>New performance capabilities, can render on one and upscale on the other to completely eliminate performance loss. In the future raytracing and physics will be offloaded to the secondary GPU>New gameplay opportunities, in modded Skyrim you can have the game running on one while having dynamic custom AI conversations running on the other>Can play a norm game on one GPU while generating high res videos on the other Essential purchase.
SLi hasn't been needed since the 8800sDo all the above on 1 chip.
>>722551987So double the price of the cards while also doubling the amount of cards you need to buy all for everything to still look and run like shit because devs aim for a minimum res and fps regardless of hardware. Devlishly jewish.
>>722552068One GPU means your 30fps interpolation will drop to 20fps due to the overhead then double to 40, with two it'll stay at 30 then double to 60-70 .
>>722552191For gaming the key is to thrift a shitty second hand card for the slave GPU as all its doing is upscaling the rendered image from the master GPU. 5070/9060+3060 seems to be pretty popular.
People have been trying to sell this 2 GPU thing for a while now and it's still barely even a thing and support for it is damn near non-existent. Call me when it actually starts becoming a thing and not when it's still "Wouldn't it be cool if.."
multigpu is a thing with laptops. the igpu is used for low power tasks like your browser. the actual gpu is used for you games. you can do same on desktop.
>>722552346Too bad you're stuck going over pcie unless you're on the same gen card, which means a performance hit versus a single card
>>722552346>demand for gpus doubles overnight>thinks cheap gpus will still be a thing
Isn't the 4090 the last GPU to support NVLink, and only in like 10 games?
>>722552724Buy a Trump x Intel GPU in 2026
>>722552468On desktops, you have to output different GPUs to different displays (or at least, different inputs on the same display). You can't just put a 5090 in a system with an iGPU and have the output on the motherboard switch to the 5090 when you play a game.
>>722552950Thats the old school method. Lossless scaling should work on everything.
>>722553064>this faggotSeriously, die of AIDS.
>>722553114Upscaling is being forced ON for new UE games, deal with it old man.
>>722553205>shills lossless scaling>tries to compare it to DLSS>this faggotSeriously, die of AIDS.
>>722551987we already did this with sli/crossfire and fucking nothing came of it because at the end of the day you were at the mercy of a good profile for the game and people thought "2 cards = 2x power" and that wasn't the case in the slightest. we also already did secondary cards for shit like physx and fucking nothing came of that.developers have a hard enough time making games work on one (1) gpu with a baseline spec from consoles and they can't even do that right. you make two gpus a standard and that will eventually become 4, 6, 18, 32, etc. it's as ridiculous as a $2000 gpu with 32gb of vram still need to upscale form 720p to hit 60 fps in a game.
>>722553274You Nvidia piggies are real sensitive, having a melty in your own shill thread.
>microstutters
>Megaeloquents
>>722553349The main problem is the bus. If there was a good link method so you could deliver on card performance between two cards, it'd be awesome. That SLI ribbon cable had 26 wires. There were also tests showing bus-level SLI (sans cable) was just as fast in most cases.>>722553390>bong uses bong slangNobody cares. Eat some beans on toast.
You ever wonder if watercooling could be done with cum as an additive? Would it be more effective?
Disregard the faggot OP, answer my retarded question instead: why can't we just slot in more VRAM like we do with regular RAM? I don't need a new fucking card, I just need more VRAM for my old one.
Are we going to return to the Voodoo and PhysX days and have a dedicated gaytracing/upscaling card?
>>722553761Some autists can solder more RAM. Manufacturers will never allow expansion slots because it will kill yearly upgrades.
>>722553761>there's no good standard for socketable VRAM modules (likely due to card orientation and blocking even more slots)>AI generation depends more on VRAM than GPU power, so companies like Nvidia don't want datacenters filling their machines with cheap gaming GPUs with memory added
just buy 2 gpus and you can play your games at native res again (select game engines only, does not include unreal engine)just buy 2 more gpus and you can install the npc dynanimc conversation packjust buy 2 more gpus and you can then install the dynamic physics packjust buy 2 more gpus and you can have path tracingjust buy 2 more gpus and you can play unreal engine games (select tier 2 titles only)just buy 2 more gpus and you can play all unreal engine games (select tier 1 titles only)
double your pricetag for 2 cards and double you energy consumption to up your performance from a 1 to a 1.3, with occasional perf hiccups because SLI was a half-baked Nvidia gimmick.
>>722551987You need three GPUS: One for videoOne for vidyaOne for your AI assistant
>>722553875For gaming, 24-32GB is more than enough. It doesn't really hurt the gaming market, since people upgrade for raw power rather than VRAM. It's an AI thing that keeps VRAM limited.With the Nvidia/Intel deal, we might start seeing unified memory for Nvidia gaming hardware, though.
