>>106790566Brand loyalty and game devs in bed with that brand. The RTX brainwash is in full gear, never mind that it looks and performs like absolute shit.
>>106790590Is type of thinking is why AMD is loosing.Truth is AMD has lost everyone that was once sympathetic to it.You also can't win trying to be Nvidia minus $50-100 because people in this day and age will gladly pay a premium for things that may or may not matter.
>>106790566their strongest card is simply to weak
>>106790566I assume it's because the nVidia performance is just much better.
>>106790566>missed the gaming era>missed the mining era in the most part>totally missed the AI eranot surprising
gee I wonder why
>>106790843>missed the mining era???the most common gpu for mining was literally the rx 480/580 and it's one of the main reasons it was the most popular amd gpu ever.all the cheap 580's third worlders are running these days come from mining operations.
>>106790566AMD is very disadvantaged in the 3D rendering space. The 9070 XT has less than half the rendering performance in Blender compared to its Nvidia equivalent.
>AMD selling a lot of mobile ryzen>that graphX doubt
>>106790781This.When you're the underdog you need to offer a much better deal to regain market share.AMD is simply too greedy and only looks at short-term gains.
Edit:>DesktopI see, that's why. Meanwhile AMD dominates in consoles and have good APUs for mobile.
>>106790566they should focus on having more VRAM than ngreedia in low and mid price range to win gaymers
>>106790986Nah, they should mud the waters even more and add sub-frames to fake numbers like nvidia.>re-render 10% of the scene, extrapolates or reuses the rest, "here's your new frame sir"
>>106790566AMD needs to prioritize ROCm as much as Nvidia prioritizes CUDA. Every Nvidia GPU has full CUDA support on day one.
>>106790996you might be right, weekend players probably don't care about fake frames, only multiplayer tryhards do and they are the minority
Is this because AMD Andrelin is so retarded and much worse than my PS5 share play?
>>106790566Nvidia have the prebuilt market on lockdown, just try to think of the last time you saw a desktop or laptop on sale that didn't have a Nvidia GPU in it.And despite what /g/ thinks thats how most people buy their computers.
>>106790973>When you're the underdog you need to offer a much better deal to regain market share.I'm not really convinced that way of thinking works.Maybe in extremes.AMD would have to push alot to break the current consumer culture around PCs.It's not like It's impossible, people buy AMD CPUs but when people will gladly overspend on most components the current AMD GPU "value" proposition doesn't work.How do you convince people to spend $100 less on a GPU when those same consumers are spending hundreds more than they need to on incremental upgrades to PSUs. mobos, SSDs etc that will probably never tangibly change their experience?
>>106791111>just try to think of the last time you saw a desktop or laptop on sale that didn't have a Nvidia GPU in it.Framework primarily uses Radeon GPUs.
>>106791135ah yes, the wildly successful framework, competing with dell, hp, lenovo, msi...
AMD is owning the midrange and budget market, you guys are nuts.
>>106791343Owning?
>>106790566GeForce 8000 series introduced CUDA, and NVIDIA made WAY more money than Radeon during the GPU wars of 2008-2012, Radeon showed their hand too early with GCN and as such NVIDIA made a huge profit with Kepler. Then Lisa Su came around to cannibalize Radeon and they fell behind a decade.
>>106790566amd very good value saaar
AMD cards are as about as popular as nvidia when sold separately in most markets. However they have virtually no presence in prebuilds. Why? Because they suck at negotiating deals with other companies.
>>106790979>AMD dominates in consoles>>106791343>AMD is owning the midrange and budget market>>106791469>AMD cards are as about as popular as nvidia when sold separatelySeething, are you all jeets?
>>106791606AMD has the Linus Tech Tips AMD Tech Upgrade Nvidia doesn't even have one and in fact has never had one AMD is winning
>>106790590youre either a child or a boomer, either way clueless, no one is playing games anymore, no one gives a shit about RTX, amd lost cause it cant compete with CUDA in AI, when it comes to important stuff, amd is laughably behind
>>106791398What's with the generic AMD radeon graphics names? Is that all of the models joined together?
>>1067913986600 bros, we're all gonna make it.
>>106791851Probably intergrated graphics, APUs and stuff. Like in the Steam deck or Ryzen mini-pcs.
