These prices are beyond insulting and frankly I'm glad they're going to take a hit competing with the flood of miner cards entering the market.
Also note that nothing is preventing Optical Flow Acceleration [0] (and subsequently the DLSS 3.0 models that they claim are exclusive to the 40 series) from running on either 2/3 RTX cards. Just like RTX Voice and other gimmick "exclusives" I expect it to be available to older cards the moment they realize their backlog of 30 series cards aren't clearing as quickly as they thought.
They're competing against an over-bloated secondhand market, AMD, Intel, much better integrated GPUs, and comparatively cheaper consoles that maintain sky-high demand with subsidized games via new subscription programs. They're vastly overestimating their brand loyalty (think Microsoft v Sony after the 360) and EVGA's exit makes more sense now than ever.
IMO they still have too many 3xxx cards and don't want to cannibalize sales, hence adding them to the top of the stack instead of replacing the 3080/ti (and renaming their 4070 as 4080 12gb).
The feeling I get from the whole thing is NV doesn't care that much about consumer graphics anymore, and doesn't care about providing competitive value. More of their revenue is coming from business applications, all of which is growing, and gaming looks like its down to 1/3 and shrinking.
Me and most of my PC gaming friends gave up on this hobby in the past two-three years due to the insane GPU prices. I moved to the Oculus Quest 2 and they moved to PS5 and Steamdeck since the price of a modern GPU alone buys you a whole stand-alone gaming console today. Crazy stuff.
I assume the 4000 series is basically a paper launch designed to make the existing 3000 inventory look attractive and make the people who were waiting on the 4000 series being good value rethink and just buy a 3000 instead.
Just to echo the other children of this comment; I'm pretty lost as to the reasoning here.
I game at 1440p on a GTX 1080 from launch day in 2016. Cost is essentially no object for me, as gaming is my primary hobby and I can easily spend 60hrs/week playing video games, so the return would be more than justified in my mind.
But I haven't upgraded in 6 years because there has been no reason to. I'm running my favorite games at 144fps, and the newer super-high fidelity stuff I can run at 60fps, which is more than reasonable.
I will say that I was excited for the 40XX lineup, but at these prices I'm more inclined to hunt for a cheaper 2080 or 3080 solely for RTX functionality, but that's an enthusiast feature even for someone like me who spends a huge amount of time gaming.
I couldn't imagine giving up my access to old titles, console emulators, mods, itch.io experiments, and low-budget indie games on Steam to save a few hundred bucks in the short-term to buy a console.
YMMV if you're not paid well in a HCOL area, or if your location offers alternatives such as PC cafes.
You’re missing the part where you get pipe cred by posting timespy scores in Discord after installing your 4090 and dedicated ac unit.
I historically have bought 1-2 gen old cards and it has worked well for me too.
The big filter right now is 4k 120hz. If you’re aiming for that you need a recent card with hdmi 2.1, which is like 3k nvidia 6k amd series. I upgraded displays recently and it finally forced me, begrudgingly, onto a more current gen card. Really wanted to wait for 4k/7k lines to ditch my old 1080Tis.
No. I am almost certainly an outlier, even among my peer group.
No kids, taking a break from work, very few responsibilities.
When I have a full time job my playtime drops, but 1-3 hours on weeknights plus large binges on weekends can keep the number surprisingly high, though.
This is mostly a matter of multitasking (which I know HN is fairly divded on).
I play games when I watch TV, and I also use games to socialize with my friends.
If an average person added up ~80% of their socialization time and all of their TV/movie/audiobook/podcast time, I suspect that it'd look pretty close to the amount of gaming I do when I have a full time job (probably fluctuates between 10-40 hours per week).
Weekends could easily be > 1/3 of that time (say 24h over 2 days),leaving a reasonable ~5 hours per week day. I don't know about typical, but I've had such a schedule at points in my life when playing games I really enjoyed.
Yes, for some games, but not on a consistent basis.
I still enjoy games, but with adult life I'm playing either after diner (if I'm not too tired) or on the weekends, and gaming competes with other entertainment (series, movies, going out with friends/dates with SO, etc). So as I've grown it takes less time overall.
I am, however, making a schedule for Homeworld 3 and KSP2 when they launch (:
- buy a $800 GPU, keep PC gaming and all the fun maintenance associated with that
- buy a cool new toy like a Xbox Series X/PS5/Steam Deck
- don't do anything, game at 720P with no AA (this sounds sort of like when I had lunchables for dinner yesterday)
Your alternative isn't wrong but they weren't looking to optimize for staying on a PC, but rather, for gaming, and were already thinking in terms of "how much of my disposable income do I want to spend on my hobby?" rather than "should I spend my disposable income on my hobby?"
I had a PS5. I sold it after playing a few exclusives because there was just nothing else to play. Games are more expensive on consoles than PC, no mods, have to pay to play online, and so on.
Plus, the very fact that it's a general purpose computer I can use for work, programming, and cool GPU stuff like Stable Diffusion are what made me get a PC over a console. In other words, I'm gonna need a computer anyway, why not get one that can play games over getting a computer or laptop and also a games console that doesn't do everything as well as a gaming PC?
I'm sure in the long term one would actually spend less on a PC than a console, given all these constraints.
Dual classing a pc as a tv and desk PC can be pretty nice once you get it set up. Something like Nvidia Shield, Moonlight Gamestream software (if you have an Nvidia card), and Playnight launcher can work great if you can hook it all together with ethernet or MOCA over COAX.
Having run an over powered desktop to enable my gaming hobby, I have been toying with putting a "thin client" on my desk and my beefy rig in the living room.
Computer performance has felt ~stagnant for the past decade. Given the number of platforms that build to be mobile friendly, there is increasingly less need to be constantly running a power hungry rig as my main configuration when I am spending most of my time browser the internet. If I want to "desktop game" I could stream from the living room to my desk in the event it is not a controller friendly experience.
> PC -> TV experiences are also not great in general.
Why not? I recently bought an inexpensive gamepad, hooked up TV via HDMI and added a wireless keyboard+touchpad to the mix (I can't get used to Steam console mode). Overall experience is pretty good, effectively indistinguishable from PS4 we had in the office.
It drops into desktop, you have to drag out the keyboard / mouse, it's connected as another monitor on hdmi, so if someone else turns on the tv, that starts interacting with your computer as a second screen. Not to mention the extra annoyances when the tv is in another room.
Streaming boxes like the nvidia shield or similar have their own issues, along with dropping to desktop and so on.
While a dedicated TV console has none of those issues.
I just do both really. There are some games that are just better sitting on the sofa that can be started fast and no fuss. Then there are some games that are maybe more of a slow Sunday afternoon at a desk vibe.
First point is disingenuous, it’s not like you only have that one option. There’s plenty options at lower price points which are still decent. And there’s the second-hand market for the budget-conscious. You can get a decent pc for around 1k, and much lower with second-hand market options.
For the maintenance, it’s not like you’re buying a car. Unless you’re enthusiastic about your pc, you can get a pre-built, plug it in and use it, same as console.
Not to mention that games are cheaper and there’s more cheap indie titles available, so if one is optimising for max gaming enjoyment for the least money, pc looks like a better option. But if the plan is to play through the best titles here and there, then console is a good option as well.
There is far more maintenance or rather troubleshooting with a PC. The Steam Deck too TBH (even on "Verified" games you run into the occasional problem with more frequency than a Switch). You're just far more likely to run into some weird problem or issue with a random title especially that 10 year old game you bought for $1.
On console you match the label to the machine and it works.
This is my story. I've been a PC-first gamer since I was a kid. Diablo was my first "big" game. The pricing of PC components and GPUs in particular have forced me to take a step back and move to console first with a Steam Deck supplementing.
For the price of one mid level 40 series card I can buy two consoles (Xbox Series X + Digital PS5). I know that the games themselves end up costing more, but I don't really care about the majority of new games coming out and the ones I do are often on game pass.
"I know that the games themselves end up costing more"
Just got PS5 and the PS Extra membership whose library has almost every game I wanted. Maxed out my SSD storage on day 1. And I think the XBox subscription is even better. I mean PC is still cheaper but the difference is not that big anymore.
I know, I have a 2080 at the moment. But I don't like where the PC industry is headed so I am not going to support it going forward unless something changes.
You don't have to buy 40 series. Actually, you even don't have to buy 30 series. Even more actually, for gaming only you may buy AMD card, more or less th same that used by Xbox and PS5.
Even at 300 dollars, a 350w 3080 still looks pretty bad. I'd much rather watch prices of the 6800xt, at least until AMD makes their RDNA3 announcement. It also runs (well) on linux too :).