>>722553761Same reason why phones don't have sd card slots or headphone jacks anymore. Jews.
>AIWho gives a shit?
>>722554018You can eliminate tranny discord voice actors from your Skyrim modlist.
>>722551987>framegen>dynamic sloppa conversationsIt's just a scam to make the consumer market fund AI chip development so Davos types can make us all redundant.These people will summon demons if they think it will allow them to live without the help of "excess carbon".
>>722554016SD cards had to go because they were slow, and people would buy these $5 "8TB" cards that were like 16GB that you overwrote constantly. Both led to complaints. >Why are my apps so slow?>Why do my files keep getting corrupted? It's Android's fault!Headphone jacks still exist on cheap phones. People with premium phones don't care because there's nothing someone who pays $2200 for a phone needs a headphone jack for.
>>722551987Are they retarded? Upscaling on another gpu will be latency hell.Maybe you can offload another map or something that isn't real time. But not something on your screen right now.
>>722551987You can't get dual gpu setups to work anymore because we've hit the end of what the ATX hardware standard can provide.
>>722554251how? i am ignorant
>>722554016it seemed to me that headphone jacks went away because wires became unfashionable and people wanted their glass candy bar phones to be slimmer and prettier for bragging rights.
>>722551987I have 7.
>>722554251>release ATX2>fully mount compatible with ATX>new power limitsProblem solved.
>>722554324>there are now 3 different ATX2s>zero compatability>takes 5 years to get any standards
>>722554297You can only have so much power on your motherboard without burning it.
>>722554251Consider the following: double ATX.
>>722554324>new power limitsis this really the answer? you can only pull so much from a wall socket and deliver it in a way that's still safe not only for your self/house but for the hardware itself.
Obviously we need multi-motherboard home datacenters with photonic intermediary layers.To run the new DEI Unreal slop at 43 fps
>>722554507The power from the wall socket is irrelevant, as the PSU handles converting that. It's the DC power being spit out of the PSU.
>>722554251After the HDD cage and 5.25" drives were gutted, there should be plenty of space.
the sad truth is we are again slowly being driven back to streaming games
>>722553031The output doesn't change nimrod what does the work changes. igpus can do "work" and passthrough the GPU.
Carbon nanotube motherboards when
Fatherboard???
>>722554605>this faggotSeriously, why not die of AIDS already?>>722554620They literally can't. The CPU's video output lines go directly to the motherboard's DP and HDMI ports. It can't be rerouted like in a laptop.
>>722551987>boils you alive in your roomnah
>>722554667
>>722554716Consider>>722553712
>>722551987> 2 GPUs so they can display the top surgery scars in 4k on the body type B they/them non binary demisexual otherkinGonna be honest with you, i'm not running to the store for that shit.
>>722554716>in your room>your room>your one room>because you're underage or a NEETAlso>having your PC anywhere near the room your keyboard, mouse, and monitor areImagine not using Thunderbolt in 2025.
>>722554821do you fucking move your pc to every room in the house retard
>>722554313Apple got rid of the jack to sell more expensive wireless buds. Now that the iphone doesn't have a jack using a wire is telling the world you don't have an iphone and retarded girls use iphones as a status check. None of it was organic it was just apple making the world worse however they could. Any "slimmer" or "waterproof" cope is pure shilling.
>>722554871I run a display port to my room and an HDMI to my living room TV
i thought sli doesnt even work these days
>>722554871>can't read"Your room" means your bedroom in your parents' house. I don't have "my room." I have "my house." Every room is "my room."
>>722554986>can't thinkRetarded ass nigger needs everything spelled out to a fucking tee because his low iq brain cant extrapolate the idea of "your room" being in reference to the room that the pc is currently residing in from the contextual clues of the original point.How would you feel if you didn't have breakfast this morning?
>>722555095How would you feel if you didn't live with your parents?Bet you can't visualize that, either.
>>722555170Yeah, exactly as I thought.
>>722555237>I thoughtNot likely.
>>722554530>>722554818Kill your GPU with Eastern slop instead.
>>722555372That looks halfway decent for a Sega Saturn game.
you don't need itjust run the game on lower settingsit's fine, you'll have the same experience
>>722554297Power requirements and heat generation. A single 5090 needs to draw more power from the PSU than a dual titan or even triple titan setup would have back in the day, and we're not even factoring in anything else like the CPU, RAM (yes that also draws a good chunk of power and it's only grown with each new generation), cooling, etc. Meanwhile the PSUs haven't changed all that much in the past 20 years, and look at how fucked up things became when those same power supplies suddenly need to introduce a new cable standard that they're not actually capable of supporting.We're not going to see another shakeup until hardware companies push for another ATX successor to solve these issues, which probably won't happen for decades because these niggers don't want to have to change their factory equipment to produce different shapes.