>>106791803AI is not important.
>>106791883according to you?
>>106791860>>106791398Talk me out of buying a 6600/xt today right now
>>106792795It's a pretty solid card, nice performance, low consumption. If you don't need gaytracing, it should do you well.
>>106790997It still pisses me off that I can't run SDXL on AMD integrated graphics or AMD AI chips when even a shitty $150 iPhone SE 2022 can.
>>106792795idk where do you live or if the used market in your country is decent but in my case (italy) the rx 7600 was only like 20-30€ more expensive while being 30% faster so it made no sense to go with a 6600.6650xt is much closer but can only be found used and retards were selling it for like only 20€ less than a new 7600 so yeah.
>>106791883OP is comparing market share and AI is very important for that atm. But at least now we know that you are just retarded.
>>106790566Mother fuckin' CUDA support
>pic
>>106790997ynr https://github.com/ROCm/ROCm/discussions/4276>which cards do you want us to support in ROCm!?all of them, morons
>>106790566RTX offers way more frames per watt and better AI frameslop on top, the market recognized the better product
>>106790566Delete your thread or nicotroon will spam-bake /pcbg/ 10 times in a row
>>106790843Retard-take, it always has been mindshare. Nvidia just won hard at this. It allowed to weather mishaps that would otherwise cause them to lose the lead.Nvidia successfully build a cult of personality. They read the manual from Apple their idol. Just look how similar the average macolyte and nvidiot are.
>>106791407CUDA didn't do jack then. It was because their mindshare was already rock-solid you dunce. CUDA only made returns nearly a decade later when general compute stuff started to go outside of pure academia. CUDA-fags make most of the Nu-nvidiots who are convinced that CUDA is some kind of ambrosia and shit-post on /g/ about it.
>>106790566marketing, it's always marketing
amd wants to sell cards that offer less than 30% of the functionality for 90% of the price
>>106797371people like you act as if it's just neural networks using it but pretty much EVERYTHING that uses the GPU uses cuda. emulators, browsers, video players, blender, you name it. the only thing that works reasonably with amd besides general purpose videogames is emulators and half of the time, for amd, it works on linux but has issues on windows
>>106797475You don't understand, everything else in the world having issues with AMD GPUs except consoles and some linux distributions is everyone else's fault and not AMD's at least that's what reddit tells me.
>>106797475Most of those were niches and exist only within academia before GPU mining took off. This is what jump start the demand for general compute crap. The ML/AI took the demand to a whole new level and Nvidia got most of the windfall from it. It had generate a generation of Nu-nvidiots who are convinced that CUDA is some kind of sacred cow is the only reason Nvidia got big. Both Nvidia and AMD/ATI look fine on both ecosystems. They each other own laundry list of "stupid, stupid" eyeroll issue that their fanboys pretend don't exist. Nvidiots are just much louder at this.>>106797531Nah, Nvidia has a ton of stupid, stupid non-sense on their ecosystem. Nvidiots are just louder and downplay it more.ie use products on both platforms for decadesThe funny part is the most of the "issues" are actually result of a CPU, memory and GPU overclocking going awry. Just because your CPU passes PRIME95 or Linpack for 1 hour doesn't mean it is 100% stable.
>>106791469>negotiatingMore like Nvidia strongly discourages OEMs from using AMD GPUs, using tactics to those that Intel used and got sued for when they were locking AMD out of OEMs in the 2000s.
>>106797663you are deep in denial. no one believes a thing you write. everyone knows all the programs use cuda (and work well) whereas the amd equivalent is a shitshow that rarely works. you're wasting your time here trying to make shit up, you're not even getting paid for this either. you should value your time more
>>106797796Funny part is that in the ATI-era, ATI rule the discrete OEM market. Nvidia was the underdog. Funny that changed around the ATI-AMD buyout.
>>106797835>Nu-nvidiot-fag getting call out this hard and getting this anally devastatedThank you for proving my point. Protip: none of these companies are your friends. Stop placing them sacred pedestals.
>>106790566AMD refuses to admit they are losing(or well, have lost at this point).They can't compete when it comes to the shitty marketing features that all the manchildren want like raytracing and AI, but still refuse to lower prices to offset it. While Nvidia is chasing the AI bubble and the super high-end market they could easily dominate the low/mid market if they just priced their shit accordingly, but I guess they are to prideful or just make to much money on consoles and shit to even care about the PC market.