I occasionally play games, including some recent AAA releases. Everything plays fine at 1k/medium-high settings or 4k/medium-low settings on my GTX 1060. No AA, but I don't care that much. The age of needing a new graphics card to play the new games is basically over. IIRC the GTX 1060 is still the most popular GPU on steam. I just "upgraded" to a Radeon Pro VII (for RoCM), with basically equivalent graphics performance, and everything still plays very well.
>the price of a modern GPU alone buys you a whole stand-alone gaming console today.
This has been true for as long as I’ve seen. Mid-range GPUs have historically tended to cost about as much as whole consoles. Pascal, one of the most well-received lines in terms of pricing, sold the GTX 1070 for $400 MSRP, the same price as buying a whole PS4. These new prices being the equivalent of buying two whole machines is what’s truly intense.
Tom at MLID pretty much confirmed as much in his recent two broken silicon podcasts. Nvidia is pricing this way so they can run the 4k and 3k lines concurrently.
I believe that you're looking at the same chart that I was thinking of when I saw this line, which likely incorporates crypto GPUs into the "gaming" segment.
The 3080 released for $800 USD in Jan 2021 which one calculator[1] says is $905 USD with inflation. So a 4080 is $900. At least they are consistent.
edit: 3080 12 GB, that is. The original 3080 was $700 USD in Sep 2020 which is $800 in 2022. So the base 4080 is priced above that.
They're competing against an over-bloated secondhand market, AMD, Intel, much better integrated GPUs
For the 4080/4090 pricing, the only competition is AMD.
Just glancing at some benchmarks, the Intel ARC A380 is worse than a GTX 1650[2] and the best integrated graphics I know of is the Ryzen 7 5700G which is worse than a 1050[3]. I don't see why these would affect 4080/4090 pricing at all.
The $900 one is the 12GB version, which is basically a 4070 compared to the 16GB version, because the memory is not the only difference, but also a different core count, memory bus size and clock speed, which is frankly even more insulting.
The 3080 (10GB) and 3080 (12GB) also had a core count difference, although they did not release simultaneously.
But I realized I did give the price for the 3080 (12 GB) model, which I corrected in an edit. So I think you're sort of right in the sense that the 4080 (12GB) is priced above the 3080 (10GB) in real terms, and that the 4080 (16GB) is priced way above the 3080 (12 GB) in real terms.
They’re different markets though. It’s like $50 bluetooth keyboards for the everyday user versus $200 mechanical keyboards with replaceable keycaps. The latter intentionally requires tinkering because it’s geared towards an enthusiast that enjoys the tinkering.
Outside of the true exclusives, console games are just a subset of PC games. So as a PC gamer, you get access to console games + the majority of MOBA, MMO, RTS, 4X.
I really miss keyboard and mouse on gaming console. It's weird that devices originally for non-gaming is better than a gaming oriented device, but gamepad is really painful for some games like FPS/TPS.
No they aren't. All signs point to ARC's first outing being a swing & a miss, and Intel themselves didn't even have any plans to come anywhere close to competing at the top end. Intel's highest end ARC card, the A770, was positioned against the RTX 3060. And Intel is both missing any semblance of a release date it set for itself, but the drivers are also a disaster and rumors are Intel is going to shutter the entire division, focusing only on data center compute cards only.
So used cards yes are a factor, yes, but if this really is a 2x performance jump then those aren't going to be a factor here. None of the $1000+ GPU buying market is going to care about saving $100-300 to get half the performance after all.
That just lives AMD as competition, and only if RDNA 3 comes out a winner. Which the rumor mill is saying it will be, but AMD's graphics track record is spotty to put it mildly. I'm sure Nvidia's internal plan is to just cut prices if RDNA 3 surprisingly competes at the top end instead of just the midrange, but in the meantime is happy to take those fat fat margins.
> That just lives AMD as competition, and only if RDNA 3 comes out a winner.
The RX 6650 XT is in the same price bracket as the RTX 3050. I'm not sure they would be considered competition even if RX 77whatever were in the same price bracket as the RTX 3050. Nvidia just seems to push just the right buttons on the software features which AMD just cannot seem to get.
Don't get me wrong. I would love AMD to be competition. They should frankly team up with Intel. AMD has the hardware and Intel can provide the software. Don't see that happening considering AMD went alone on their FP8 proposal and everyone else joined up on a separate paper.
Intel's GPU software is currently, and always has been, a complete disaster. AMD's software is miles ahead of Intel's. Nvidia does have more stuff if you care about it though, although the suite of that is increasingly becoming more annoying than valuable. Like GeForce experience constantly demanding to be logged in & collecting telemetry so I can do... Driver updates? So dumb.
But there's occasionally the random great piece of software like DLSS 2 or RTX Voice, definitely. AMD kinda tries with stuff like FSR or Anti-Lag, it's just not as well done as Nvidia's version.
Bingo, Nvidia isn't making enough A100, A10G to fill demand from GCP/Azure/AWS. The idea that they are worried about their position in the gaming industry is off mark.
What's happening is that with mining dying they know the gaming market won't magically expand, there will be a few years of reduced interest. The only way they can make it look good on their balance sheet is if the entreprise segment grows and that means reassigning their capacity towards it even if it means reducing volume on the consumer side.
From a business point of view this makes perfect sense.
My guess is, the problem is the lower end. I'm guessing they bought a lot of fab capacity for the 40 series given fab capacity was highly constrained over the recent years and they couldn't produce graphics cards fast enough.
Now masses of 30 series are being dumped onto the used market from a combination of ex-mining card and ex-gaming cards being sold in preparation for the 40 series. Gamers who were looking for 3060/3060ti/3070's can probably pickup a used 3080 for comparable money on ebay.
So what does the demand for the 4060 look like? I'm assuming a 4060 is a binned down part that didn't make the cut for higher models? You wonder if nvidia is left with a lot of chips that would've gone to the 4060 which they can't sell.
In 2019-2021, IMO gaming market was important because new PS and Xbox were start selling . It was time to get customers to their ecosystem before they could get new console. Now shortage is end, so selling latest model is no longer important.
AMD's decision to have different architectures for gaming and datacenter is still a major mystery. It's clear from Nvidia's product line that there's no reason to do so. (And, yes, Hopper and Ada are different names, but there was nothing in today's announcement that makes me believe that Ada and Hopper are a bifurcation in core architecture.)
Moreover, CDNA is not a new architecture, but just a rebranding of GCN.
CDNA 1 had little changes over the previous GCN variant, except for the addition of matrix operations, which have double throughput compared to the vector operations, like NVIDIA did before (the so-called "tensor" cores of NVIDIA GPUs).
CDNA 2 had more important changes, with the double-precision operations becoming the main operations around which the compute units are structured, but the overall structure of the compute units has remained the same as in the first GCN GPUs from 2012.
The changes made in RDNA vs. GCN/CDNA would have been as useful in scientific computing applications as they are in the gaming GPUs and RDNA is also defined to potentially have fast double-precision operations, even if no such RDNA GPU has been designed yet.
I suppose that the reason why AMD has continued with GCN for the datacenter GPUs was their weakness in software development. Until today ROCm and the other AMD libraries and software tools for GPU computational applications have good support only for GCN/CDNA GPUs, while the support for RDNA GPUs was non-existent in the beginning and very feeble now.
So I assume that they have kept GCN rebranded as CDNA for datacenter applications because they were not ready to develop appropriate software tools for RDNA.
Some guy on Reddit claiming to be an AMD engineer was telling me a year or so ago that RDNA took up 30% more area per FLOP than GCN / CDNA.
That's basically the reason for the split. Video game shaders need the latency improvements from RDNA (particularly the cache, but also the pipeline level latency improvements, each clock an instruction completed rather than once every 4 clocks like GCN).
But supercomputers care more about bandwidth. The once every 4 clocks on GCN/CDNA is far denser and more power efficient.
But what kind of latency are we talking about here?
CDNA has 16-wide SIMD units that retires 1 64-wide warp instruction every 4 clock cycles.
RDNA has a 32-wide SIMD unit that retires 1 32-wide warp every clock cycle. (It's uncanny how similar it to to Nvidia's Maxwell and Pascal architecture.)
Your 1/4 number makes me think that you're talking about a latency that has nothing to do with reads from memory, but with the rate at which instructions are retired? Or does it have to with the depth of the instruction pipeline? As long as there's sufficient occupancy, a latency difference of a few clock cycles shouldn't mean anything in the context of a thousand clock cycle latency for accessing DRAM?
EDIT: So it looks like my memory was bad. I could have sworn RDNA2 was faster (Maybe I was thinking of the faster L1/L2 caches of RDNA?) Either way, its clear that Vega/GCN has much, much worse memory latency. I've updated the numbers above and also edited this post a few times as I looked stuff up.
