giving a shit about gaming gpu's and put those chips into the pro cards. AMD literally does the same shit as Nvidia and the CEO's are related. AMD isn't giving you shit because they don't give a shit about you.
>>750533463>AMD bad driver memesI bet you still call them ATI as well. Time to move on grandpa
>>750551443>>750549842>software support is le bad actually>just buy the latest product, goyAre you pretending to be a braindead fanboy, or...?
>>750533307so i can finally look forward to a new GPU worth getting an upgrade from the 7900XTX
>>750551630There's no point having software support for hardware that can't work. It's like you're upset you can't boot Linux directly from the GPU and type commands into the DVI port.
>>750535164>>750535728yeshttps://github.com/Void-Compute/AMD-Ghost-Enviromenthttps://github.com/Speedstu/CUDA-for-AMD-Windows
>ai discovered nvidia gpus had kernel level restrictions that halved their compute rate>retards still pay $2000 for a 5080lolLAMO
>>750552837AI can undo it thoughbeit
>>750552837>>ai discovered nvidia gpus had kernel level restrictions that halved their compute ratesource?
>>750553519Jewvidya uses the same production pipeline for both datacenter and consumer cards. They gimp performance in the latter via software. The signed firmware checks also keep Loonix from getting freetard drivers as you can only run goyvidya-approved drivers or else the cards won't reclock from low power mode
>>750535561I think the average american is rich enough to buy india
>>750552837>>750553743>ai discoveredno, retard, everyone already knew this.
>>750554097The thing is they've recently come up with an unlock that boosts performance (raises thermals as well, will likely need a better heatsink)I don't get how the government allows goyvidya to scam their customers like this
>>750554426It doesn't effect gaming performance at all, just local AI models on older cards
>>750534310>2 mole weeks!Fuck off chang.
>>750534527>4gb VRAM>64-bit memory bus
>>750536627nta but chinese labs use huawei hardware and google trains their models on their tpu's, so its not really a cuda thing, more of a finance/availability thing
Give China your energy and all your IPOnly they will save us
>>750533397fucking AHH zoomer. You didn't exist at that time but it happened
>>750533473software isnt a moat anymore ironically thanks to AI
>>750552837.....At the cost of Wattage being doubled + additional 50% chance to explode your GPUNothing is free
>>750549956its gonna be a low volume product, most of their money comes from consoles and datacenter parts anyway. I'm more interested in their low end dgpu's and apus honestly, 9060xt had amazing performance for a 32 CU chip, if they can double that for rdna5 it would be an awesome "budget" card
>>750533307Genuinely find it pathetic how some people spend their entire day discussing pc components. Amd releases some new obscure gpu tech that nobody will use and suddenly you have thousands of spergs arguing about how amd saved gaming or nvidia is killing the industry.I genuinely cannot comprehend how you can be so invested in what products some multibillion-dollar technology company is marketing this quarter. When I built my first desktop gaming pc, i bought a 4070 ti super, put the thing together and that was it. I wasn't autistically invested in every new technology announcement, memorizing sensor sizes, comparing transistor counts or following every fucking gpu release like it was the second coming of christ.What exactly do you get out of this? do you actually enjoy gaming or do you just enjoy discussing the hardware you might or might not buy someday?Are these mostly poorfags who get off masturbating over the thought of owning hardware they'll never actually afford, or do some of you genuinely have nothing better to do??
>>750545457>For the longest time, Intel were the only CPU manufacturers people cared about. AMD equaled the playing field with Ryzen.It happened in the early 2000's, too. P4 vs. AthXP generally favored the Athlon and it seemed for a while there that all the big gaming rigs were running Bartons. Intel finally abandoned Netburst and they traded blows for a while, but after the Core series took off, AMD was out for a long time.If memory serves, of course.
>>750553743>jewvidya>goyvidya>loonixYou do know speaking like that - like a juvenile retard - makes everyone disregard everything you say, right?
>>750554426its not a scam. consumer gpus are half precision. enterprise gpus are full precision. this is why workstation gpus are $40,000 and your 5050ti is $200
i struggle to find game past 8-10gb VRAM needed for games at ultra settings why would you need this much especially when next gen games are going to be held back by current consoles like the ps5/series X no publisher in their right mind would sell games only on next gen meaning all this VRAM is just going to go to waste not to mention the price in the current market they could say this thing is 500$ and it would instantly go for 3k upwards
>>750533307I'd rather just stick with what I've got than "Upgrade" to an AMD CPU, desu.Never again. They got me once, never again.
