Well I am shocked, SHOCKED I say! Well, not that shocked.
Don’t think I’ll be moving on from my 7900XTX for a long while. Quite pleased with it.
Rimworld doesn’t need a new gpu
What it needs is a new multi-threaded engine so I can actually use all these extra cores. XD
Sounds like version 1.6 is supposed to get multithreading.
Can’t wait
i wish i could remember how i got it to run on apple silicon last time because i can’t do it now
I think the Steam Deck can offer some perspective. If you look at the top games on SD it’s like Baldurs Gate, Elden Ring, Cyberpunk, etc., all games that run REALLY poorly. Gamers don’t care that much.
Paying Bills Takes Priority Over Chasing NVIDIA’s RTX 5090
Yeah no shit, what a weird fucking take
But why spend to ““eat food”” when you can have RAYTRACING!!!2
I’ve been waiting for a product that makes sense.
I’m still waiting. I can keep waiting
All I want is more VRAM, it can already play all the games I want.
But with our new system we can make up 10x as many fake frames to cram between your real ones, giving you 2500 FPS! Isn’t that awesome???
Bullshitted pixels per second seem to be the new currency.
It may look smooth in videos, but 30fps upframed(?) to 120fps will still feel like a 30fps game.
Modern TVs do the same shit, and it both looks and feels like ass. And not good ass.
I don’t mean to embarrass you, but you were also supposed to say “AI!”
Points with a finger and laughs
Look at that loser not using AI
In the US, a new RTX 5090 currently costs $2899 at NewEgg, and has a max power draw of 575 watts.
(Lowest price I can find)
… That is a GPU, with roughly the cost and power usage of an entire, quite high end, gaming PC from 5 years ago… or even just a reasonably high end PC from right now.
…
The entire move to the realtime raytracing paradigm, which has enabled AAA game devs to get very sloppy with development by not really bothering to optimize any lighting, nor textures… which has necessitated the invention of intelligent temporal frame upscaling, and frame generation… the whole, originally advertised point of this all was to make hi fidelity 4k gaming an affordable reality.
This reality is a farce.
…
Meanwhile, if you jump down to 1440p, well, I’ve got a future build plan sitting in a NewEgg wishlist right now.
RX 9070 + Minisforum BD795i SE (mobo + non removeable, high end AMD laptop CPU with performance comparable to a 9900X) … so far my pretax total for the whole build is under $1500, and, while I need to double and triple check this, I think the math on the power draw works out to a 650 Watt power supply being all you’d need… potentially with enough room to also add in some extra internal HDD storage drives, ie, you’ve got leftover wattage headroom.
If you want to go a bit over the $1500 mark, you could fit this all in a console sized ITX case.
That is almost half the cost as the RTX 5090 alone, and will get you over 90fps in almost all modern games, with ultra settings at 1440p, though you will have to futz around with intelligent upscaling and frame gen if you want realtime raytracing as well with similar framerates, and realistically, probably wait another quarter or two for AMD driver support and FSR 4 to become a bit more mature and properly implemented in said games.
Or you could swap out for an Nvidia card, but seeing as I’m making a linux gaming pc, you know, for the performance boost from not running Windows, AMD mesa drivers are where you wanna be.
Saved up for a couple of years and built the best (consumer grade) non nvidia PC I could, 9070XT, 9950X3D, 64gig of RAM. Pretty much top end everything that isn’t Nvidia or just spamming redundant RAM for no reason. The whole thing still costs less than a single RTX 5090 and on average draws less power too.
Yep, thats gonna be significantly more powerful than my planned build… and likely somewhere between 500 to 1000 more expensive… but yep, that is how absurd this is, that all of that is still less expensive than a 5090 RTX.
I’m guessing you could get all of that to work with a 750 W PSU, 850 W if you also want to have a bunch of storage drives or a lot of cooling, but yeah, you’d only need that full wattage for running raytracing in 4k.
Does that sound about right?