>>722553515NTA but don't bongs have higher iqs than yanks?Just interesting
>>722555510It's not the greatest, but for what is essentially TotK with graphics I think it's pretty impressive.
>>722555826Their cutesy kindergarten slang makes me think otherwise. Also, they haven't figured out the whole "cooking" thing, which cavemen mastered.
>>722551987>SLI 5090
>>722555705But I can have the same experience...with a higher framerate.
>>722551987I had two GPUs in 2011.Microstutter was always an issue, and VRAM was a bottleneck since each card had to have to same data loaded into memory.
>650w power supply>Currently using a RTX 3060 Ti>I'm considering a RTX 5070 Ti as a 2nd card for Black FridayI'd do it only to be able to play (5070 Ti) and Gen AI (3060 Ti) on dual screen monitor setup.Not sure if I can run both on that power supply tho. Can I do it?Worst case, I'll just swap to the new one.
>>722555905Seems like you don't understand the effects of the second industrial revolution
Are modern games even worth it? I don't think so.
>>722556363I'm running 9070XT on a 1000w and I feel like that's not even future proofed enough.
>>722556512Emulate an FPS locked 360 era game on one GPU then use lossless scaling to sloppa it to 60+
>>722556553Why not use Smooth Motion like a heterosexual?
>>722551987God I hope not SLI was so fucking stupid.Retards really used to buy 4 fucking 980's for 4 way SLI.>buying 4 fucking 980's in 2015 costs about the same as buying 1 5090 todayGod this fucking industry is in the toilet.
>>7225569424 way 5090 won't have the same bottleneck issues though.
>>722556639because then you can't do mixed brand dual gpu
>>722551987Would GTA6 need this?
>>722557634Yes, and a second CPU
This is a baseline gaming build.
>>722556194Kek
>>722552192But what if the game runs native 10fps?
bump
>>722558030cump
>>722557451Why would you ever do that?
>>722551987>talking about doubling tge expending on overpriced GPUs>first thing that comes to mind is modded SkyrimLike pottery
>>722558428To be fair there aren't that many examples of local run AI models in games right now.
>>722551987>SLi and Crossfire are coming backLike VR, this fucking trend will never die
>>722551987>so you render at 240p on your latest memetracing card>and you use another memetracing card to upscale it to 4k
>>722558556Explain what's wrong with this.
>>722556363>Can I do it?>5070ti max w at 300 3060ti max w at 200Sure; make sure you post pics of the melted hunk of plastic and metal that your desktop turns into
>>722558556>t. AMD user
>>722551987We're so back
>>722558296AMD for gaming and Nvidia for AI
>>722551987I had two 9800 GTX running in SLI and those bad boys ran every new game ok till like 2015. It was such a fucking bitch to set up though.Everything about my PC was fine normally but it took me a dozen hours of troubleshooting to finally figure out how to make the video output work with SLI enabled. For some reason I had to move my RAM to something insane like slots 1 and 4 in order to make the SLI mode output video. Never heard an explanation for this, no one can figure why the fuck RAM moving to some seemingly random and incorrect slots would fix my video issue that only existed when SLI was enabled.
It's all part of the racket to bring the price of gpus up, back in the day you could run 2x 460 and get better performance than a 480 for less money.Now that nvlink got so strong and the pcie5 bus is so fast and underutilized they can't link 2 gpus? Horseshit. You don't even have the memory mirroring problem of SLI anymore they just don't want 2x 5070s to outperform a 5090 for half the price.
>>722558951>AMD for gaming>when their flagship GPU this year can't compete with a 3 year old Nvidia GPUGood job.
>>722559287if you need more than 70+ class then your game is unoptimized.
>>722557848I have 4 GPUs and it works greatOne for Frame RatesOne for Resolution. One for GraphicsOne to supplement the other three. Takes 5 power supplies to run the whole thing.
So in theory could we have one 5090 run at native 4k then have a second upscale to 8K?
>>722559550Yes. And it is possible.
>>722559574Nice, now we finally have a use for those gayass TVs.
>>722559550In old games, sure. But then they wouldn't be supported anyways so
>>722559645Use memeless scaling
>>722559550Nope, you have to render it at 240p. Then you can upscale it to 4k with another GPU, then you can upscale it 8k with a third one because apparently people think it works like that.