>>750558883Minimax H3 is My Video game and its struggling on 16gb VRAM
>>750558883I play 1440p for framerate and never see anything more than 10gb VRAM usage. 12gb is fine for ultra 1440p in any game
>>750557803it doesnt, not on this site, newniggerleave, faggot
>>750533409But seriously, who didnt see this coming?Nvidia will drop consumer markets and go all in on AI Datacenters because AMD ie the controlled opposition is there to follow up and monopolize the consumer market, making sure there's no niche for a third party to jump in. Both CEOs share the profits as they take over everything and control the market.
>>750559925>kike shifts the narrative
>>750533307>AMD is coming to the rescue!>by this time competing with only a somewhat outdated Nvidia card and dumping even more vram onto it!>totally for consumers btw please don't get eaten up by datacenters and stuff teehee
>>750533307Cool.Looking forward to people deciding whether to buy a used car or the new high end vidya card.
>AMD prices skyrocketing at my local pc shopsWhat a great rescue.
>>750533307>36 GB of VRAMI can't afford a GPU for $7000
>>750546237and -50% performance
>>750560003Wrong, dumbass. i just know what's upNeither Trump, not whatever blue face looking to replace him will let the AI bubble pop. It will not get better, deal
Intel is more likely to beat AMD than for NVIDIA to be dethroned. Though Intel is giving up the consumer market too it seems and going for datacenters. We'll see when the AI boom dies down if they decide to come back but for now I'm enjoying my Arc B70 and enjoyed my B580
>>750560454Nvidia literally said they aren't making new GPUs so you will either buy AMD or continue using your 3070 while telling everyone you plan to buy a 5080 when the prices become reasonable for the 8th year in a row.
>>750560573Nvidia can stop making new consumer GPU's for the next 5 years and AMD would barely start competing them by that 5th year. Then they would just roll out a new generation for consumers, if it's even viable by then, because everyone is chasing the datacenter market. The world is shifting away from consumers in real time as much as many justifiably loath cloud carbage.
Nobody with a functioning brain is going to willingly spend $2,000 on an AMD gaming gpu.
>>750561061Nope, but they will when the alternative is a $10000 Nigvidia GPU that's for AI more than gaming.
>my 3070 would let me game and watch stuff on my other, lower refresh rate, monitor fine>my 9070xt crashes often depending on the game I'm playing in similar situationsGreat otherwise but holy shit it's fucking annoying. Tried so many different settings and it still happens. AMD still has quite a ways before upstaging Nvidia.
>36GBimplies a 384 bit bus and gddr7even if it's usual AMD craptastic fashion of poor mem controller and the cheapest chips that implies 28gbps chips and a 1.34TB/s bandwidthThese will get snatched up by people who want to run AI and the price will skyrocket.But it's a bullshit rumor anyways unless Nvidia all but tells everyone they are exiting the gaming gpu market.
>>750533307No amount of VRAM is going to beat DLSS. Nothing AMD has is close to DLSS 4.5. I'm not even an Nvidia shill, I usually alternate between AMD and Nvidia cards when upgrading, and was about to buy a 9070XT before I managed to get a 5080 at MSRP. DLSS 4.5 is insane, it makes the games run better while making them look better somehow. And it's not slop like DLSS 5, the game looks either identical to native, or even better. I have no idea why they didn't market that instead of going all in on frame gen and DLSS 5, which are way more divisive.
>>750533307Never buy an AMD GPU their drivers are complete dogshit
>>750561220>>my 9070xt crashes often depending on the game I'm playing in similar situationsWhy does /v/ have every single time 50 IQ when it comes to setting up PCs. Like Jesus you fucks are Orcs of Morder.
>>750562178>blah blah blahUnfortunately this isn't helpful for my situation so AMD is still behind Nvidia by quite a bit.
>>750561886FSR4 and DLSS is ported, you can stop glazing
>>750534479>nvidia has dogshit linux driversNvidia is all in on Linux now since it's what's best for AI. Their drivers are fine.
>>750535272Disgusting amounts of money, but surprisingly reasonable for Jensen, all things considered.