Eitherway… yeah… imagine an alternate timeline where marketing and industry direction isn’t bullshit, where people actually admit things like:
Consoles cannot really do what they claim to do at 4K… at actual 4K.
They use checkerboard upscaling, so basically they’re actually running at 2K and scaling up, and its actually less than 2K in demanding raytraced games, because they’re actually using FSR or DLSS as well, oh and the base graphics settings are a mix of what PC gamers would call medium and high, but they don’t show console gamers real graphics settings menus, so they don’t know that.
Maybe, maybe we could have tried to focus on just perfecting frame per watt and frame per $ efficiency at 2K instead of baffling us with marketing bs and claiming we can just leapfrog to 4K, and more recently, telling people 8K displays make any goddamned sense at all, when in 95% of home setup situations, of any kind, they have no physically possible perceptible gains.
1000W PSU for theoretical maximum draw of all components at once with a good safety margin. But even when running a render I’ve never seen it break 500W.
And then to stick it to the man further you’re running Linux of course, right?
I tried Mint and Ubuntu but Linux dies a horrific death trying to run newly released hardware so I ended up on ghost spectre.
(I also assume your being sarcastic but I’m still salty about wasting a week trying various pieces of advice to make linux goddamn work)Levelone techs had relevant guidance.
Kernel 6.14 or greater Mesa 25.1 or greater
Ubuntu and Mint idt have those yet hence your difficult time.
The entire move to the realtime raytracing paradigm, which has enabled AAA game devs to get very sloppy with development by not really bothering to optimize any lighting, nor textures
You clearly don’t know what you’re talking about here. Ray tracing has nothing to do with textures and very few games force you to use RT. What is “allowing” devs to skimp on optimization (which is also questionable, older games weren’t perfect either) is DLSS and other dynamic resolution + upscaling tech
Doom the Dark Ages is possibly what they’re referring to. ID skipped lighting in favour of Ray tracing doing it.
Bethesda Studios also has a tendency to use hd textures on features like grass and terrain which can safely be low res.
There is a fair bit of inefficient code floating around because optimisation is considered more expensive than throwing more hardware at a problem, and not just in games. (Bonus points if you outsource the optimisation to some else’s hardware or the modding community)
That is a prominent example of forced RT… basically, as I described with the TAA example in my other reply…
idTech 8 seems to be the first engine that just literally requires RT for its entire render pipeline to work.
They could theoretically build another version of it off of vulkan-base, to enable you to be able to turn RT off… but that would likely be a massive amount of work.
On the bright side… at least the idTech engines are actually well coded, and they put a lot of time into making the engine actually very good.
I didn’t follow the marketing ecosystem for Doom Dark Ages, but it would have been really shitty if they did not include ‘you need a GPU with RT cores’.
…
On the other end of the engine spectrum:
Bethesda… yeah, they have entirely lost control of their engine, it is mangled mess of nonsense, the latest Oblivion remaster just uses UE to render things slapped on top of Gamebryo, because no one at Bethesda can actually code worth a damn.
Compare that to oh I dunno, the Source engine.
Go play TitanFall 2. 10 year old game now, built on a modified version of the Portal 2 Source engine.
Still looks great, runs very efficiently, can scale down to older hardware.
Ok, now go play HL Alyx. If you don’t have VR, there are mods that do a decent job of converting it into M+K.
Looks great, runs efficiently.
None of them use RT.
Because you don’t need to, if you take the time to actually optimize both your engine and game design.
I meant they also just don’t bother to optimize texture sizes, didn’t mean to imply they are directly related to ray tracing issues.
Also… more and more games are clearly being designed, and marketed, with ray tracing in mind.
Sure, its not absolutely forced on in too many games… but TAA often is forced on, because no one can run raytracing without temporal intelligent upscsling and frame gen…
…and a lot of games just feed the pixel motion vectors from their older TAA implementations into the DLSS / FSR implementations, and don’t bother to recode the TAA into just giving the motion vectors as an optional API that doesn’t actually do AA…
… and they often don’t do that because they designed their entire render pipeline to only work with TAA on, and half the games post procrssing effects would have to be recoded to work without TAA.