>>106797932AMD already lost the war years ago. They are just shifting their attention elsewhere. Graphics are simply not sexy. Nvidia knows this as well. Nvidia doesn't give a crap about their graphics division. Just look how their treat their gayming market as of late. /v/tards are just desperate paypigs trapped with stockholm syndromeWhy do you think noise from nvidiots these days are all about "MAH CUDA" and "MAH AI/ML" and gayming stuff quietly took a backseat?
>>106791803me and all of my 3 friends play videogames daily , and we all have amd cpus thus im gonna upgrade my 2060 gpu to a 9070xt
>>106797957Yeah, that's kind of my point. Seems like it would be a pretty easy market to tap into for someone like AMD. I imagine there is a pretty hungry market for decently priced hardware focused on video games since that doesn't really exist anymore.Shame Intel's GPUs flopped. Would have been nice to have another player in the market. How come the chinks never made a push into GPUs? They are usually the price equalizers.
>>106790566Because 3dfx went belly up in 2000 and Nvidia got all their IP and most of their employees.
>>106797847Remember during the AGP era when the Radeon 9800 dominated and Nvidia played catch up the whole time? I had a 9800 Pro 128mb and it played everything under the sun in 2004
>>106797663>Nvidia has a ton of stupid, stupid non-sense on their ecosystem.And AMD's equivalent works better and has wider adoption? >>106797869>Stop placing them sacred pedestals.That seems to be your issue though, in fact your obsession with nvidia is really odd
Both brands are crap and have been for a while but thankfully no good demanding games have been released in a decade now so I can skip all these generations of cards comfortably, chang can have them for his deepsuck server or something
>>106790566Because the 9070 XT isn't 450 dollars, the 9060 XT isn't closer to the 9070 XT in performance for 250 dollars, and there is no 9080 XT that offers more vram than the 9070 XT. AMD thinks they can compete for the inflated midrange when they should take over the 200-500 dollar GPU market instead.
amd software is catastrophic.
has nicotroon ack xirself ?
>>106798002Problem is investors go "well, why are you focusing on consumers/gamers instead of the future, like AI"
AMD does not want to win back marketshare, that is 100% for sure.Otherwise they would undercut NVIDIA pricing.The MSRP of the RX9070 makes 0 sense.It should be 50 bucks below the RTX5070, then more people would buy it compared to the 5070.There are a lot of folks who bought the 5070.
>>106790965Barely anyone uses Blender.
>>106791398RTX 5070 ti has been a great upgrade from my RTX 4070.
>>106791803People don't really care about local AI. These cards are sold for gaming.
>>106799290It has gotten to the point if you really care about local AI, then you need to pay for cloud GPU time.
>>106790566NV has 10x the R&D budget.
>>106790997AMD's inability to keep up on the software front makes their hardware less valuable than Nvidia in spite of being more or less peer (and sometimes less expensive, though that varies depending on factors).
>>106790781AMD barely cares about gpus, it probably makes them a little money so they keep on working on them. Their real money maker is ryzen.
>>106799769I think AMD is aiming for paradigm shift with incredibly powerful APUs like strix halo.
>>106790566CUDA is just straight up better than anything else out there plus NVIDIA heavily invested in software and libraries using/supporting CUDA. For example suitesparse has CUDA support due to NVIDIA, which means that anything using linear algebra with gigantic sparse matrices (happens quite often) is going to use an NVDA GPU (or will use iterative methods but those suck if you require extreme accuracy). AMD never even bothered with functional drivers let alone fundamental library support, so they will slowly die out in anything other than gayming.
>>106790566AI slop economy, I imagine. AWS is further developing their own GPU line but in the meantime Nvidia was the fastest to jump on board to fuel the slop factories.
>>106790566It's because CUDA. Any other reply/reason is uninformed. Nvidia is winning because of CUDA full stop.
>>106799769They only care about datacentre gpus, huge market for them
what the hell is CUDA
>>106790566>why are they losing>They are constantly selling out of every card they makeThey aren't losing, they just don't have the same production volume.