>>750561886I still have ptsd from using my 7xx/9xx/20xx ngreedia cards on linux, as nice as DLSS is I can't go back
>>750533307I use Linux and even I’m not buying amd
>>750533307you know what the fucked up part is its probably going to work Jenson knows that the goodwill and bridge hes burned will never be repaired so his going go ask his sister to pick up the slack as AI fumbles and collapses (China)
>>750562737supporting AI slop and supporting the general desktop and gaming experience are two completely different worlds
we won
>>750561652>implies a 384 bit bus and gddr7it's not exactly viable to make a 48gb 512bit card right nowwould be nice if they did support this level of flagship now that they can use the same dies on instincts
Just stick with 1080p gaming like a normal person and dont be bothered with anything for next 10 years
>>750559925How long can they realistically keep this up? Okay, you release a GPU with 30GB of VRAM. My 8gb VRAM GPU is running RDR2 on high graphics settings without an issue. For what reason would I ever in the next 50 years need a 64GB VRAM GPU? There will never be a game that will require anything like that. They can release their DLSS5's, RDNA's, and all the other cosmetic shit, but one day there will be no more reason to upgrade your/my PC. So who will these companies sell to then?
>>750533307the name is the price
>>750564017The use case is tricking consumers into thinking they are on their side, when they are just making a datacenter card. A PR stint, since Nvidia is starting to slash the upper end of their own consumer lineup.
>>750564017>but one day there will be no more reason to upgrade your/my PC. So who will these companies sell to then?Anon, we're already there. The only reason to go buy the latest most expensive card is ultra High ultra wide ultra gay graphics in your AAA garbage. Even the 5060 can run everything at 2K/144 fps.These cards sell on fad logic: they're new, therefore you HAVE to get them. You have no actual need for them, none whatsoever.
my 9800X3D and 9070 XT are perfect. upscaling, frame gen and ray tracing are all gimmicks for faggots.
>>750564319i have the exact same i only use FSR4 for Anti Aliasing, Ray tracing doesnt change anything in games im surprised, i still dont understand the need for frame gen it creates input latency and you already have 100+ frames why turn it on?
>>750564162I tested my old RX 580 and it can suprisingly run newer Unreal Engine slop at around 30-40fps, which is basically PS5 tier.
>>750533307If it won't cost a retarded sum of money, like 5090s do, then I will get it.
>>7505646985090's price comes from its nonexistent supplyat worst you can add a 32gb ram kit to find out what's the maximum the price would've risen if nvidia cared enough to keep making them and not funneling all gb202s to datacentres
>>750562603DLSS runs like shit on AMD cards.
I'm mildly hopeful because amd won me over with the 7900xtx.
>>750564017Inherent firmware compatibility issues would be the only reason to upgrade in the near future >Oi goy, new GPU is Windows Super Subscription exclusive only>And it's the best one we made>Pay us $3000!!!
>>750533307>36GB VRAMSo it'll be $5700 MSRP and every single one will be stolen off the truck instantly and sold for $14,500?
>>750564017>8gb is enough *(to play 10 year old game)>play saarwars outlaws>uses 19 gigs at all timesI know shit game, shitty optimised but that's the current year standard. Your 8gb is for 10 year old games. 2 years and rdr2 is considered retro.
>>750564017It's not just about VRAM moron. Performance also matters. Games get more bloated year after year despite the fact that graphics have long hit diminishing returns and improvements are marginal at best. Gotta keep making the goyim throw away working hardware
>>7505431145090s were selling for $5000 when there was still stock on several stores
>>750533307>36GBAlright so technical paper MSRP of $5999, a shelf price of $8999 permanently out of stock and all retail units stolen straight off the assembly line and up on ebay for $14,999?
>>750545457>>750533451AMD didn't beat Intel, Intel committed suicide. The only way AMD standa a chance against Nvidia is if they completely stop doing anything.
>>750533409>are cousinsNone of their parents are siblings with the parents of the other one.
>>750559925>Nvidia will drop consumer marketsYou could have just said that you are retarded instead of doing this whole expert larp.
>>750565105nvidia has been cutting back on stock for higher vram cards for months now, this is just stock being critically lowhttps://www.pcguide.com/news/nvidia-reportedly-cuts-back-on-16gb-gpus-shifts-focus-to-8gb-as-memory-shortages-cause-havoc/
>>750564017Mr. Gates, is this you?!
>>750534019He's right about AMD not being able to compete on those fronts at least. >>750534156>you don't need that, this shitty open source thing is just as goodfuck off
>>750534534VR really likes as much VRAM as you can get, I regularly max out the 16g on my 4080 Super trying to drive essentially two 4k monitors at the same time
>>750534534>AMD's more sensible pricing.A 20% worse card for 15% less isn't actually cheaper.