The weird part is that this latency difference has to be due to a terrible MC design by AMD, because there's not a huge difference in latency between any of the current DRAM technologies: the interface between HBM and GDDR (and regular DDR) is different, but the underlying method of accessing the data is similar enough for the access latency to be very similar as well.
Or... supercomputer users don't care about latency in GCN/CDNA applications.
500ns to access main memory, and lol 120 nanoseconds to access L1 cache is pretty awful. CPUs can access RAM in less latency than Vega/GCN can access L1 cache. Indeed, RDNA's main-memory access is approaching Vega/GCN's L2 latency.
----------
This has to be an explicit design decision on behalf of AMD's team to push GFLOPS higher and higher. But as I stated earlier: video game programmers want faster latency on their shaders. "More like NVidia", as you put it.
Seemingly, the supercomputer market is willing to put up with these bad latency scores.
When we look at RDNA, we can see that many, many latency numbers improved (though throughput numbers, like TFLOPs, aren't that much better than Vega 7). Its clear that the RDNA team did some kind of analysis into the kinds of shaders that are used by video game programmers, and tailored RDNA to match them better.
> I've never seen any literature that complained about load/store access latency in the shader core. It's just so low level...
Those are just things I've noticed about the RDNA architecture. Maybe I'm latching onto the wrong things here, but... its clear that RDNA was aimed at the gaming workload.
Perhaps modern shaders are no longer just brute-force vertex/pixel style shaders, but are instead doing far more complex things. These more complicated shaders could be more latency bound rather than TFLOPs bound.
Nvidia have been making different architecture for gaming and datacenter for few generations now. Volta and Turing, Ampere and Ampere(called the same, different architectures on different node). And Hopper with Lovelace are different architectures. SMs are built differently, different cache amounts, different amount of shading units per SM, different rate between FP16/FP32, no RT cores in Hopper and I can go on and on. They are different architectures where some elements are the same.
No, the NVIDIA datacenter and gaming GPUs do not have different architectures.
They have some differences besides the different set of implemented features, e.g. ECC memory or FP64 speed, which are caused much less by their target market than by the offset in time between their designs, which gives the opportunity to add more improvements in whichever comes later.
The architectural differences between NVIDIA datacenter and gaming GPUs of the same generation are much less than between different NVIDIA GPU generations.
This can be obviously seen in the CUDA version numbers, which correspond to lists of implemented features.
For example, datacenter Volta is 7.0, automotive Volta is 7.2 and gaming Turing is 7.5, while different versions of Ampere are 8.0, 8.6 and 8.7.
The differences between any Ampere and any Volta/Turing are larger than between datacenter Volta and gaming Turing, or between datacenter Ampere and gaming Ampere.
The differences between two successive NVIDIA generations can be as large as between AMD CDNA and RDNA, while the differences between datacenter and gaming NVIDIA GPUs are less than between two successive generations of AMD RDNA or AMD CDNA.
Turing is an evolution of Volta. In fact, in the CUDA slides of Turing, they mention explicitly that Turing shaders are binary compatible with Volta, and that's very clear from the whitepapers as well.
Ampere A100 and Ampere GeForce have the same core architecture as well.
The only differences are in HPC features (MIG, ECC), FP64, the beefiness of the tensor cores, and the lack of RTX cores on HPC units.
The jury is still out on Hopper vs Lovelace. Today's presentation definitely points to a similar difference as between A100 and Ampere GeForce.
It's more: the architectures are the same with some minor differences.
Turing is an evolution of Volta, but they are different architectures.
A100 and GA102 DO NOT have same core architecture. 192KB of L1 cache in A100 SM, 128KB in GA102 SM. That already means that it is not the same SM. And there are other differences. For example Volta started featuring second datapath that could process one INT32 instruction in addition to floating point instructions. This datapath was upgraded in GA102 so now it can handle FP32 instructions as well(not FP16, only first datapath can process them). A100 doesn't have this improvement, that's why we see such drastic(basically 2x) difference in FP32 flops between A100 and GA102. It is not a "minor difference" and neither is a huge difference in L2 cache(40MB vs 6MB). It's a different architecture on a different node designed by a different team.
GP100 and GP GeForce has a different shared memory structure as well, so much so that GP100 was listed as having 30 SMs instead of 60 in some Nvidia presentations. But the base architecture (ISA, instruction delays, …) were the same.
It’s true tbat GA102 has double the FP32 units, but the way they works is very similar to the way SMs have 2x FP16 in that you need to go out of your way to benefit front them. Benchmark show this as well.
I like to think that Nvidia’s SM version nomenclature is a pretty good hint, but I guess it just boils down to personal opinion about what constitutes a base architecture.
AMD as well. The main difference being that Nvidia kills you big time with the damn licensing (often more expensive than the very pricy card itself) while AMD does not. Quite unfortunate we do not have more budget options for these types of cards as it would be pretty cool to have a bunch of VM's or containers with access to "discrete" graphics
Nvidia's datacenter product licensing costs are beyond onerous, but even worse to me is that their license server (both its on-premise and cloud version) is fiddly and sometimes just plain broken. Losing your license lease makes the card go into super low performance hibernation mode, which means that dealing with the licensing server is not just about maintaining compliance -- it's about keeping your service up.
It's a bit of a mystery to me how anyone can run a high availability service that relies on Nvidia datacenter GPUs. Even if you somehow get it all sorted out, if there was ANY other option I would take it.
While I do wish they priced lower, I think when you put things into perspective, it blows my mind that I can buy a card with 76 billion transistors(!) for just 1600$. I suspect the demand for top of the line RTXs will come from enthusiast gamers and ML researchers/startups. Most gamers would be more than fine with 3090 which handles almost anything. The market will be very interesting to watch.
Virtually any contemporary game with maximum settings will struggle to maintain 4k120 without a sacrifice on my 3090. Cranking the render scale in WoW to 6K or 8K (retail or Classic) will easily push the GPU to 100% and drop fps to around 60.
Gaming-wise, I think the 3090 struggles at 8k and 4k 144hz gaming, which is very high end gaming experience and people that spend the money on those monitors will spend the money on the 4090.
For other purposes (like scientific computing/ml), while 3090 can do a lot of things that 4090 does, it's half the speed (or even slower) and a lot less energy efficient, say 1x4090 = 2x3090, 4090 uses 450W, dual 3090 uses 700W, of course you can power limit both of them further, but you still get best bang for your buck out of the 4090.
What games are they running?! While I know not top of the line - I can run the HTC VIVE pro 2 at 5k@120hz at max visual settings in beatsaber on a 2060 based gaming laptop.
Beatsaber is a great game but the graphics aren't very stressful (nor do they need to be).
Keep in mind people are also starting to mod VR into more and more new PC games that were never designed for it. Stuff like Elden Ring, FF7R, the Resident Evil remakes, etc.
$1599 for 4090, $1199 for 4080 16GB, $899 for 4080 12GB (aka 4070). See Nvidia's News page for the announcement.
Edit: Apologies, got the 4080 numbers wrong, accidentally used GBP, corrected above. For the curious, the 4080 16GB is £1269, the 4080 12GB is £949. Meanwhile, the 4090 is £1679.
During the shortage, AMD kept their reference units at MSRP (I was able to pick up a 6800 XT at $649 through sheer luck when others I knew were spending upwards of $1200 on GPUs). With supply loosening up, I expect AMD will continue to play the good guy to steal market share and keep prices sane in comparison to Nvidia's play.
You could buy GPUs directly from AMD, which also had measures in place to ensure no scalpers had the upper hand. This would happen at a certain time of a certain day almost every week, so you could get one if you were patient enough.
Meanwhile NVIDIA stopped production pre-holiday season last year in an attempt to bolster their company value to 1 trillion dollars. Quite a risky play that didn't pay off [1].
AMD outsources it to DigitalRiver, NVIDIA has done that in the past but this time they outsourced to BestBuy (with contractual requirements to sell at MSRP). Both BestBuy and AMD take some nominal measures against bots but it's always been a problem for both sites.
> Meanwhile NVIDIA stopped production pre-holiday season last year in an attempt to bolster their company value to 1 trillion dollars. Quite a risky play that didn't pay off [1].
The source for that article MLID and wasn't borne out by actual production numbers. Actually there was a nominal rise in shipments in holiday '21.
Also, we're currently in a huge glut so obviously there was no availability problem in early Q1 '22 or Q2 '22 either - like, why is pulling back production for 2022 a bad thing, when you're already sitting on a huge stockpile of chips? Were they supposed to keep cranking and make the glut even worse?
People take these techtubers as gospel and a lot of what they say just ends up somewhere between slanted perspective and factually incorrect. But people are willing to believe literally anything as long as it paints NVIDIA in a bad light - just like people went from being absolutely sure that partners were scalping gpus and selling right into mining farms, but then the EVGA CEO says he didn't make any money at all during the mining boom and NVIDIA was capping prices and forcing him to sell at a loss suddenly everyone forgets about the board partners selling $2,000 3080s in their first-party stores. Or the "800-900w TGP" rumors for 4080 that ended up being off by double.