>>106801186It's basically an API language to make the GPU do tasks that would traditionally run on the CPU. Legit it's probably one of the most important features of a GPU. the AMD version is called ROCm but unlike CUDA it doesn't work for all AMD cards, in fact it only works on a handful of them which just showcases how fucking braindead AMD is, because even the shittiest 750TI still has CUDA support albeit with a lower feature set since there have been improvements over the years. If I'm remembering right even the fucking RDNA 4 cards don't all support it and thats just insane.
>>106799559>care about local AI>pay for cloud GPU see how retarded you sound?
>>106790997they're working on it at least
>>106804698and how many months now has the 9070XT been out that its finally being worked on? how about the HDMI 2.1 situation on Linux that both Intel and Nvidia solved ages ago? All these years and still no actual solution other than "buy expensive HDMI to DP adapters and hope it works". What a god damned meme
>>106790566I like AMD, but the truth is that they're always a step behind (of both Nvidia and Intel) and their marketing team sucks ass too. At least their APUs are doing okay.
>>106790566AMD could reclaim entire market with cheap high VRAM GPUs 64Gb and over, and better support for AI gen scene.
>>106790566>manufacture 1 gpu for every 20 nvidia does>wtf why charts low???gee I dunno
>>106790566Nvidia Jew tricks entrapped themShit like AI upscaling and RT/PT came out of left field and blindsided Radeon pretty badlyAlso AMD just plain doesn't care that much about discrete GPUs. They make millions off of snoy who guarantees multiple tens of millions of orders of APUs over 5 years
>>106790566Software. Nvidia has CUDA man, they don't have competition.
>>106791803more people are playing videogames than at any other time in human history
>>106790566arent all nvidia gpus ending up at some datacenter?
>>106790566AMD are basically controlled opposition and even if they weren't, Nvidia pushes for a lot of anti-competitive shit.The CEOs of both companies are cousins and they're practically just middlemen for TSMC fabs, which is another company that they have close ties to.>>106790781>people in this day and age will pay a premiumPeople have always been willing to pay a premium on cheap tat so that they can LARP as richfags, exploiting groupthink has always been a thing, it's just gotten worse as marketing has become more omnipresent.If you make the card cheaper without removing the stigma of not being the "premium" option, then you change very little.
>>106790566Jeets vs Chinks
>>106790566>>why? they have superior software stack
i wish that amd gpu will finally die and their shitty hardware stop influence graphics api like they used to be.most of graphics api complexitities today are due to amd crappy hardware architectures, especially GCN which caused the whole vulkan/dx12 mess.
>>106801186compute shaders, but designed to be ergonomic and programmed in plain C. unlike garbage with have today in vulkan.
>>106790566>data by jprshipments does not equal sales when people will learn thisand you gonna see this onthe next Q since china has basicly buttfucked nvidia
>>106799271it is being used more and more in the industry. AMD perfs on blender are ridiculously bad. AMD once gave $100 000 to the blender devs but they used it to make a cuda renderer lmao.
>>106805413>they're practically just middlemen for TSMC fabs,since when tsmc is designing logic circuits?
>>106806281for example, cuda uses C memory model, with cudaMalloc(), cudaMemcpy() etc, very simple. now compare it to vulkan memory model, where you need to write like 2000 LOC just to upload some buffer.
>>106790997They need to ditch the middle men of ROCm/CUDA and just offer straight GCC to let people run their normal code on the GPU. https://gcc.gnu.org/onlinedocs/gcc-15.2.0/gcc/AMD-GCN-Options.html
>>106806466cuda is a real world example of why API ergonomics matters more than "explicitness" or "performance orientation".
>>106804757It looks like they've been working on it the entire time.
>>106806281What does Vulkan have to do with CUDA?
>>106806244>especially GCN which caused the whole vulkan/dx12 messAMD just pushed the deficiencies of their architecture on API users and called it "low level".
Wait it's just an APIwtf then how can AMD not have an equivalent that's just as good then
>>106806473is it good if like 90% of that page is listed as (Experimental)?