>>750536663Wow, retard. You're so smart. Turns out that making bananas is really hard and that scaling up banana fields takes a lot of money and time.
>>750565301cite your source
>>750537092>it's the people who have the best cards, the best support and the best features who don't care >not all the other companies that can't keep upCan somebody explain this logic? Are you a racing fan? Do you go "that guy who won probably just hates driving, that's why finished so quickly?".
>>750538320Scalpers aren't magic. They only work when there is less supply than demand. >>750539357Yeah, I'm sure you are right, the card will be very cheap and plentiful. You're not right, you're dumb.
>>750565387where's all the stock then?why is nvidia basically only making entry level shit now?
>>750546253>what do you mean this high end card has more than the bare minimum?crazy
>>750555917You bought an Nvidia card. Of course you're happy and don't feel the need to seethe about the industry all day.
>>750533463Works on win 10 no issues
>>750533307>36GBlol
>>750533473CUDA is really good for all sorts of data processing including AI, and as far as I know only NVidia cards can use it. Until that changes the only value these 36 gb AMD cards will have is maybe taking the ram out to mod nvidia cards with.
>>750533397I've only ever had Nvidia cards, but with SteamOS + Nvidia vibeshitting + Nvidia moving away from gaming focus + prices, I might give AMD a chance. If anything just to get off winshit once and for all
>>750566639amd has rocm and for the buck it's the best option for a regular anon
>>750533409Nepotism is the only way to play the game.
You are a sucker if you bought AMD. You got played. Now you can do nothing but gaming and even then you have all these driver issues and software problems. Just to save a few hundred $$. You get what you pay for, in the end
>>750567456you're a sucker if you bought jewvidiayou're stuck in the goycattle pen of using windows 11 and have a card with a connector that can just kill itself but it's ok because you fell for AI jeetslop
i really REALY wish they would give a hint as to the price of these things so people can save up for them more easily
>>750567560Save at least $5000 because it's going to be the minimum price for high-end GPUs.
>>750534310CXMT's RAM is literally on the same price as the other threes. They ain't saving shit.
>>750567690>judging prices based on nvidia's low stockpeak retard lmao
>>750567456I'm on Nvidia, but your post is retarded nigger. An AMD gpu and/or cpu is not going to stop you from any of the following>Computing>Video editing>Audio editing>Making high end/resolution digital art/illustrations>Streaming>Running your office shitAre you implying that it can't be used as effectively, as Nvidia cards, to generate photorealistic pedophilic imagery and videos via AI? Sure, but that's something 99% of people can live with
I'm currently thinking about rating gfx card warriors as the fanbase even more retarded than snoytendies. They still need a good name - maybe cardcattle?
I have been using my AMD card for AI stuff with ROCm for 3 years now.Literally just replace pytorch+cuda with pytorch+rocm and it works out of the box most of the time. Not as fast as a Nvidia though, but not an obstacle either.
>>750567726Retard-kun what do you think will happen with the price of AMD GPUs if prices of nV GPUs go even more insane?Look how much is 9070XT right now, it will get worse.
>>750567998What AI stuff are you doing?
>>750533409/thread
I have a friend that is I HATE AYMDD saar, which also happens to be console cuck,which has only had used generations old Novideo GPUs in his old PC,still he always larps how bad AYYMDDD saar redeem is,and this seem to be the case with all the cattle, all while praising raay traaaysin saar upscarlers saaar and using AYYYMDDD on their piece of shit consoles discussing GPUs here is like going to free Piglestine protests and trying to discuss well anything at all
>>750568113amd is actually making the 9070xt dumbass
>>750567456this.
>>750533307considering amd already has some pretty severe compatibility issues with various things, i dont think people are gonna get excited for this.
>>750567456>you are sucker if you bought ps5>you are sucker if you bought xbox>you are suc...oh waithehe
>>750568639ranjeet, we both know you have no computer at all
>>750533307For the low, low MSRP of $1199.
>>750535272How does Jensen not have more money than all of those? I'd expect Elon and Bezos to be at the top, but I always imagined Jensen at the top 5 at least. Literally everyone and everything IT related buys Nvidia - from gaymers to AI superclusters.