>>722551987>Every game is gonna be on Unreal SlopThis is a very grim future for gaming.
>>722551987>can render on one and upscale on the other to completely eliminate performance lossanon, FSR4 at 4K in Performance on 9070XT takes just 1.4 ms
>>722551987Wait, what if>one gpu that renders everything at 240p>second gpu to upscale it to 480p>third gpu to upscale it to 720p>fourth gpu to upscale it to 1080p>fifth gpu to upscale it to 1440p>sixth...
>>722559889Not think. Know. That's how it works, techlet.
My fucking CPU AND Mobo have iGPUs in them yet there's no way to put it through my dedicated. It's bullshit. I want to use all 24gb of vram for rendering while playing some shit on the other hardware I paid for but noooo- I need a second monitor.
>>722559959that's slower than 0ms
>>722551987Two GPUs is a a nightmare for AI generation and gaming I'll just fallback on ram even if it's slower
>>722551987SLI and AMD Crossfire had too many problems that we could never find any solutions for, the fundamentals could never be fixed. Mainly the problem is CPU performance being problematic for gaming for a long time, multicore scaling was a problem. Also DX12 left optimization to studios who preferred not to bother with multi-GPU setups.Frametimes were all over the place as well. You'd need twice the FPS to get similar frame latency to a single GPU running the same content, for example.
>>722560097>>722560264That's not the point refer to: >>722559964
>>722559321Shouldn't you spell it "unoptimised," since you're poor?Also, "needing" depends on settings.
>>722551987>literally no one on earth writes the micro code for the bridge>which is why everyone abandoned this steaming turd that was sli>GUYS BUY MORE GPUS GUYSRabbi please. Be happy that the US state still throws money at AI scam start ups like perplexity and drooling retards keep buying subscriptions for it. Bubble will burst within 1 year
>>722551987people are using one to do that frame upscaling frs stuff and the other to play the game?
>>722560409provide one game that needs a 5080 for recommended that is well optimized.
BUY AN INTEL ARC 770ITS 16GB VRAM AND COSTS UNDER 200 USD
>>722551987Yeah how will that make forged alliance more fun to play?
>>722560495>no one on earth writes the micro code for the bridgeNvidia did double duty pushing huge patches for SLI support and they still got shit on for it, it was an expensive PR disaster.
Cant believe there are goycattle that unironically buy new nvidia cards in 2025do you love the jews THAT much?
ITT: OP tries to justify his poor purchase/getting ripped off again
>>722560585Will it be useful for generating slop in slot 2?
>>722556220any frames over 70 is a meme desu
>>722551987im not holding your AI bag
>>722554185id have thought people who pay 2200 would have high end taste, like prefer high quality audiophile headphones, which are all wired. nobody with any money or knows about sound buys bluetooth headphones
>>722560721yes
>>722551987Naw. We need 1 GPU to cost as much as 4 GPUs currently do so the entire enthusiast and professional market can collapse into an endlessly-expanding gaping void leaving the only viable consumer option as APU-driven devices.
>>722560864nice, give link. I'm only seeing listings above 300$
>>722560818But listening to music on your phone isn't audiophile territory. It's "I want to enjoy music while I'm at the grocery store" territory. Cords are annoying, have multiple points of failure, and provide no real benefit in that situation.
>>722552424its zoomers who were too young to experience the SLI meme. the jews are back to trick a new wave of kids like OP
>>722551987>>722553515I had this with two 980s back in the day and the only time I ever noticed any difference was on a heavily modded skyrim game. It actually did help by a good margin, gained like 20+fps. Everything else it was unnoticeable.
If not multi-GPU, what are the other slots meant for?
>>722551987ITT: OP had a retard attack and accidentally clicked '2' on the amount section of newegg while buying his new GPU then made this cope thread
>>722561017People used to put sound cards in them. nowadays everyone uses onboard audio or an external DAC if you have autism.Other main use is expanding the number of ports. Shit like this.
>mfw I'm just playing games from 1995-2015Can't name a single game from the last decade actually worth playing. Maybe Dragon's Dogma II, but that's a very personal exception, and certainly doesn't turn your computer into a bomb.
>>722561017>multi pcie slot sorry bro you fell for the asrock jewish tricks to make you pay 400 usd for a 99 usd motherboard but painted black and a useless 2nd slot for le gamers
>>722561249I have three slots...
But guys you are missing the important point here hello? You can literally daisy chain GPUs, to upscale from 240p->480p->720p->1080p->...->...By using a separate GPU for every layer.
>>722561751At that point you should just render a single pixel and have the GPU guess what it's supposed to represent every frame.