>>106806339this descrete gpus for nvidia means 99% datacenter gpus and 1% consumer level shitthis just doesnt say anything
Nvidia ships 11 million discrete desktop GPUs a quarter so nobody will buy themAMD ships 700,000 discrete desktop GPUs a quarter so nobody will buy themHaven't you seen the Mindfactory data? AMD outsells Nvidia. Mindfactory nearly went out of business this year but nevermind that.
>>106791343>budget marketIntel B580 is actually a decent budget option.I don't think any serious user opts AMD, despite the industry and e-content creators DESPERATELY wanting AMD to be more than controlled competition and paper launch volume products.An AMD GPU designed only to benchmark well is just never going to find purchase against NVIDIA (or Intel, assuming they keep playing the game) - NVIDIA/Intel sell products at real volume, and need real users to be happy.
>>106790566this graph doesn't count APUS but does count mobile sales. The split on desktop is 30 by the data used to make this graph.
>>106790566RTX has won by forcing you to use fake frames and "AI" upscaled resolutions with half assed gaytracing, powered by....... "AI."AMD was always lagging in performance, but the software lock-in is slaughtering what's left.
>>106804927they won't do this because that would make sense. kek
>>106797471Exactly. Everyone in this thread is retarded. Fact is the software both from AMD and third parties is shit.The only way they can recover market share is to produce a cheap AI card with a shit ton of VRAM and sell it for cheap, and sell their gaming cards with regular VRAM at break even prices.
>>106807830It's almost as if Jensen and Lisa were cousins and she doesn't wanna cannibalize the other family business.
>>106807880ya. I honestly don't understand how this blatantly anti-competitive bullshit is legal, desu.
>>106805413Nonsense. People are building datacenters out of 3090 and 4090 chips. GPUs nowadays are a B2B market.
>>106807923pretty sure consumer grade GPUs are licensed for non-commercial use
>>106807923Nvidia are shipping 3090s in 2025?
>>106791135framework is fake diy jeet trash. oh yea its totally repairable and upgradeable so long as you buy our parts to do so!!
>>106791398you niggers are so retarded lmao OMG look at this market data, that is the only thing that matters! economic performance is a direct 1:1 correlation tp graphics performance. you stupid faggots HAVE to be corporate marketing shills, there is no other explanation for being such a blatant dickheaded retard.
Because AMD is owned by ((them))
>>106805413>People have always been willing to pay a premium on cheap tat so that they can LARP as richfagsMaybe.I've always seen it as people are paying a premium because there is too much "just in case" mentalityLike people buying a platinum versus gold PSU because they don't "trust" the lower end option.Or people buying a higher end AIB GPU because they are afraid the MSRP one will be noisy or hot. This goes for basically everything incuding mobos, ssds, etcThis is distinct from a decade ago when most people would try to find the best value proposition and be rational about it, If you bought something higher end like a premium AIB model of a GPU it's because you got a MUCH better card for a $20 premium at the most. Hell I remember looking for a 1080 TI and the mid-range AIB option was CHEAPER than a reference blower style card.There has always been people just willing to blow money on whatever but the difference today is more "normal" consumers are falling for this trap of paying just that little bit more and AMD cannot compete by selling NVidia performance for $50-100 less.What's the deal AMD is proposing? Going against what everyone else has just to save $100 on a $1000+ PC?
>>106790566ATI has been losing to NVIDIA since ~2005. This is nothing new zoomer.
>>106790566It died with ATi
>>106790590Who the fuck cares about rtx? It's the first thing people disable so they actually get some fps.Meme setting
>"AH DOOD you gotta get RTX ON ray tracing is the future and you can DLSS and you can't even tell and you can frame gen and you can't even tell and you can reflex and you can't even tell! i would never TOUCH amd because FSR looks like SHIT">game like Borderlands 4 comes out>"WHAT THE FUCK! you incompetent developers can't even optimize a game correctly!?!?!? i had to turn on DLSS and frame gen and it looks like BLURRY SHIT i fucking HATE this shit. you need to fix your game so I never have to use DLSS ever for any reason!!!! and don't even get me started on ray tracing, it just makes the game run terribly and i can't even see the difference"
>>106810144Is that even possible in current games made with an engine like UE5?