>>750538761>>750536308The reason NVIDIA is so dominant is because of CUDA, and CUDA is so dominant because everything has already been written for it over the years, all the libraries etc
>>750568163Image gen, model training, finetuning, text generation, vocal isolation.Everything is local. As long as it uses pytorch, it's doable.
>>750568937Pretty much. AMD is always behind trying to create a competitive solution but they take for fucking ever to deliver anything. Only now is ROCm working on Windows and is relatively easy to use, and over in the gaming side it took FIVE years to create a proper AI algorithm for frame generation and upscaling.It's like they don't have the engineering team, which is weird seeing how big and important they are.
Can AMD do DLSS5?
>>750550035>Don't buy a 16GB GPUWhy?
>>750549535Very funny you say that because Nvidia on Windows has been causing problems non-stop for over a year.
>>750568726That would be a great deal, won't happen.
>>750534142Rocm is also native on windows now as of the summer.
>>750569071I tried image gen but it tedious as fuck.I don't have a purpose for those images so I stopped.What I really wanted to use it for would be video upscaling, but storage is expensive as fuck as well.
>>750569408Which is why they make different cards for workstations and game stations
>>750569802Video upscaling requires a lot of VRAM. 16GB won't cut it.
>>750564017idk, your GPU and/or other components will not last for 50 years?
>>750560267hehe should've boughted crypto ;)
>>750560286>>750570167
>>750533307cool, I'll think about it again when it comes out.
>>750569947I don't even want to do 4K there is a lot of 480p shit, even 2x would be great.
>>750569521Child, you're confusing MSRP with actual price. That's the joke.
>>750570630That MSRP is very optimistic to begin with.Even at 200% of that it would be too cheap in the current market.
>>750570768Nta. All cards of this gen were overpriced for the current market. A supply issue skyrocketing prices doesn't change that. This gen was underpowered for its msrp.The current market is for retards who want to buy weak cards for stupid prices.
I'll wait for a few years and nab it once the card inevitably got cheaper, got my 7900XTX that way and I love this card to dead.
>>750533307>AMD is coming to rescue Nvidia tardsrescuing drowning retards is not a good idea, they will kill you.
>>750533397AMD has beaten Nvidia before. Nvidia just bounced back very quickly, often with unfinished chips, just so they could take back the performance crown.
>>750533397Do me a favor and remember you said this
>>7505709349070XT was decent at MSRP but it was fake MSRP. AMD could keep their prices high because nV had not a single decent card. 5090 is in its own league and everything else was just 4xxx +10% perf. Shame intel gave up.
>>750571271>600 bucksMhh yeah I never seen one at 600 bucks, that would have been a reasonable price.5090 is also just 10% perf (well it's 20 but that's also emberassing) basically this gen was just last gens oc editions but for next gen after next prices. And that alrdy at msrp. Everyone asking for even higher ones because we have a supply shortage has been giga mindraped by buyers remorse copium overdose.
>>750536427>AMD is NVIDIA little bitch so NVIDIA can avoid MonopolyThis. AMD's CEO is Jensen's cousin ffs
I'll use my 4060Ti 16Gb as a work horse until it kicks the bucket.
>>750571480>Mhh yeah I never seen one at 600 bucks, that would have been a reasonable price.I actually bought my Pulse for 640€ (tax included) over at Spanish Amazon. Pretty good deal, I think.
>>750533397It happened for us uncs
>>750571997That's ATi though.
>>750571271>5090 is in its own leagueThe problem is it's not worth buying. The price even at release was mind-numbingly stupid if you're not a multimillionaire
>>750572374If you bought it as a professional card for near MSRP it was a great deal.If you mostly play VR it was also awesome.It was still obtainable for lone nerds with disposable income in developed countries. The biggest problem with that card is the retarded melting connector.
>>750572374>multimillionairebro anyone with a part time job can afford it if they arent a complete retard with money
>>750572889I'd only buy a 5090 if I were a streamer or something. It is complete overkill for mere mortals and will become obsolete by the end of 2027 anyway.
>>750573372Then buy a new one in 2028 poor idiot?
>>750572185based retard
>>7505733726080 of whatever they will call it will be slower than 5090 for sure.6090 if it even exists will be $6090 MSRP.Those cards won't get released before 2028.
>>750573489Youd have a heart attack if you saw how much my stock portfolio is worth. That is just one of the perks of not wasting cash on effectively worthless and overpriced shit.