>>722551987I'm not going back to SLI after last time. More headaches than it's worth honestly. Having to track down the right SLI bits with Nvidia profile inspector for every game because the drivers never did it correctly was a pain in the ass. Also didn't Nvidia actually officially kill off higher bandwidth bridges? therefore it's actually dead and never coming back in a way that would matter. What the fuck are you talking about? also can be prone to stuttering and hitching. Trust me dual GPU ain't ever worth it.
>>722561978We have already seen how upscaling from 240p to 4k produces no visible artifacts. AI will solve gaming, 1 pixel to 4K is the future.
Also one experience I can share of my many years of SLI. If you value game stability then SLI is not for you. It's not uncommon for game crashes to happen. It is way more prone to crashing games because the way SLI works is very sensitive and can be error prone.
>>722562209But we are not using it for rendering retard. Rendering happens on one GPU at 240p, the GPUs just take the frame and upscale it, pass it to the next GPU for another upscale pass.
>>722562251Does it still need to connected by a bridge? is that how the cards are communicating? Nvidia can market whatever way they want and I can see the inevitable scenario of where fucks up anyway.
>>722562095We can go even further though. Why start with 1 pixel? If we don't render any raster graphics at all it's an infinite performance gain. Have the entire game consist of prerendered pictures generated by AI and automatically arranged to produce a seamless environment. This is real by the way, look up 3D gaussian splatting.
>>722562345I had a duall 8800GT setup, and the PCIe bus was fast enough at the time that you could go without the bridge for about the same results.
>>722553875>kill yearly upgradesI hold on to my PC 8 years minimum. Typically, I'm not "upgrading" until the thing just dies. There's just no need for that anymore. Games requiring stronger cards are always inferior products. They are not even games. Not even close. They are tech demos and propaganda ware.I'm not buying "le new pc parts" for that lmaoo
>>722562351AI is not there yet.
Didn't you all feel like the ultimate peasants whenever SLI was not supported in a driver update for a specific game profile? like the ultimate sucker? I can't believe they're going to get away with this again. Fucking Nvidia marketers should all be shot for trying this shit again in regards to trying to make two gpu sales. And that's a point I want to make about all of this. If they're really going to get two cards working for games again, how well do you think it will be supported by actual game developers? you have to remember that it's developers who have to think to include this in a driver profile for their game. And the reason why SLI support in the past was so shocking because developers saw SLI users as rightfully niche. Therefore they never supported it in game drivers. Hence why you were forced into Nvidia profile inspector fuckery. On a seperate note note: in which some of those profiles completely fucked up shadows in some games when people found a 'fix'. I understand what you're talking about is different from the old ways of SLI, but you fuckers are still gullible and if I could find out who's suggesting we go down the route again I'd beat them to a bloody pulp just for trying increase Nvidia profits rather than making sure people were getting a good game experience.
>>722551987spending 5k on a pc setup isnt a replacement for a personality
>>722563017It's a different world now. It's not about performance. Literally all we care about is VRAM, and two cards = twice the VRAM. Sure it's not accessible the same way, but there's no other fucking option to just plug in more VRAM, so it does somehow have a usecase.
>>722563238And you don't even need VRAM. Literally use a shitty GPU to render everything at 240p, and use another GPU with more VRAM to upscale it.
>>722563306doesnt that give a bunch on input lag?so you play the game in minimum settings but use the 2nd card to turn it into 1080p?
>>722563571Yes, and you then upscale it 2160p by passing it through a third card. The amount of time you save on rendering a 240p render target exceeds the time spent on scaling, i.e, latency is not an issue.
>>722551987Everyone that pays for the latest GPU in this day and age JUST to play video games or asking some basic questions to AI shit is a sucker.
Jew nose practically poking through my screen with that post.
>>722563238It always sounds better on paper than the reality of how well it functions. And I don't like that you're potentially selling snake oil again. I see it as scummy marketer tactics. People should just buy the best single card they can afford and be done with it. I disagree with you or OP whoever said it's 'essential'.I hope my post serves as a warning to younger anons who don't know about Nvidia's shady history as a company when it came to dual GPU setups. It's not the 'past' when the idea you're suggesting is still at risk of the same flaws of the old model. I could get a whole lot more educational in this thread about a life's experience when dealing with this company for anyone who is interested.
>>722564078>I hope my post serves as a warning to younger anons who don't know about Nvidia's shady history as a company when it came to dual GPU setupsYour post won't be listened on this board because /v/ is shill central. Same board that defended the shitty performance of so many new games by telling you to enable DLSS (which still isn't enough, by the way).