>>106810192Average American truck owner. "I want a 4WD V8 crew cab but SO HELP ME GOD if I get bad gas mileage, road noise from my mud terrains, and I struggle to fit in the McDonalds drive thru." "You would never catch me in a fucking CAMRY like a poor faggot." "Of course I need a 5090, I'm a gamer and I want the best of the best!" "What do you mean it's loud, hot, power hungry, and I'm expected to use the upscaling features that I paid for and brag about? THAT'S NOT FAIR! I should get 4K 120fps without that FAKE AI CRAP!"
I have a 9070 XT and was scammed 400 dollars because I wanted a 9060 XT but couldn't waitfag long enough after I had already been using a used 1070 ti for years. However I really enjoy the 256 bits of graphics over 128...
>>106790566value is the only thing AMD has and even then they just barely undercut NVIDIA AMD gpus are worse performance per watt also the windows drivers were dogshit until recently
>>106790566AMD has far less proprietary support for many things. You gotta jank your way into it working if you get an AMD GPU.That being said if you're willing to jank it AMD is far more of a value offer with their lower cost, and matching or even higher vram offerings.
>>106790590Basically this. AMD should be eating Nvidia's lunch just because Nvidia's GPU pricing is insane but they have a stranglehold on memeTX and other meme Post Processing shit that AMD (try as they might) can't replicate 100%.People gotta stop caring about memeK and memeTX and demand better pricing on their GPUs.With that said, the only reason (right now) to choose AMD over Nvidia is because their Linux drivers absolutely stomp the shit out of Nvidia's. Propritary and "libre" coded ones.
>>106810883ThisMy new build was me jumping ship from Windows and while I have to fiddle with memory parameters and find the right tools that work for AI stuff the 9070 has literally 'just worked' for games and video editing.
Intel had a chance to crack the nut open from the workstation/virtualization GPU front while AMD works on gaming, but now Nvidia is on the way to absorbing them. Not a monopoly BTW, Jensen Huang is Trump's friend now. Nvidia can continue to keep a stranglehold on all GPU fronts, funded by an effectively infinite cash flows from their stocks and $30,000 datacenter GPUs sold to other mega corps with infinite cash flows from their stocks. Again, not a monopoly.
AMD simply doesn't have an answer to this
>>106811071nigga nobody wants frame hallucinations. it looks like trash. antialiasing is all that matters and my 9070 tears at 170 FPS. u probably don't have physx either
>>106811134Wrong
>>106810192that why people pay premium for nvidia, so that they don't have to cope like that later
>>106810883there is nothing wrong with nvidia linux drivers. when people say that nvidia "drivers" are bad what they usually mean is that some obscure tiling WM developed by trannies doesn't support nvidia hardware, because it doesn't come out of the box on linux.
>>106810144>Who the fuck cares about rtx?the only people who say this genuinely are running old GTX 1000 series cards or AMD users.
>>106811071>fsr 4 doesn't exist>afmf doesn't exist>ray tracing doesn't work on amd cardsretard
>>106811367all those suck
>>106791883>>106799290If people don't really care about AI, then why are billions being invested into new datacenters? That's all they ever build now, they don't even build offices anymore.
>>106811390>fsr 4 is pretty much on par with dlss 4>most people agree that afmf is actually better than the nvidia framegen slop>ray tracing performance is like only 20% worse compared to nvidia now and that only makes a decent difference in games where ray tracing is too heavy even for nvidia gpus anywaysit's your shitty rotten nvidiot brain that sucks, not amd gpus.
>>106811519no they're trash and I 100% know you don't even use any of that fucking garbage
>>106811496>If people don't really care about AI, then why are billions being invested into new datacentersFake as shit bubble that's going to leave the economy in ruins. If billions are being invested into it, then when are the returns gonna start flowing?
>>106790566The bankruptcy of 3Dfx damaged the GPU Market irreparably This is just the end result of the past 25 years
people want to make porn of their ex girlfriends and colleaguesyou can't do that on amd
>>106808063Maybe not anymore but people were taking the chips and putting them on custom boards with more memory.>>106807972Lmao tell that to the Chinese.>inn4 amerikek cope
>>106791851amdgpu drivers on Linux. Situation on Linux is totally inverted vs Windows wrt GPU performance. AMD gets a huge leg up from drivers and their shit works. nVidia doesn't unless it's CUDA dependent.