>>750562178How is it that you retards seem to be the only one this shit ever happens to then? It's always some faggot in these [Brand] war threads -- when, I, someone that just buys whatever is the best value seem to have entirely no issues with my AMD card. You're either lying, or fucking retarded and don't even know the most basic shit like DDU.
>>750572889>bro anyone with a part time job can afford it if they arent a complete retard with moneyWhat kind of part job would allow you to spend 10k on a graphics card?
>>750573932Meant for>>750562417
>>750573840Mine is $99999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999999 but of course I only got it by eating beans and living in a cave so can't change that now can I.
>>750533764>competent yellow menkek
>>750533307>AMD is coming to the rescueWe are doomed.
When these insane GPUs out I just imagine the comanpy owners ringing their hands in anticipation for retards with absolutely no need for the GPUs to buy them with their walmart paycheck. Why would any gamer need 36GB of VRAM when things are about to stagnate for years due to inflation?
>>750535572>elon begging for a spot on the island.txt
>>750573980I live rent free with my mom, eat light, don't drink or smoke, rarely go out with friends. Like I said, if you're not a retard then you should have enough savings for the things that you enjoy and spend time on the most.
>>750533307i'm buying italong with a next gen vr headset
>>750574178>rent free with my momFucking lol, and you think you can educate actual functioning adults, you should be saving money for when you inevitably have to move out then not buying worthless overpriced garbage you fucking retard.
>>750574178This is a dumb fucking argument. I don't care if you live off on bread, you shouldn't be spending your paychecks on a 10k graphics card and depriving yourself of everything else.
A.I. bubble says nope.
>>750574178>I live with my mom so I can afford to play vidya on an RTX 5090>t.
>>750574178You sound like one of those brown people that buy rolls royce to park outside their section 8 apartments if I'm being honest, just completely retarded.
>>750574178If you bought the 5090 on release, then it was a good investment. Sell it and you can move out of mommy's house.
>>750533307>36GB of vramNo one needs this much ffs, make it less and more affordable
>>750574178oh you got them mad with this one lol
>>750574295>when you inevitably have to move outWhy would you have to do that instead of just inheriting all the properties?
>>750574959Then buy low end card with 8GB.
>>750569163yeah it uses frame gen just like nvidia to get above 60fps, 9070xt is basically a 5070ti, you use danielblanc's port of it
I sold my 5070ti because the drivers just dont work for gaming compared to AMD on Linux or Windows 11/10.
>>750564886it runs like shit on nvidia cards who the fuck are you trying to kid 5090s struggle with this shit you lying faggot
>that'll be $3499.99 + tip
>>750575334RNDA4 version of DLSS5 runs significantly slower than nVs newer model.
>>750536548lol. It's an AI world not whether you want to accept it or not.
>>750575532can they focus on making AI do things that it's actually good at and not injecting their slop into every application purely to pump up usage metrics on a bar chart for some coked out product manager to jerk off to then
>>750567456>and even then you have all these driver issues and software problemsYeah, that hasn't been the case for a very long time now. I can't even remember when I encountered an issue with Adrenalin, and its user interface is comfy too.The only thing Nvidia has in terms of advantages is DLSS, but the most recent vibecoded one has questionable use anyway.In terms of graphical performance, there's basically no difference between the two in their respective price classes.
i have 2 AMD PCs and i've had issues with both no one fucking told me to build an nvidia machine instead
>>750533307Just for 10k
only winjeets use nvidia and intellinuxchads use AMD
>>750576319Same sir, I really should have done the needful and bought Nvidia...
AMD will never make a halo card better than Nvidia because of their business structure. AMD designs both CPUs and GPUs and ever since they spun off Global Foundries, they must buy silicon from a supplier. Silicon is bought by the mm2 (as wafers), and AMD has to figure out how many mm2 to use for GPUs and how many for CPUs. AMD knows their margins for their various products and the mm2 necessary for each (hint: CPUs are small, GPUs are fucking huge). From that, it's simple math to predict the $/mm2 yield for a particular product allocation. And what does the math say? It says AMD makes a lot of money allocating to CPUs and almost nothing to GPUs. IOW, there's no money in it for AMD to make GPUs at the scale that Nvidia does. They can make mid-tier GPUs because those are smaller, they can make APUs because those are integral to CPU dies, but they can't make the fuckoff huge GPUs Nvidia does because doing so would rob their CPU business, the business that actually makes money, of precious silicon mm2.