>>722551987THE MORE YOU BUY THE MORE YOU SAVE THANK YOU JACKET MAN
>>722564078>Nvidia's shady history as a company when it came to dual GPU setupsElaborate
>>722564078I'm not OP and think it'd be retarded to buy dual GPU for games, I meant more for AI applications that simply need more VRAM and nothing else. But yeah it's scummy tactics all the way down. For that exact reason they artificially restrict VRAM on every card, scummy tactic #1.#2 is rolling software releases based on hardware despite no functional differences in architecture. Oops you bought a $999 GPU last year? Sorry that doesn't support DLSS 17 we just made lol.Then there's the scummy pricing structures. The intentionally confusing tiers. Inferior editions timed for maximum FOMO. Ratfucking of OEMs, which made EVGA drop out of the GPU business entirely. I could go on. One of these days I bet we'll get another 3.5GB fiasco.Yet at the end of the day, they're the only game in town. AMD is a joke on the software end and intel's response doesn't exist yet. It can't be helped.
Can I play Borderlands 4 if I have two 5090s in SLI configuration?
>>722564290AMD and NVIDIA is owned by the same people anon. It's just a tactic to avoid being declared as a monopoly.
>>722551987console gamer imagining how technology works thread
>>722561017More sata ports.
>>722553515>bong slangWhat the fuck, that isn't "bong slang"
>>722564078You're missing the whole point of this thread. This is not Nvidia shilling, this is people shilling lossless scaling. It's a 5 dollar program that lets you use any two gpus, one for scaling one for raw rendering, Intel amd Nvidia doesn't matter. It works like they say, but looks pretty fucking horrendous and laggy. People without don't know any different though, so they like it and shill it.
this is unironically kino.get your dusty old 1080ti or 5700x out of the closet, set your main card to run a high end game at 480p at 200fps, use the 2nd one to upscale to 4k and you have a faster pc than the guy who bought a 5090 on launch
>>722564737i mean use the dusty old card as the 480p one and your main card as the upscaler, my bad
I will continue to use my one 3090 to play games at 1080p 240hz.
>>722564595>melty welty jubbly wubbly bongy wongy talky walky slangy wangyIt sounds like either a 4 year old or a retard came up with it. It's bong slang.
>>722564737That is what I am saying, just render everything at 8x8 pixel resolution, then upscale it to 16x16 using another GPU, then 32x32 with a third GPU, and you keep daisy chaining them until you reach 4K.
>>722564779hz over 100 doesnt do anything you cant even see the framesyoud be better doing a higher resolution with less frames
>>722564802This wouldn't work because you'd need too many power adapters for one PSU. Even if you had the wattage there's not enough cables.
>>722564779Me but with my 4090, I am fine for the next 10 years. 1080p 60Hz
>>722564837Not if all they are doing is upscaling.
>>722551987The skyrim idea is nice but the ai gen takes a toll on your other ressources as well like processor and ram so it doesn't workIf you want something to process ai conv it'd be better to get an alternate server / pc and use it for that
>>722564779Same. I like how we all still use the same resolution+framerate since 2005 and roughly the same graphics from 2015 but everything is more expensive and runs worse somehow.
>>722564907Yep, and what's funny is you can boot up MGS3 on a PCSX2 emulator, render it at higher resolution, and see the custom lighting, shadows, atmosphere, etc that looks better modern games.
>>722564976youre not even gonna be alive in 10 years
>>722565109but the brand Nvidia is gonna survive and it ain't getting my money.
ITT: Gullible idiots or deceitful marketing shills?Is there even a difference?
>>722565225Gullible idiots do it for free.
>>722565225this doesnt cost you any more money, unless you sold your old GPUjust put the card you replaced with your current one back in alongside it and use it as a scaler
>>722565223Not that there will be an alternative with AMD soon to be named a FTO by Trump.
>>722565389I'm not gonna add my old RX 580 and redo my cable management just for this meme scaling retardation.
>>722551987I will not consume the product
>>722565539I will still buy NVIDIA products, just not in the must consoom way. Like anon said: >>722565646I had a 1080ti, jumped straight to 4080 when the 50XX scam was announced. And I looking good for the next 10 years.
>>722565554>4k 200fps for free isnt worth plugging in a new cableyou can take a horse to water but you cant make it drink i guess
Lol, this just sounds to me like "SPEND MORE MONEY!".Don't forget to also grab a second PSU, and two additional physical CPUs.
>>722565835>4K 200fpsYou kikes said the same thing about anyone not buying the newest jewvidia card for DLSS and look where that got us. Eat shit.
>>722565835Post a video of you getting 200fps 4k in any of the newest games, I wanna see how fucking ugly it looks rendering 240p and upscaling it to 4k.