There is just a ton of anti-nvidia sentiment in general and it completely distorts everyone's perception of reality. People will believe any random thing, even if it disagrees with common sense, or their own personal observations, as long as it scratches that "green man bad" itch. And there is a whole little cottage industry of people like MLID and kopite7kimi who thrive on catering to it for clicks. Once those negative frames get started, they don't go away, even if the information used to construct them later turns out to be false.
But people rushing to the defense of poor board partners who say they didn't make any money at all during the crypto boom, heck we actually lost money really takes the cake imo, like that's either actual goldfish-tier memory or massive "green man bad" brain, take your pick. They're the ones who handle sales of all the mining cards too - NVIDIA doesn't run that directly either, it's all partners.
Because the demand for AMD graphics cards has never been all that great. AMD cards languished on the shelves when there was no RTX card to be found. I mean the RX 6650 XT can be found for the price of an RTX 3050. They have been that price for a bit now. If price to performance were a thought to the majority of people buying cards, there would be no way that would be the case. Obviously, gaming performance is not the only consideration. I've been wanting a new graphics card for some time now. I won't pay the Nvidia/miner tax and AMD does not provide one with easy machine learning access.
Not the person you responded to, but rumors, and also, none of AMD's board partners have come out and said they are gonna stop making AMD boards, especially none of the AMD exclusive board partners like EVGA did last week.
I think it's fine honestly - remember we have inflation, semiconductor shortage, and 3090s retailed at the same price like two years ago. Nobody honestly needs this level of hardware for "gaming", like honestly 12-24gb vram? I'm not saying you shouldn't buy it if you want it but it's definitely a luxury product or for non-partnered AI work.
I think this hits the nail its head. Nobody "deserves" a top of the line gaming GPU. And if most gamers can't afford it, game developers will not develop for it as a minimum. Especially for multiplayer games where the total number of players in integral to the game being played at all (which is why many are free to play).
This is not true unfortunately. For high-end 4k gaming you can easily become memory throttled on 8 or 10GB of VRAM. This is the performance cliff you see on cards like the RTX 3070 8GB card. Granted, it's not for everyone, but it's certainly something 4k gamers would want.
Seems like it's higher than it "should" be with GPU mining winding down due to difficulty maintaining profitability. Then again, companies will generally behave with an asymmetric pass-through if they figure it will not impact sales enough by pricing out consumers--prices increase like a rocket, fall like a feather type deal.
I don't know when GPU mining really started to popularize, but there was a significant leap in pricing from the 1080 TI and the 2080 TI. When you roughly account for inflation, it's about on par with current pricing.
Looks like they could trim prices to 1080 TI levels, but I don't see that happening unless sales slow a bit.
Also, I'd like to add, I used general Consumer Price Index to figure this out, so it could vary if inflation did not impact the GPU supply, manufacturing, transport, and other influencing factors similarly.
Is it actually reasonable to be comparing the TI pricing rather than original version pricing? Are they skipping a TI release this series or something? If not then the launch version table looks like this:
Both 3080TI and 4080 are roughly 600mm2+ Die Size. 2080 has a 700mm2+ on a mature node.
1080TI, the only outliner in your example has less than 500mm2 of die size.
And all of that are before taking account of GDDR memory costing, the board, higher TDP requirement and heatsink differences.
While I dont think 4080 pricing is good value or cheap by any standard. I also dont think it is ridiculously expensive. From a transistor stand point, you are getting 2.7x the transistor between 3080Ti and 4090. With 4090 You are buying 76 billion transistor, with 24GB of GDDRX for only $1600.
$/transistor could be the biggest variable here. My understanding is that we haven't been seeing this lower even with node shrinks. Analysts have speculated that this is why Xbox released a trimmed down console (series s)- because a cheaper "slim" wasn't realistic. Similarly Sony raised prices because inflation is outpacing their ability to reduce costs.
>$/transistor could be the biggest variable here. My understanding is that we haven't been seeing this lower even with node shrinks.
Purely from a Wafer, Transistor per Die Space hence Cost per transistor, we are still getting it cheaper per node generation. The ratio, cost per transistor may not be dropping "as much" as what we were used to. But it is still dropping.
However the Initial cost for product development on new node is increasing. This fixed cost, ( think CAPEX ) amortised over "fixed volume" would mean the cost per "die" would also be increasing.
So once you add those together you will see we are at the end of the power curve.
Is it actually reasonable to be comparing the TI pricing rather than original version pricing? Are they skipping a TI release this series or something? If not then the launch version table looks like this:
- 1080 (2016) = $599
- 2080 (2018) = $699
- 3080 (2020) = $699
- 4080 12GB (2022) = $899
- 4080 16GB (2022) = $1199
Doubling the price of the flagship version of the launch product over just 6 years is well, a pretty large jump.
That's non-sense. I omitted inflation data because the parent post had that for similar numbers and did not list the source used. My attempts gave significantly different numbers, so if I gave inflation adjusted numbers, it would be a misleading comparison vs the numbers in the parent's post.
My experience buying my last GPU (1080ti) leads me to fill in the blanks. GPU mining became popular shortly after the 10xx was released. I know this because I couldn't find one in stock and even when I did, it was much more expensive than the launch price.
> Also note that nothing is preventing Optical Flow Acceleration [0] (and subsequently the DLSS 3.0 models that they claim are exclusive to the 40 series) from running on either 2/3 RTX cards.
Are you sure about that?
The page here says it uses "Optical Flow Accelerator of the NVIDIA Ada Lovelace architecture" which sounds like a hardware feature?
You're right, at least for me. Been waiting for a reason to switch, given that 4G decoding capability is available on my AM4 mobo that would give this a further performance boost and available VRAM.
NVidia can suck it and go solely target cloud computing - they already dedicated a significant portion of the announcement to just that. Why didn't the fact A) this is extremely lucrative and B) they dominate there already reduce their price for regular consumers? Corporate greed.
If I see NVidia cards start shipping in consoles, I'll know they decided to rip us off so they could get great deals for the same hardware to manufacturers. A total middle finger to us PC gamers.
I'm getting an AMD vs Intel vibe like I did from years back. I switched then, I'll switch again.
eBay. Make sure the seller has a long history of good reviews and is in your country/region. There are some sellers with over 100 cards they’re trying to offload.
I think it’s great the high end keeps getting bigger and faster.
Nvidia has primarily been associated with gaming. With Machine Learning that’s starting to change. With tools like Stable Diffusion every single artist is going to want a $1600 GPU and honestly that’s a bargain for them!
I think it’d be sweet if you could spend $100 to $10,000 and get a linear increases in cores, ram, and performance. That’s not how it works. But philosophically I have no problem with it.
Hopefully AMD steps up and brings some real competition.
>With tools like Stable Diffusion every single artist is going to want a $1600 GPU and honestly that’s a bargain for them!
Most artists who are not tech savvy to spin up their local Stable diffusion instances are probably going to pay and use one of the online services like DALL-E 2 or something.
It’s already been turned into single-click installers and Photoshop plugins.
Packaging Python is a dumpster fire of local environment complexity. But this is an extraordinarily solvable problem. I won’t call it solved today. But we have Proof of Life of solutions.
I mean shipping people a .zip that has a simple executable that can be reliably run by end users zero external environment variable bullshit or dependencies. No Anaconda, Pip, or any of that non-sense. No fetching of dependencies either. Dead simple turn key usage.
And yeah it’s not super duper hard. It’s just not how Python projects are structured by default. Simple Diffusion isn’t setup that way. Some random person on the internet spent some time to package it up nicely and in exchange they hit #1 on HN. Possibly more than once!
The 3090s were more than the 4090s, and there's been a great deal of inflation in that time. I won't call them cheap, but it seems like a reasonable price trajectory?
For what it's worth on October 2 2020 I was lucky enough to get a 3080 for a retail price of 699.99. (This was only possible due to using a bot for personal use since Covid may everything nuts). Adjusted for inflation this would be around $800 today. There's two more months until launch so the 4080 card is maybe 80 to 90 dollars above inflation.
my thoughts too. As someone that has been buying generations of video cards from matrox, riva tnt and onward. This doesn't exactly seem off to me, I hope AMD brings some competition with RDNA3 this time around.
For most of that time, the Euro was worth more than USD. The exchange has been dancing around the parity line since June. It may still be roughly the same price in USD->EUR.