>>750576425The true chad move is to use AMD with Windows and Nvidia with linux
>>750576840you need gpus to exist to sell cpus, dumbassand amd does have a much better strategy than nvidia going forwardchipletsonce they figure out how to glue 16 CU complexes together, they can make cheap as shit glued together monsters
>>750575524its like 20fps lets not bullshit ourselves and they both use x6 frame gen it runs like shit on both
>>750576319Let me guess, Linux? I've had 0 issues with drivers for both Nvidia and AMD on Windows
>>750576979Been hearing this since RDNA2. To AMD's credit, they did do that for the Instinct cards, but for reasons, they never do it for the consumer cards.
>>750576979>my cousin gurpinder has much better strategy than anyone>once he figures out nuclear fusion he can make many energies!!!!!
>>750576319>OH NOOOOES THE PC CRASHED ONCE! BETTER BUY A JEETVIDIA FOR $50K!!
>>750577193https://www.tomshardware.com/pc-components/gpus/amd-patents-configurable-multi-chiplet-gpu-illustration-shows-three-diesthey're obviously planning to do this when they figure it outreminder that crossfire existed
>>750533307Without the kuda stack it will never matter.
>>750577341Genuinely what are these retards doing that causes specifically AMD products to "crash", it's starting to feel like a lie/shill that's been used as long as the internet has existed as I've used AMD GPUs for the last almost decade now and I think I've only had one driver timeout the entire time and it was on Diablo IV which runs like shit anyway so I don't even think I can blame AMD for that.
>>750577426Right well the 7900XTX was supposed to be triple die and that fell through, then RDNA3.5 came out and it was APUs only and then RDNA4 came out and it's mid-tier GPUs only and now we're hyping for RDNA5 or UDNA or whatever we're calling it now and forgive me if I'm just not hyped. Show me the goods and I'll happily buy, I have nothing against AMD but the last Radeon I had was the 9700 Pro and ever since then the goods just haven't been so good.
>>750578139they don't have the tech to stitch small CU dies together yetit's not about multi die, it's about completely decentralising everything into memory, i/o and shader diesrdna5 won't be made up of tiny shader dies but the die being usable in both radeons and instincts mean we'll actually get a flagship this time
>>750564017You'll need the latest hardware to run Windows 12, just by itself, the way that shit has been going. Seriously, Win11 lags when you right click on the desktop trying to load the options, it's insane. Opening the start menu also lags and spikes both the CPU and GPU. Hell just opening the goddamn image viewer takes forever, even on a strong computer, because of all the bullshit bloat and still lacks some basic functions. In the future, only $2000 Nvidia cards will be able to run Windows as intended. Everyone else will have to use a stripped down and intentionally gimped version.
>>750578510Power consumption is also fucked in Windows. Just doing nothing on the desktop has my fans spinning up and down. Literally nothing going on, just the screensaver.
still on RTX 2070 Super
>>750533307how much is it?
>>750535910>Nvidia card can do everything else and thats why they won and become richest company everESL pajeet
>>750533307>intel's not even 3 gens in yet>intel already beats them in features and driver stabilityhow is amd fumbling so hard
I feel like a god
>I feel like a god
>>750535413>Rothschilds>barely aware ofAre you retarded? You haven't stumbled across some secret nobody knows about. The very fact the Rothschilds are as memed as they are (and have been for literal centuries) should make you question your assumptions in light of your own dipshit post.
>>750533307>No DLSS>No CUDAAin't buying it.
>>750533307Imagine the amount of money AMD could be making if they knew how to code drivers
>>750539498>i can sell my 5090 at 10 times the price i bought beforeI lived that shit when I bought an RX580 right before the crypto craze. Why would you do that tho? you won't be able to replace that gpu at a reasonable price
>>750571243I'll post this when AMD fails to deliver, again.
>>750584736The actual 5090 chads will be the result. Poors with a 5090 who got lucky will just sell it to AI jeets
>everything I want works on my XTX including textgen>it also makes /v/ seetheNice. Might buy this new one if the price is right.
>>750541240I legit wonder why, at this point, gpus aren't just an external box you hook up via like, a usb4/lightning/whatever the fuck high speed connector that has its own dedicated power supply.
>>750580847Intel can't even run chat bots with normal quants. Shit's broke.
>>750579213Why would anyone buy a 5080 unless they were sending it to a chinklab to be retrofitted? Literally double a 7900 XTX and less function.
>>750582662You can run both with a simple driver mod. Indians could never