So if you summarize all that: the ‘design for raytracing support’ standard is why many games do not let you turn off TAA.
…
That being said: Ray tracing absolutely does only really make a significant visual difference in many (not all, but many) situations… if you have very high res textures.
If you don’t, older light rendering methods work almost as well, and run much, much faster.
Ray tracing involves… you know, light rays, bouncing off of models, with textures on them.
Like… if you have a car with a glossy finish, that is reflecting in its paint the entire scene around it… well, if that reflect map that is being added to the base car texture… if that reflect map is very low res, if it is generating it from a world of low res textures… you might as well just use the old cube map method, or other methods, and not bother turning every reflective surface into a ray traced mirror.
Or, if you’re doing accumulated lighting in a scene with different colors of lights… that effect is going to be more dramatic, more detailed, more noticable in a scene with higher res textures on everything being lit.
…
I could write a 60 page report on this topic, but no one is paying me to, so I’m not going to bother.
GTX 1060 6Gb still going strong!
That was a beautiful card, bought to use with vr, my gf is still rockin that system
Runs FFXIV at 1440p.
Runs HL Alyx on my Rift.
Runs everything prior to this gen.
If I need to run a more modern game, I’ll use my PS5.
Jesus christ man. I thought I was slumming it with a 3070.
hahahahahahahaha.
rx580
RX580 remains a power efficient champ. The old hot hatch of the GPU world.
Uhhh, I went from a Radeon 1090 (or whatever they’re called, it’s an older numbering scheme from ~2010) to a Nvidia 780 to an Nvidia 3070 TI. Skipping upgrades is normal. Console games effectively do that as well. It’s normal to not buy a GPU every year.
As long as you make an upgrade that’s equivalent or better than the current console generation, you’re then basically good-to-go until the next generation of consoles comes.
I don’t really care if my current graphics are better or worse than the current console generation, it was just an illustration comparing PC gaming to console gaming.
Ain’t nobody got time (money) for that!
Still on a 1060 over here.
Sure, I may have to limit FFXIV to 30fps in summer to stop it crashing, but it still runs.
I’m running Linux for everything and my GTX 1070 is still chugging along trying to power my 1440p 144hz monitor ^^’
Well, I mostly just play strategy games and CS2 (which I do have to run on almost the lowest possible settings without FSR. I basically turn everything to lowest except for lowest still AA setting and dynamic shadows to not have a disadvantage and get 110 - 180 fps depending on the situation)
But I’m planning on buying a used Radeon 9070 XT and just inserting it into my current build (i7 6800k based lololol) and on eventually buying a new build around it
(A 750W 80 Plus Platinum PSU should be able to handle a new 970 XT)
They are talking about skipping 1 or 2 generations not taking 10 years off
Hey, it’s not 2026 just yet!
Hey, I’m also on a 1060 still! Admittedly I hardly game anymore, although I am considering another Skyrim playthrough.
Me too, I was considering an Intel B580 mostly cause I feared another price surge…
Unfortunately gamers aren’t the real target audience for new GPUs, it’s AI bros. Even if nobody buys a 4090/5090 for gaming, they’re always out of stock as LLM enthusiasts and small companies use them for AI.
Ex-fucking-actly!
Ajajaja, gamers are skipping. Yeah, they do. And yet 5090 is still somehow out of stock. No matter the price or state of gaming. We all know major tech went AI direction disregarding average Joe about either they want or not to go AI. The prices are not for gamers. The prices are for whales, AI companies and enthusiasts.
5090 is kinda terrible for AI actually. Its too expensive. It only just got support in pytorch, and if you look at ‘normie’ AI bros trying to use them online, shit doesn’t work.
4090 is… mediocre because it’s expensive for 24GB. The 3090 is basically the best AI card Nvidia ever made, and tinkerers just opt for banks of them.