Sure, but the salaries in the EU didn't go up 20% in the same time so Nvidia is clearly not targeting the same segment of the population anymore, but a 20% richer segment now. Companies adjust pricing to different markets all the time, but for some reason Nvidia didn't do it this time and EU consumers are much more price sensitive than US ones, so EU will likely turn red or blue in the near future and green will lose its hold. Also with crypto gone and cost of living increasing sharply I think this is a huge miscalculation on Nvidia's side and this will burn them.
NVIDIA not giving you a 20% discount on their hardware to adjust for currency exchange rates isn't really the scandal you think it is. I really really wish during the last 20 years when the euro was higher that I would have gotten a "USD discount" on all my purchases from the EU, but that's not how it works.
Nvidia's manufacturing is in Taiwan. US has traditionally much lower prices on electronics than the EU, now with this change EU is completely priced out. SWengs don't earn 800k there, but typically plateau at around 100k at late stages of their careers. This will have a major impact on Nvidia's ability to service EU market. If you tried to buy something from EU in the past, you probably had to pay large import duties which likely skewed your perception.
People buying expensive GPUs never really struggled with increased cost of living. For the budget sector the difference between a 300 Euro card and a 500 Euro card is significant but for the enthusiasts or business owners it doesn't matter if it's 1800 euros instead of 1500 Euros, they'll still buy it.
Nvidia wants to be the Porsche of cards, not the VW, so they're pricing themselves accordingly.
I am in their "Porsche buyer" category as I bought 1x RTX A6000, 1x Titan RTX and 4x RTX 3090 for Deep Learning, but I am staying away from 4090 until they get some sense with pricing.
From my point of view, 2x NVLinked 3090s give me the same benefit as one 4090 at 150W higher power consumption and nothing is super appealing about 4090 until drivers on Linux and PyTorch/TensorFlow implementation get stable enough for production use which took like 6 months last time, so there is no pressure in buying it.
Why do you need a 4xxx for PC gaming? A 1070/1080 is still a very decent card. There's no game released that won't run at reasonable res/fps on 1xxx series cards.
Agree: the PC gaming market size[1] continues to grow and is definitely very very alive.
I am guessing the dead rumour is because the console market is growing faster. Conveniently ignoring that the mobile market is growing faster than the console market (mobile has 50% of market value by some analyses).
Conversely, AAA games are mostly not exclusive to PC keyboard/mouse control (games market could be split by input devices: mobile touchscreen, console controller, PC keyboard + mouse).
> competing with the flood of miner cards entering the market.
Remember: don't buy miner cards and let them rot. They artificially generated scarcity thus raising the prices for everybody during the pandemic. And don't believe that a card that has been on 24/7 is safe to use. "but it was undervolted!" means nothing.
Just let them become e-waste and buy something new. Turn on the stove and open all Windows when not at home while you're at it. SMH.
And that "not safe to use"? You think it's gonna explode? Worst case it doesn't work, but even that's probably pretty low risk. Hardware doesn't "wear" that way that quickly.
Exactly. The thing that stresses silicon the most is extreme change of temperature. Guess what doesn't change during mining: Temperature.
When mining, you want a steady rate and will most likely downclock and undervolt it in order to get maximum efficiency out of your card.
The only thing I could imagine to break are capacitors, but I still have to run into anyone who has issues like that.
GPU mining is poised to be permanently done. There is no "lesson" to teach miners, because there is no expected repeat of such a run on GPU mining again.
Meanwhile, the implication from your post is that it is better that the tens or hundreds of thousands of GPUs miners currently own and may offload would be better in a dump, just for retribution.
Nicely said. Actually there is a point - vengeance and moralizing. But you're right this probably wouldn't have much effect on GPU availability and reduction of pollution in the near future.
His argument is that by buying a miner card, you're validating strategy of miners, which are considered "bad actors" by gamers and environmentalists. If you intentionally decline buying a miner card, you're helping to make miner activity less profitable, thus doing the "right thing".
>Why don't you go to a beach and scoop up some oil from the BP spill?
If it was trivial, why not? If you told him "don't buy any GPUs" then it would make sense. But either way he will be using some "oil" (GPU) in this case so I don't see why he should waste perfectly good "oil." I suppose you would rather leave the oil in the ocean?
I have never understood this argument. How much environmental damage is done by the mere creation of all television shows cumulatively, the creation of plastic tat given away at conventions, plastic wrapping at grocery stores?
But graphics cards are what we REALLY need to focus on and you better believe you are tuning in for the next big show. It's almost manufactured.
I would actually be quite interested in seeing some numbers to back that up, but I'm not sure if there have been studies on the damage of mining yet.
In some ways, letting a GPU run at a constant load (and thus constant temperature) is less damaging than a typical gaming workload I imagine, where the temperature fluctuates often. In other ways it probably isn't (e.g. atoms of the chip 'migrating' over time with the constant electron flow, though I have no idea how much of an issue that is with modern chips).
> I would actually be quite interested in seeing some numbers to back that up, but I'm not sure if there have been studies on the damage of mining yet.
The few investigations have generally found mining cards are perfectly fine and if anything above-average in used quality. Because yes the primary damage is from thermal cycles, not just being on. And since mining is a constant load & constant environment, it has very few thermal cycles. And miners primarily care about efficiency, so they're power-limiting them[1] which further reduces strain on the cards.
Fans might be closer to death than a typical used card would be, but on pretty much any GPU those are cheap & easy to replace. That's really the only "risk"
1: And before someone inevitable scoffs at this, it's common enough that this is built into tools like nicehash. It's literally "click 1 button" territory, not exactly something a miner wouldn't attempt.
Do you have any idea about where these were kept? Not all mining operations are commercial-like white rooms with people going around in sealed suits. Maybe someone had half a dozen cards in his chicken coop, exposed to dust and moisture and energy fluctuations. You will never know. Would you happily buy that at say, 50% MSRP?
Yup, assuming it works and checks out the first 24 hours I have it.
Solid state electronics are pretty robust. If they survive whatever torture you put them through plus being shipped to my door, chances are it will work just fine. In my experience there are some problematic GPU models that are more prone to early failure under heavy use (HD6990 I'm looking at you!), but they are more rare than legend may lead folks to believe.
At an appropriate discount it's far cheaper to just buy these at pennies on the dollar and just be alright with self-insuring for a rather rare early failure. When you can buy 2 or 3 for the price of new, it starts to get rather compelling.
As long as there isn't any corrosion that will lead to premature failure, I don't care. Dust is easy to clean off and I would expect anyone trying to sell a $500+ GPU to at least try to clean it up a bit. Even if not, I have an air compressor and a blow gun. "Energy fluctuations"...what? Power supplies and VRMs are built to deal with this. Check your home with a power analyzer with a high sampling rate and see what happens when your AC kicks on. At the same time, measure the 12v rail going into your GPU when that's happening.
How do you tell there's no corrosion though? You need to at the very least take off any coolers on the card and look at the PCB itself to have some level of confidence that there's no damage, and even then I've seen corrosion sneak under components where you need to pay really close attention to notice it.
If I can inspect a card in person, I would be plenty happy to buy an ex-mining card, but most of these are going to be sold online sight-unseen. This isn't a problem exclusive to mining cards though; people have been selling damaged GPUs as 'working' for years.
> Are you claiming that it is somehow unsafe to use?
For your wallet, yes. Do you have any information about the environment in which a card was used? Temperature, dust, air moisture were like in a home or there operated in an industrial warehouse? Fan bearing status after all these hours? Voltage applied? Stock BIOS? were they in an enclosure or in a sort of a case?
Thermal paste has been changed? Ever been disassembled?
I would not trust a miner's card even if it was a gift.
Miners screwed the market over and now they want you to think that their used crap is such a bargain. Let them deal with their waste, don't help them.
> Do you have any information about the environment in which a card was used?
That would be even more true for cards owned by non-miners; most miners will operate at a large enough scale to use some best practices, which definitely isn't true about most people who buy GPUs for personal use.
This is but one test, but Linus Media Group did some testing and found that mining GPUs worked just fine. Fan time and thermal pad degradation are about the only real detractors from buying a used GPU. https://www.youtube.com/watch?v=hKqVvXTanzI
Of course, if a card's running at thermal throttle for long periods of time, that has a chance to damage the card, but given that most cards are running underclocked, the likelihood that you'll run into that problem is low.
So the 4090 is 38% faster (in FP32 FLOPS) than the current top server GPU (H100) and 105% faster than the current top desktop GPU (3090 Ti). And it's also more than twice as efficient (in FLOPS per watt) as all current top GPUs, even compared to the H100 which is manufactured on the same TSMC 4N process. This is impressive.
The computing power of these new GPUs (in FP32 TFLOPS) is as below:
The FP32 comparisons show only a very incomplete image about the RTX 40 series vs. server GPUs.
Both the NVIDIA and the AMD server GPUs are about 30 times faster at FP64 operations. At equal FP32 Tflops, the server GPUs would be 32 times faster at FP64 operations.