This thread is retarded
>>722564779Guys, I think this is the year I finally go from 144hz 1080p to 144hz 1440p
if you have an amd CPU you can already do this without needing a second cardhttps://www.youtube.com/watch?v=RCw5Fj3QApY
>>722566840Why do you keep shilling this bullshit?
>>722566840doesnt have to be amd, just any cpu with integrated graphics, use your real card as the upscaler
>>722566897thats my first post ITT retard-kun
>>722566950>since the last time I shilled it
>>722566897t. nvidia employeeno i will not buy your 2000 usd GPU yes i will play at 4k settings on my 6600xyes you will seethe
>>722567014>yes I am foreignI'm sorry to hear that.
>>722556194A couple years ago a guy from work bought a 2kW PSU for his new rig. I called him silly.
>>722551987Upscalers are a meme they where a meme in the emulation and console retro scenes and their still memes
>>722567000>shillits literally free. only goys would hate it because they wasted tons of money for nothing
>>722551987>You need two GPUsNo, I do not. Fuck of nVidia.
>>722567102i got a 1000w psu like 15 years ago for cheap. still works a treat. i guess the fact it never runs above 30 percent makes it last longer>>722567075>280hzthats pointless. its clear that monitors have reached a point where they just have to trick people to sell new ones. theres 9001 blind tests online showing nobody can tell the difference between 120 and 240 nevermind 280
>>722567128>I'm not a goyIt's 100% mask off on 4chan these days.
>>722567075Still wary of OLED for my monitor, feel like I'm just asking for burn-in with the way I leave it on the desktop and have it on for like 9 hours a day.Plus I don't feel like anything above 144mhz is really doing much, would rather try and go higher res.
>>722569910https://www.rtings.com/tv/tests/longevity-burn-in-test-updates-and-resultsDon't listen to the OLED obsessed faggot in these threads. ALL OLED HAS BURN IN. All of them. Even the most modern.
>>722570430>needing to use a screensaver>having to blank my screen instead of being able to glance at it from across the room to see if anything has changedI have a bathroom that adjoins my bedroom with my office, and I like to leave the door open so I can see if anything updates on my monitor while I'm taking a long, relaxing shit.
>>722569910why even buy a monitor in 2025just get a 75 inch 120hz gigachad tv for like 500 usd
>>722570947Ulcerative colitis.
>>722551987>bought rx 9070>all I've been playing is Dusk, Megabonk, Brotato and other indie shitNah, I'm good. AAA games are trash.
>>722571312If I had that tatercard, I'd play Indian trash, too.
>>722556427Its not really the second industrial revolution its just in 90s and 00s Bongs decided to become Chavs who decided to say fuck society and do whatever they wanted to do. Including having lack of manners or family in the future. Chavs were a mistake.
>not having a GPU farm in 2020+5What are you, poor lmao
>>722570528>>722570947Oled nigger you will never win. We've already moved to miniled you fucking loser faggot. Talk about backing the wrong horse you stupid bitch.
>>722551987schizo-tier OP, this will simply not happen. Its way too inefficient, and if Njudea made this, AMD+intel would simply make an alternative that would have all the required components on the same card, for example. Many motherboards aren't even setup for a dual card system anymore, the second pcie-slots aren't even accepting a gpu now in many mobos.
>>722552192You need a minimum of 60 to start with for FG to not look and feel like ass.
>>722552468every person i know who ever tried a dual-gpu setup, whether on a laptop or pc, hated it because of the micro-stutter. It was a failure that will never be repeated. OP is a coping nostalgia-retard.Now if you'll excuse me, im gonna do a few more runs of Megabonk, a game you can play on the switch just fine.
>>722561185i still use the soundcard i bought 15 years or so ago. built-in dolby headphones sounds good
>>722572157You're talking about SLI. This is Hybrid Graphics. Different things entirely.
>>722572108this is dumb and gay. How are you supposed to have an AI card and a regular card then on same motherboard?
>>722572216you're supposed to buy expensive workstation grade cpus and mobos there's lots of AM5 mobos there where you can have two slots x8 lanes each, x8 is fine for AI generally, even games as long as it's not some old ass PCIe version
>>722554185>and people would buy these $5 "8TB" cards that were like 16GBThat just sounds like they needed an excuse, and found one.
the dual gpu era is upon usreminds me of those chads who bought dual xeon cpus for 100 bux that outperformed intels slop of the range stuff back in the day
>>722572358Like I said, you can still get Android phones with SD card slots, just not the premium phones.