>>106807701Why are you even mentioning intel in this? AMD is right now outselling intel in cpus in both server and workstation chips and intel GPUs are not selling at all. AMD is incredibly bad at marketing. They have GPUs with the most consistent frametimes compared to nvidia equivalent but for cheaper but they literally never mentions that. The only reason for gamers needs to buy nvidia is that makes sense is if they want the XX90 class card because even local AI is now a solved problem and will work on Intel and AMD cards as well as on Nvidia.
>>106790590its AI you fucking nigger
>>106814076ThisMy only gripe has been the limited amount of AI tools that work well with ROCm.
>>106798002Nvidia's mindshare is too damn strong. Discrete gayming SKUs aren't sexy to investors and are quite frankly a dying market. It is only being held by old guard of /v/tards. iGPUs are taking over the masses. There's simply no big money involved in discrete GPUs to justify the R&D costs.Nvidia and AMD regard discrete gayming GPUs to be little more than a dumping group for professional-tier silicon yield rejects.
>>106798085Retard-take, both companies aren't your friend. Paypigs Nvidiots are just to blind to the simple fact. Just howl and scream like rabid fools on /g/ and /v/.
>>106798843They tried this strategy for a decade+ before giving up. Nvidia's mindshare is that strong. AMD previous attempts always resulted in the same result. AMD gets bled out from R&D costs for no real tangiable gain in marketsahre. Nvidia can easily undercut AMD. Do have any idea the insane margins Nvidia commands for their SKUs even shitty yields? The crypto-mining rush open up the floodgates with price discovery.
>>106801029more /g/tard takes who parrot the choir in their echo chamber Free marketing at its finest.
>>106806429kek, gaymers really do deserve the nvidia monopoly they're living under as fourth class citizens.
>>106804927Nope, that strategy will not work. You vastly underestimate the mindshare that Nvidia build over the decades. They have to practically throw themselves for nearly the same duration for any competitor to catch-up. Just look how well they weathered flops like 12VHPWR and Blackwell launch debacles. Intel's fortress in x86 CPU market didn't fall apart overnight. It took over decade of mishaps.
>>106811827Econmony is already in ruins. We are deep in the denial phase. Most of the AI bubble is fueled by funny money that makes Enron and Worldcom debacles from dot.com era look like street hustles by comparison.
>>106816023am of demoralize from Michigansk oblast.
>>106816010>Intel's fortress in x86 CPU market didn't fall apart overnight. It took over decade of mishaps.More like 26 years. People forget how locked down Intel had the motherboard and pre-built PC market when Athlon came out. They had exclusivity agreements all the tier 1 companies were afraid to break until things got really out of hand.
>>106811827Nah they're going to use those data centers to do forever war on their domestic population once the labor market shits.
>>106790590>game devs in bed with that brand.all game devs work console first so being in bed with AMD would make more sense
>>106816816That's not how dev partnerships work. nVidia has more money than God from Quadro and AI, devs are always cash starved -> nVidia pays devs for special relationship. AMD doesn't have to pay anyone. They're the base case which must be supported.
>>106799271Still not paying for any autodesk subscription garbage. Blender won. Go shill somewhere else.
>>106790566> Shows old data right before new GPUs> AMD is losingCouldn't you be even more biased? AMD is going up right now. Nvidia lost 2% global market share in Q3 2025.
>>106817189>2%That's barely a margin of error
>>106790566Anyway the reason is OEMs. That's really it. If you don't count OEM it's 80 for Nvidia and 20 for AMD, more or less.
>>106817247You're right but the chart shown could have been updated. It's not surprising AMD only maintained status quo, not much changed. Nvidia high numbers are mostly about OEMs anyway. AMD needs to surprise Nvidia to go up, otherwise it will be in the same spot.
>>106817122>They're the base case which must be supported.Then should they have the most support and sales by default
>>106790781>loosingQuite so, sar.
>>106817953That doesn't follow. Desktop Windows and AI slop are different market environments from consoles. There are no games that don't run on AMD, but when you're looking for extra shit to add on for PC, that's where nVidia and their promo funding steps in.