A 16 factor comes from having an 1:2 FP64/FP32 speed ratio instead of an 1:32 ratio, and an additional factor of 2 comes because the server GPUs have FP64 matrix operations, which do not exist in consumer GPUs.
So the difference in FP64 speed is really huge. Moreover, the server GPUs have ECC memory. When you have a computation that takes a week, you would not want it to provide wrong results, forcing you to redo it.
Of course, server GPUs remain kings of the hill on FP64 workloads, as they have extra FP64 units. I focus on FP32 performance because it is representative of general compute performance on all other workloads, such as FP32, FP16, FP8, or integer.
The more significant changes have not been detailed.
- What will MIG look like on these GPUs?
- Are there multiple graphics queues now?
Otherwise, the RAM remains the greatest limitation of NVIDIA’s platform. They should really design a RAM add in standard and decouple it from the chip.
IMO NVIDIA’s greatest competition is against Apple, who have achieved a unified memory architecture and a naturally multi queue (actually parallel) graphics and compute API. Everyone from normals to the most turbo “ML boys” do not comprehend yet how big of a deal the M1 GPU architecture is.
It seems like they're really emphasizing the difference in RT performance from the previous generation, but I think the gaming market at least will care more about the difference in raw frames and memory size from the previous generation and AMD's offerings.
Personally, I like using RT for some single player showpiece games with a 3080 Ti, RT was useless on my 2080, but the games that I play the most do not use RT at all. DLSS is always great on any title that offers it, but again the real issue is that most of the games that people put real time into are the types of games for which RT is irrelevant. Graphical showpiece stuff is just a lot less relevant to the contemporary gaming market than it used to be.
Eh, the current 3080 can already do 60FPS @ 4k HDR on every AAA title I've thrown at it, and that's with an mITX undervolt build (cliffing the voltage at a point). "60FPS @ 4k is readily achievable" has been sort of the gold standard we've been careening towards for years since displays outpaced GPUs, and we're just about there now. The raw frame difference and memory size is nice especially if you're doing compute on these cards, but these weren't holding the gaming market back, at least. So for those segments, you need some juice to go on top of things. I can see why they advertise it this way.
Personally, people say RT is a gimmick but I find it incredible on my 3080 for games that support it. In a game like Control the lighting is absolutely stunning, and even in a game like Minecraft RTX, the soft lighting and shadows are simply fantastic. (Minecraft IMO is a perfect example of how better soft shadows and realistic bounce lighting aren't just for "ultra realism" simulators.) It's already very convincing when implemented well. So I'm very happy to see continued interest here.
Try MSFS in VR on ultra settings. With mostly high and ultra I'm sometimes able to get 30fps on my quest 2. Nevermind something like the Pimax 8k or upcoming 12k.
Yeah that's fair, VR titles are still a place where GPUs need more performance. I personally have no interest in them but yes the compute needs are in general way higher. Not to mention Flight Sim is a particularly intense case...
I finally got a chance to check. DLSS is set to "auto".
General settings are: Ray Tracing Ultra, Texture Quality High, all advanced settings set to "high", ray tracing is set to "Ultra" not "Psycho".
Looks like I was wrong- the benchmark shows average FPS of 30, not 60. When I change DLSS to "Ultra Performance", then the FPS is 59.29 (I assume vsync) and I didn't see any visual differences in the benchmark. With DLSS at "Quality", average FPS is 27, with drops as low as 9, and obvious visual frame skips.
At the very least, seems like I should force DLSS to Ultra Performance, so I bring the average FPS during complex scenes up.
The 3080 Ti and 3090 have the same framerate on that page. I updated with a comment to show the FPS at various settings- only see 60FPS with DLSS set to Ultra Performance (all the other settings are maxxed).
I went back to run some more benchmarks and it turns out with DLSS totally disabled, 4K is unplayable (3-6FPS). Didn't realize DLSS was contributing so much. Simply disabling raytracing at that point (so, DLSS disabled, 4K, everything else set to max but RT disabled) gets me back to 50-60FPS. So basically DLSS is compensating for RT. But disabling RT doesn't really seem to make a big difference- it still looks pretty damn nice. hard to tell.
Yeah RT is a killer in CP2077, CDPR did a bang up job with the raster graphics. With everything jacked I mostly noticed the RT in the puddles/water reflections vs. the screen space reflections but not enough to enable it. I'm sure it does other stuff lighting wise but I didn't notice it personally.
Op claimed to get 60 FPS at 4K with "pretty much" everything at max - I hope that wasn't implying "everything but raytracing". I got the card for raytracing, without raytracing even my 5 year old card worked ok and looked good enough.
How about in the actual game? I have a 3090 at 1440p and swing 60 a lot at maxed everything and with dlss at quality, but more often than not it sits in the 45-50 range, that said I suspect I may be CPU bottlenecked atm.
No, but not for lack of interest. I simply upgraded from 2k res + 2080 Super to 4k res + 3080 at one point in the last year or so, but I had already beat the game by then and haven't returned to it.
I suspect they also highlight RT performance (and AI acceleration which is more-so focused on a different market than these gaming cards) because it is their key differentiator with competitors.
Most upper market cards can already run most games well at 1440p or 4k.
I’d be interested in better RT performance for the purposes of VR gaming but, unfortunately, the high fidelity PCVR gaming market died with the Quest 2.
No, average people never play VR with wired headset and highend PC so the market can't grow. Quest2 is acceptable device, and it can be upgraded by wireless Oculus Link with highend PC.
While RT is a single player gimmick I mostly turn off, it won't take many more increases like this to make it a very real feature. What will we see when developers start targeting these cards, or their successors?
If it's actually another 2x performance improvement, we're really only another generation away at most. On my 3080 RT caused a performance drop but I was able to decide the trade-off for myself, often it was still playable, I just needed to decide whether I wanted high refresh rates or not. Another doubling or two and it'll barely be a thought at all. The caveat with this still is it requires DLSS for best effect, which is barely a trade off in my experience.
I interpret it to mean that the graphical enhancements that RT can give you are only really appreciable if you're playing a single-player game that allows you to stop and check out how beautiful it is, while in a competitive multiplayer game, you're moving around too fast to really appreciate (or even notice) the difference RT gives you over fake reflections.
> you're moving around too fast to really appreciate (or even notice) the difference RT gives you over fake reflections.
You're also at a disadvantage running the graphics on High.
In any game that I want to play at a "competitive" level, I always put the settings to the absolute lowest settings so the computer always gets 100+ FPS.
No reason to even try to optimize. The point is winning, not it looking good. In fact, in some games, the developers have had to go back and make it more fair for higher settings because of low settings hiding tall grass.
It's a spectrum, not an either-or. Many people play competitive multiplayer for the sake of overall enjoyment, not solely to compete. For that, you want the graphics to be as good as possible, to the limit of what you can notice while playing the game normally.
We're past Quake-level graphics, but competitive eSports games, by and large, have deliberately simple graphics optimised for readability over flare.
If you look at LoL, Fortnite, Valorant, Overwatch, they're all kind of the same — fairly flat light, bright colourful characters against a backdrop of pastels and otherwise muted background colours. Texture detailing is discrete and everything almost looks like block colours if you squint. Rainbow 6 Siege, CSGO, PUBG are the same, minus the colourful characters.
Even if those games didn't optimise their engines for performance over visual fidelity, their visual design would make them fairly light weight anyhow.
Another reason why those game graphics is like that is they want to support many devices, including very low-end Nintendo Switch and MediaTek smartphones. Toon style looks good in low quality.
The reason behind the 4:3 low resolution is outside of scope for ray tracing or performance overall: most of pro players found out that widened picture and thus player models are easier to target precisely in motion (AFAIK the phenomenon hasn't being studied properly) and you're limited by vertical resolution of the screen thus 960 is the upper bound for a common 16:9 1080p panel.
For PC, at least judging from Steam charts, single player gamers are a substantial minority, but still a minority in terms of what games are popular. CP2077 is the only single player game currently in the top 10, and there are only a couple other primarily SP games in the top 20.
At no point of history have multiplayer gamers outnumbered singleplayer gamers. Even for MP focused titles like StarCraft 2, the majority (iirc the number was around 54%) of customers never clicked on the multiplayer button once.
You're looking at number people logged into an online service and thinking it represents anything about gamers that by definition aren't online. A typical sampling bias.
Sounds like you haven't paid attention to the gaming industry since eSports took off.
Overwatch and Valorant alone may have increased female participation in video games by double digit percentages. Also, the TV show for League, which was obviously aimed at women.
> You're looking at number people logged into an online service and thinking it represents anything about gamers that by definition aren't online. A typical sampling bias.
steam, ubisoft connect, origin, blizzard launcher, and rockstar are all online today, even for single-player games. I'm not sure where a significant number of non-platform-delivered games would even exist in 2022, I haven't bought a single game that didn't use steam or another platform from a vendor in ages.