>>722570309yep oled is trash, LMCL could've been something but greenies denied us this (it's kinda like a dual layer LCD, you can get OLED-tier contrast but the backlight needs to be powerful which causes increased power usage => greenies say no no you can't sell this to consumers, real smart and green because now everyone has to buy disposable display tech)
>>722572353Sounds like a market someone could exploit for gaming and AI enthusiasts.
>>722572636What do you mean? AMD and Intel are already exploiting anyone who wants lots of PCIe lanes by lane-starving consumer grade CPUs/sockets. Buy a sTR5 mobo + Threadripper and you shall have all the lanes you need. AI enthusiats on a budget tend to go for second hand workstation equipment. If you want both games and AI, best bet is probably AM5 motherboard which allows x8/x8 bifurcation for slots (and has slot spacing which allows two cards to fit in and cooled properly). Many already exist.
Also one 24GB card is already enough for some AI enthusiasm and you can use it for games as well.
>>722570430>this technology is superior, you just need to do something that hasn't been a standard thing for nearly 3 decades>also, it costs fucking twice as much for a decent 32 inchYeah, I'll be waiting for mini-LED. IPS is good enough for now
>>722560721Ahh yes, you mean will it do something you'll never use? Yes!
>>722567496>uses $ at the end>accuses others of being poorPost fucking hands right now
>>722572902No, I mean as in offering that option on the cheaper end, if nobody else is offering the option at that price point, people who want a budget option will go for that.
>>722573154nta and esl but how do you guys really say>i have dollars 500??
>>722573941no, but the symbol always comes before the amount of the thing. its just how english works
all that stuff is old hatwelcome to the futurehttps://www.youtube.com/watch?v=gSzgeyuyblM
>>722569910I think 170mhz is the minimum because it allows you to hit the magic number 85fps with black frame interpolation. 85hz is where flicker and related eye strain starts to disappear for 98%+ of people. A 144hz monitor is only gonna hit 77fps.
>buy 2 gps to play this unoptimized console port of a game you'll play once and never touch
>>722571082Monitors have better latency by design because they are designed for interaction. TV's are just meant to be gawked at so they tend to have tons of input latency and smear that is shit for vidya.
>>722573164But who would do it and how? AMD and Intel decide the amount of PCIe lanes in socket designs and CPUs. Motherboard manufacturers can't do anything there. In theory more of the PCIe 5.0 lanes going out from the CPU could be converted to PCIe 4.0 lanes (right now only a few are used for this) but bandwidth stays the same and you win nothing except if you wanted to install lots of OLD cards specifically.
>>722574529doesn't make sense to me but oki'm spanish we just write it like we say it
>>722551987>he doesn't have a 12 GPU frame miner rigDaily reminder to stop being poor
>>722554681it works with AMD cards/chips only
Also since AMD can apparently force all X870(E) motherboard vendors to dedicate some lanes for USB4 (this is the reason you can have a full speed x16 slot for graphics card + two NVMes directly connected to CPU on eg. Asus ProArt X670E but can't on ProArt X870E, second NVMe drops the top slot x8 and nothing can be done about it, there are other motherboards which suffered the same fate going to X870E vs. the older X670E design) I'm sure they can also force them to not sell something with too many old-gen PCIe lanes (assuming it's a market worth targeting anyway) if they think it's going to eat into sTR5 sales
if i want to do this but have a x16 pcie16 slot and a x4 pcie16 slot, which card goes where, do i put my shit card in the x4 2nd slot or the main slot
No, you're not being ambitious enough.What's really needed is the ability to combine computation of multiple computers, so you buy 4 Asus ROG desktops, plug them over ethernet and they combine into one big computer.You could use it for real work like AI training or protein folding but you could also use it to play DCS with ray tracing and 4K graphics, at a stable 240 FPS.
>>722576528I'd put shit card in x4 slot
>>722575174wtf
I would buy a second GPU if SLI/Crossfire came back
>>722578928Yep. It's been doable for six or seven years now.
Finna buy 2 5090 so i can play games with dlsss ON as it was meant to be
>>722551987As someone who is into CRTs I can already tell you that this isn't feasible because the bandwidth on PCIE lane isn't high enough.The future will actually be rendering on one GPU and firing it out of your ethernet into another dedicated box for upscaling.
>>722579198>>722576772
>>722579249Sorry I didn't read the whole thread and am surprised someone else mentioned it. I was thinking of a setup similar to Groovy Mister, but probably it will be something we see people actually doing in the future.
>>722557848house fire build
you wont have to do this because chinese gpus are going to give us 5090 performance for 200 dollary doos within a couple yearshttps://www.youtube.com/watch?v=Y0wFuOB0aXw
>>722551987SLI was always a stupid trend and shame on the game magazines that pushed this shit to consumers.