>>106799271Other than anyone making indie games you mean
>>106811241No, I haven’t checked in about a month but the 575 series nvidia drivers were absolutely terrible last time I tried them, maybe they updated it by now but I doubt it. It had stuttering with simple hd video files and streams and there was god awful stuttering in video games, it definitely has something to do with how it’s using the vram buffer. They really fucked up the consumer side of things with the 50 series. They were too busy selling overpriced products to ai farms.
>>106818402It's probably nothing to do with the drivers. You need to do a clean install. And probably increase the shader cache size in the driver settings to 100GB or unlimited because a lot of recent games have had shit shader design causing them to use up the cache like crazy.
>>106790566since the AI boom AMD has been allocating like 1/4 as many wafers for radeon as they used totheyve been making a stupid amount of money selling super computers lately, and radeon has been abandoned
>>106816549Datacenters are going to massive deadweights. Skeletons of "silicon belt" once the AI/ML bursts that mirrored the old heavy factory facilities that litter USA landsacpe.Datacenters consume energy and resources comparable to aforementioned facilities. The zoning laws have caught up yet.
>>106821772You sound as stupid as Paul Krugman with his comments about the internet in the 90s. Datacenters aren't going anywhere and if you fail to grasp why that's on you.
>>106821839Most of the new build are future deadweights. They are inflated by "funny money" that is driving the whole "ML/AI" bubble. The actual valve that automation of white collar labor stuff will bring will be much smaller and will not require most of the new construction. Part of the "ML/AI bubble" is fuel by Cold War 2 paranoia fearing that China will surpass NATO (namely USA).
>>106821839RetardSome of the biggest DCs being constructed right now is for AI, there are 5-10 year projects just for AIYou can't just pivot the millions of AI/ML specific machines to general compute/storage, even if you where willing to swap machines you'd end up with a massive utilization problem and it may not be sustainable to keep these DCs occupied without investor money.
>>106810858they are still shit to this day. they need to fire their whole driver division, it is so bad. I have been in email chains with them about issues and one driver update fixes it and the next fucks it up again. they are incompetent monkeys. t. retard who bought a 7900xtx
>>106790566software, software, software, as simple as.amd hardware is fine, software is so incredibly dogshit you can't really use the hardware.reminder that mi35x hardware was unusable from up to a ONE ENTIRE FUCKING YEAR AFTER RELEASE because the drivers were not working.I know for a fact that many companies got burned hard because of this, atleast two cut their losses and went nvidia/intel and banned anything amd for the foreseable future.good hardware is nothing without good software, for some reason amd does not want to become a trillion dollar company, they just have to shell out a few millions for one year for 10 white male engineers to get there, it's actually crazy why they aren't doing it.there is a reason why some swe are paid multiple millions a year at nvidia, not sure there one person who isn't exec at 3m/yr or more at amd
>>106790566Still buying AMD sorry.
>>106821953k snoy
>>106821901The hardware and software are also rapidly evolving that buildings and supporting hardware could become deadweights before it reaches full utilization. AI hardware manufacturers doesn't give a crap, because they are selling the shovels.
>>106821923>parroting memes that have not been true since early the 2000sSad>>106821944Nvidia has own closet full of software skeletons.t hat the cult of "MAH CUDA" pretends doesn't exist.
>>106790566I still think it has upside
>>106790566subpart hardware plus subpar software tends to do that to you.
>>106822022>Nvidia has own closet full of software skeletons.t hat the cult of "MAH CUDA" pretends doesn't exist.as it should be, write, release and drop fast is a sign of good engineering culture, there is a reason why globohomos rewrite many of their software every 3 to 5 years, the ability to start over fast and kill dead end/old projects is the difference between a team of swe and a team of retards with near-zero velocity.
>>106822022wish I was as delusional as you guys. lemme guess all the reports about constant artifacting and severe crashes (that take minutes to recover from or a hard reset) are reported on amd’s forum are all nvidia shills?
>>106790590>Brand loyaltyamd retards did not offer normal priceamd dumbassess thought errbody gon run2buy their superspec high fps cardsamd tardoids fked up and lost guaranteed sales>>106791123if amd offered reasonable prices for their gpus i would get them on releasethey did not
>>106822225Almost of them are from users who overclock their stuff like idiots (CPU, memory and GPU). Don't overclock if you care about stability and reliability. You are not going to notice marginal at best 5% performance gains, but sure hell you will notice the instability and crashes.