The exception being GOG I guess, and to be fair there's a decent number of new-ish titles on that (horizon zero dawn) and I always make an effort to buy titles there when possible, but even still GOG's marketshare is minuscule in comparison to steam.
so no, actually, I disagree that there's some large repository of single-player games that are not showing up on steam or another equivalent.
There are more single-player games than multi-player, but overall more hours have been spent (or are currently being spent) on multi-player games. For example, someone might play BioShock Infinite once and complete the story in under 15 hours and then never touch it again, but easily rack up hundreds of hours in Apex Legends.
As you've said though, there's a sampling bias. Going by the "current players online" metric might diminish how many players are playing single-player games without being online.
I don't think that is a good metric. There are far more single player games available on Steam than multiplayer.
A multiplayer lives and dies by its community, so it makes sense that multiplayer gamers would coalesce around a relatively smaller number of games.
Multiplayer games are also designed far more around player retention. The unpredictability of human opponents, combined with various design decisions meant to keep players engaged, means that a multiplayer gamer may primarily play the same game for years. A solo player reaches the end of their game, maybe they even exhaust all the side content and achievements, but then they move on to the next single player game.
> What does "single player gimmick" in your sentences here mean?
When I've seen this written it's often explained as to do with the fact that single-player is a much more curated experience, where the developers can stage-manage where you go and even where you look, and discard resources they know you won't see again.
That's why single-player often looks way better than multiplayer, where a) the experience isn't that curated and b) fps matters much more.
“In /my opinion/, this doesn’t add enough to the game experience to justify the performance penalty. Also, my hardware isn’t strong enough to use it. Therefor it’s a gimmick! Herpa derpa!”
That may actually be giving too much credit. It’s probably more like “I don’t like it so it’s a gimmick that no one else should enjoy”
Divining people's real motivations is a skill wasted on HN news comments. If you're as good at it as you think you are, you should be brokering multi-billion dollar deals.
Gamers don't see it that way though because often, faked reflections are "good enough".
In some cases though, the faked reflections are noticeable and immersion breaking. I notice it in MS Flight Simulator. I'm flying over water and approaching a large city, and I can see the reflection of buildings in the water. But when I push the nose down, as the image of the buildings goes off the top of my screen, the reflections go away too. In reality, this wouldn't happen. But in the sim, it does because the reflections are generated by just inverting the rendered image of the buildings.
There's a map in Apex Legends where a section has a shiny floor. It fakes a reflection by rendering an environment-mapped texture. It does a decent job at making the floor look like it's reflective, but as you move around and compare what's being rendered to the actual geometry, you notice how it falls apart. But in a competitive FPS, you're not looking at the environment long and hard enough to notice.
Duke Nukem 3D had proper reflections, but it did it by actually mirroring all the geometry, a technique that only really works well if the geometry is simple.
Not everyone is a pro CS:Go player. I really like RTX shadow improvements and wish more games supported it. I've bought games for their "showpiece stuff".
I wish Valve would put a bit of investment into TF2, as it's still a very popular title with ~100k players on average[1], yet it's 32-bit ONLY and stuck on DirectX 9. The Mac version doesn't even work anymore because it was never updated to be 64-bit. The dedicated server, SRCDS, is also 32-bit only and struggles with 32 players and some fairly light admin mods. It's also mostly single-threaded, so it tends to run better on Intel CPUs with high clock speed.
> I'm not sure I understand why memory is important for gaming? For most games, with every settings maxed up, it'll be a stretch if it uses 6GB of VRAM.
Depends on what kind of gaming you do, not all games consume the same type of resources from the GPU. Simulation games will usually be limited by RAM size/speed, VRAM size/speed and CPU, rather than actual clock speed of GPUs. Action games would require better latency and not so much about VRAM/RAM size and so on.
For the majority of games, no it doesn't matter at all. For a small number of games, it's not hard to eat up all the VRAM with certain settings and especially with 4K gaming. I use a 1440p monitor so it doesn't really matter that much to me, but it does matter more to people who spend a lot on good 4K displays.
I've yet to see a game that my laptop 2080 can't handle on max settings. The only games I've done that asked for raytracing I did was Control and Resident Evil 8.
The pool of games that ask for extremely high performance is very small and pretty easy to ignore by accident.
> DLSS is always great on any title that offers it
It is very subjective. DLSS for me is always a blurry mess. I'm turning it off in every game - I'll better will turn off some RTX features than suffer that blurriness torture.
What the hell Nvidia. Post EVGA breakup, this is a bad look. Seems like they're setting MSRP ridiculously high in order to undercut board partners down the line.
It’s funny, I just had to have 40 feet of wood fence replaced and paid ~4K for parts and 1 day of labor. They also took away the old fence. My point, electronics are cheap. I know this point will not be received well though.
I had a similar sentiment when I had to purchase a set of new tires for my car. It is amazing how an assembly of rubber and steel can cost roughly as much as a device that has billions of transistors on it.
Bigger and thinner tires looks ridiculously expensive than it should (as a noob). I don't know the reason, does every tire manufacturers want to earn more from riches (who can buy a car with bigger and thinner tires)? This is important factor when I choose car.
I paid 1600€ for my last full PC (kept the old chasis and GPU) and 1700€ for my current work laptop. No, 1600$ (1900€ in Europe) is not cheap by any stretch of imagination.
I feel like both of those might be frivolous, but wood and labour are a more easily arguable expenditure. They're physical and do something physical with finite resources. A gpu often just gives you imaginary shadows faster. TVs just show you stuff, but for the same price of a nice one you could eat food for a year or travel overseas.
For $2000 you could get better light beams in a game you're already playing, or if you're physically capable you could take a mountaineering course and get a helicopter to and from a remote climb
So if you're in the business of having almost anything else to do with the money, it's going to seem like a really tenous financial choice.
It's why people pay thousands of dollars for a handbag cost maybe tens of dollars to produce. People like playing status games.
This effect is why third richest person on the planet is Bernard Arnault of Louis Vuitton, someone you've likely never heard of. He's currently wealthier than Bezos. He didn't actually improve our lives or revolutionize industry. He just latched into our primitive status seeking behavior.
To be fair to Arnault, he isn't just CEO of Louis Vuitton but LVMH which owns something like ~100 brands, ranging from clocks, wines, fashion, cosmetics, boats, jewelry and other retailing. Although to be fair to you, most of what they sell could go into the "luxury" category.
It's likely he wouldn't be the third richest person with just Louis Vuitton and it's probably mostly because of LVMH.
Graphics cards are not some idiotic luxury good. They are a tool, used for work or leisure, whose price points seem to be set at luxury level for essentially arbitrary business reasons.
It absolutely can. There's only 2 players in this business, 3 if you count Intel but their GPU future doesn't sound promising.
Nvidia sets the bar and AMD follows. AMDs prices have slowly crept up as well. So no matter what, you're paying much more for a top spec GPU today than you ever have before.
It's in the best interests of both Nvidia and AMD to keep these prices high, and it's a detriment to the consumer.
AMD did it with their CPUs as well. They undercut Intel with some great price/performance deals. Grabbed their market share back and are now bumping those prices back up to parity with Intel. Now neither company has to cut their prices.
Nvidia will use artificial scarcity to push these prices. We know they can produce a shit tonne of GPUs, they did it for years.
While what you say is correct, the recent Ryzen 7000 announcement has shown that AMD remains much less greedy than NVIDIA, because with the top Ryzen 9 7950X at $700, the price per core has remained less than $50, about the same as it has been since the first Ryzen, even if the cores are now much faster.
Before the first Ryzen, only Intel Atom CPUs had a price around $50/core, while the better Intel Core CPUs had prices around $100/core. Then Intel had to also follow AMD and drop the prices for the top models to around $50/core (the "efficiency" cores in Alder Lake and Raptor Lake count as a half core).
So there is still hope that the RDNA 3 GPUs will have a more decent performance/price ratio, even if it is unlikely that they will be faster than the RTX 40 Series.
>Wasnt like half of all US dollars printed in last 3 years ?
No. "Printing Money" is a shortcut term used for many of the measures used by the federal reserve recently, but it's not literally printing money in this case (significantly less than trillion has actually been printed since 2019). Going into the details of the economics of this is a mess, and inflation is absolutely high and causing problems right now, and very likely is higher than official CPI numbers but it's not anywhere near 100% in 3 years. Keep in mind lots of prices dropped in 2020 and were still recovering/suffering in 2021, so that paints a skewed picture vs comparing 2022 to 2019 before covid wrecked things.
> It feels more like 700$ in 2013 would be worth now 1400$??