Businesses tend to buy RTX Pro cards, rent cloud A100s/H100s or just use APIs.
The server cards DO eat up TSMC capacity, but insane 4090/5090 prices is mostly Nvidia’s (and AMD’s) fault for literally being anticompetitive.
Nvidia doesn’t really care about the high-end gamer demographic nearly as much as they used to, because it’s no longer their bread and butter. Nvidia’s cash cow at this point is supplying hardware for ML data centers. It’s an order of magnitude more lucrative than serving consumer + enthusiast market.
So my next card is probably gonna be an RX 9070XT.
even the RX9070 is running around $900 USD, I cannot fathom affording even state-of-the-art gaming from years ago at this point. I am still using a GTX1660 and playing games from years ago I never got around to and having a grand time. Most adults I know are in the same boat and either not even considering upgrading their PC or they’re playing their kid’s console games.
Every year we say “Gonna look into upgrading” but every year prices go up and wages stay the same (or disappear entirely as private-equity ravages the business world, digesting every company that isn’t also a private equity predator) and the prices of just living and eating are insane, so at this rate, a lot of us might start reading again.
It makes me wonder if this will bring more people back to consoles. The library may be more limiting, but when a console costs less than just a gpu, itll be more tempting.
I just looked up the price and I was “Yikes!”. You can get a PS5 Pro + optional Blu-ray drive, Steam Deck OLED, Nintendo Switch 2 and still have plenty of money left to spend on games.
I remember when High-end-GPUs were around 500 €.
I have a 4090. I don’t see any reason to pay $4K+ for fake frames and a few % better performance. Maybe post Trump next gen and/or if prices become reasonable and cables stop melting.
fake frames
And that’s my main problem with what the industry has become. Nvidia always had sizable jumps generation to generation, in raw performance. They STILL get better raw performance, but now it’s nowhere near impressive enough and they have to add their fake frame technologies into their graphs. Don’t get me wrong, they always had questionable marketing tactics, but now it’s getting even worse.
No idea when I’m replacing my 3060ti, but it won’t be nVidia.
I don’t think the 5090 has been 4k in months in terms of average sale price. 4k was basically March. 3k is pretty common now as a listed scalp price, and completed sales on fleabay seem to be 2600-2800 commonly now.
The problem is that 2k was too much to begin with though. It should be cheaper, but they are selling ML cards at such a markup with true literal endless demand currently, there’s zero reason to put any focus at all on the gaming segment beyond a token offering that raises the margin for them, so business wise they are doing great I guess?
As a 9070xt and 6800xt owner, it feels like AMD is practically done with the gpu market. It just sucks for everyone that the gpu monopoly is here, presumably to stay. Feels like backroom deals creating a noncompetitive landscape must be prevalent, plus a total stranglehold with artificial monopoly of code compatibility from nvidia’s side make hardware irrelevant.
Technically Intel is also releasing some cheapo GPUs in similar capability to nVidia but they all have the same manufacturers anyways.
There’s major issues with those GPUs in some commonplace use cases and they have major scalping issues. Sure in some use cases there’s zero issues, but this aint like the early 2000s when there were many brands that all basically worked.
Now you’re either nvidia with every feature, amd with most features (kinda like a store brand), or intel with major compatibility flaws with specific games because it’s technically a GPU.
I think patent laws and proprietary software supported by major OS have always had some impact, even the 90s, but yeah it’s definitely a poor state for the market to be this concentrated.
One issue is everyone is supply constrained by TSMC. Even Arc Battlemage is OOS at MSRP.
I bet Intel is kicking themselves for using TSMC. It kinda made sense when they decided years ago, but holy heck, they’d be swimming in market share if they used their own fabs instead (and kept the bigger die).
I feel like another is… marketing?
Like, many buyers just impulse buy, or go with what some shill recommended in a feed. Doesn’t matter how competitive anything is anymore.