USD$700 circa 2013 adjusted for inflation is USD$890 in 2022. Feel free to pay more if you feel like it should be more though.
When it's not a phone, but an always-on 6.5" tablet computer with an excellent camera, persistent wireless high-speed internet, amazing screen, thousands of incredibly useful apps that obviate the need for dozens of other electronic devices, and near-desktop performance for web browsing anywhere, I don't mind paying mid-range laptop prices every few years when I use it literally 4-6 hours per day, every day.
I’m a little envious of people that can get away with that. I never will as a developer. But I would definitely switch from Apple to Samsung if I only had a phone. Life would be a lot easier with Dex docking.
In the US, at least, its common to simply pay that $1k phone over time on a plan - or simply never stop paying off a phone through an upgrade program, either through their carrier or direct (e.g. iPhone Upgrade Program).
$30/mo doesn't hit as hard as $1100 up front. For some consumers, its a fundamentally different way of thinking.
I wish there was a more open adoption of CUDA like functionality. I use an astrophography imaging tool called PixInsight and it's very parallel math heavy. It supports CUDA, and you can offload some tasks to the GPU and it makes a world of difference (computing a mask of stars takes seconds instead of 5-15 minutes). It's certainly got me locked into NVIDIA.
Kind of. Vendor lock-in is only a part of why it's the only game in the town. Another, larger part is that AMD's software support is awful, and also that ATI/AMD were ignoring GPGPU/ML stuff for a while, letting Nvidia indoctrinate academia into using their stack.
As a side-note, the X-Box series S was the cheapest and most effective way for me to buy a 4K-UHD blu-ray player.
I did try to do some CUDA based AI rendering stuff on my 2080S but 8GB didn't seem to be enough.
Its weird to comprehend the 'stretching' of technology advance over time as I age, especially on the value side. There hasn't ever been the same 'feel' for me from the first leap of moving from software rendered Quake to a 3D-card - despite the various advances since, although I remember bump-mapping as another massive leap.
It's totally unfair in some ways. As an example the control you have over lighting a scene (either in a game or something you're rendering) is way beyond what multi-million dollar studios were using in my younger years.
What happens to society/reality when technology capable of producing video content indistinguishable from reality is affordable to many? Its already happening. Its going to become more commonplace.
The problem of 'truth' becomes massive - is the thing presented to you something that actually happened or was it fabricated?
24GB is enough for some serious AI work. 48GB would be better, of course. But high end GPUs are still used for other things than gaming, from ML/AI stuff to creative work like video editing, animation renders and more.
I completely agree. In the past I dropped a ton of money on gaming rig hardware that aged like milk. With a console you get the advantage of exclusives, majority of PC game releases, and a longer upgrade cycle versus a gaming rig. If you own a PS5, you got the PS VR2 coming out soon at a decent price point. If you own an Xbox, add in the amazing value of the Xbox Games Pass and I just don't see the need to be subsidising Hardware Manufacturers' bottom lines anymore.
I have a desktop PC that I sometimes use for gaming, everything from demanding simulators (like Flight Simulator and X-Plane) to grand strategy games and other more mainstream games. Last GPU I bought was a 2080ti that still is my GPU, and I bought that something like 3 years ago. I can still game mainstream games on very high settings without issues. So not sure why your setup aged like milk but mine didn't...
On my home machine, I'm running SLI'd GTX480's. They still do 99.9% of the gaming I want to do. I have only found ray tracing (which I don't like anyway) and for some reason water reflections to cause problems. But I can still run most games on high or better settings.
I don't know why OP's setup would age like milk. As long as you keep them clean, games haven't really advanced that much in the last few years, in terms of graphics requirements.
Also, if anyone has advice on building a new machine, I'm listening. I've been out of the game so long that I don't even know where to start, and have children that want towers built.
Yeah, it's kinda bizarre - I've been running GTX970 for 5 years and it still runs things acceptably today (although I've replaced it with 3080 at launch).
It really didn't last any less time than PS4 or any other console.
Consoles are subsidized and benefit from quantity. PC gaming has always been more expensive and benefitted from more cutting-edge technology. I love that consoles make gaming more accessible to more people, especially now with online multiplayer, which was once reserved for PC gamers. Even phones can do VR now, but PC is still the highest resolution, best texture quality, most detail, etc. Driving games in particular, with a Fanatec wheel, IEMs, Index or Vive Pro 2, and TOTL GPU + CPU are pretty damn spectacular, but the fact that you can get something 50-70% as good for ~1/5th the cost with a console is impressive.
ps5 hardware is sold at a profit (they hit profitability within 6-9 months of launch) and while microsoft claimed to a court that their hardware was subsidized as a justification for refusing to open the platform to alternate stores (so, somewhat self-interested of course), if you believe the production costs are fundamentally similar to PS5 then this is most likely "hollywood accounting".
It's real easy to have the xbox division license their branding from a parent company and if we set that at $100 a console then oops, there went all the profit.
It's not the PS3 days anymore where sony is losing 30% of the value of the console on the initial sale (and actually that wasn't the case for the other vendors even back in the PS3 days). Consoles are sold at a slight profit these days.
That said, having a "big APU" where everything shares a single bank of RAM and a single cooler/etc is a massive cost advantage. There is a lot of redundant cooling and memory in an ATX/PCIe spec PC and it all adds up. A console is one product with one assembly/testing line and one cooling/fan system and one set of memory. Clearly there is a market for "console-style PCs" which apply similar cost-optimizations, with a similar model to Steam Deck. Just so far there's nobody who's been willing to do the up-front cost and yet will choose not to lock down the resulting platform - Valve is somewhat unique on that front.
A Series X will not render very many AAA games at 4k/120fps (if any at all). And it definitely won't do at Ultra level settings.
PC gaming at the high end has always been about spending a lot of money to get the best graphics possible. Consoles always have much better bang for the buck early in a new console generation.
The 4K 144hz you get on the Xbox Series X is almost certainly lower graphics settings than you could run at 4K 144hz on the RTX 4080. I don't think the difference is enough to justify the price difference but comparing resolution and framerates isn't apples to apples.
And you'll still be playing the same game. If you want to spend almost a thousand dollars so your game renders more eyelashes that's your choice. My point is that NVidia's pricing is making PC gaming inaccessible to people who don't have wads of disposable cash to spend on diminishing returns. I would strongly recommend anybody who's looking into getting into current gen gaming to steer clear of playing on PC as the pricing of PC components is completely out of control and console pricing is looking increasingly more reasonable.
How is it making PC gaming inaccessible? You don't need top of the line hardware to play even the most recent PC games. You need top of the line hardware for "more eyelashes", but that's very much a choice (that you get with PCs but not with consoles).
4K 120hz is merely the limit of the HDMI standard used on the Series X. It is not a hard requirement needed to publish on the platform. This would be akin to claiming that the PS3 was a 1080p 60Hz console, when the vast majority of the library managed closer to 720p 30hz.
While developers are free to make 4K 120hz their target, few games attempt it, and the ones that do, often make significant graphical compromises or are graphically undemanding to begin with.
Raising MSRP doesn't undercut board partners, it allows board partners to charge more which is what they want.
Personally I think what they're doing is maximizing launch profit without totally crashing the 30x0 market which has a supply glut, something else board partners are worried about.
I'm really not sure why you think this is bad for board partners - you haven't really explained your reasoning and what you did say doesn't make much sense. This is all positive for board partners. It's the cheap founders edition cards that undercut board partners.
> Raising MSRP doesn't undercut board partners, it allows board partners to charge more which is what they want.
My understanding regarding the EVGA debacle is that Nvidia sets really high MSRP along with charging for the chips & giving themselves a headstart and the partners can't go above the MSRP then Nvidia comes in and slashes prices that only Nvidia can compete at (since they can give themselves discounts on the chips along with all of the other advantages they have).
What board partners likely want is either: Nvidia is time-limited to a few months with Founder Cards or Nvidia allows them to sell cards for more money and Nvidia can't make any card except Founder.
GPU prices will trend towards market prices either way. I'd rather them somewhat overshoot MSRP rather than undershoot, which is what happened with the 30-series GPU. It's not fair that scalpers were getting rich while EVGA was operating at a loss.
I am pretty sure EVGA sold directly to miners, likely at a good margin: my day-one EVGA queue was never fullfilled, while miners showed warehouses full of EVGA cards.
Mine was, something like 9 months later. But I got my queue email & had 24 hours to buy. Maybe dig through your emails to see if you just missed it? Or maybe the model you waitlisted on was basically cancelled.
wonder if he's one of the people who clicked the "I don't want to convert my queue to the LHR model, and I'm aware EVGA expects all future production to be LHR so this queue probably will never be fulfilled" button.
For awhile some of the Titan lines though got major features unlocked from their workstation graphic cards. I specifically bought a Titan XP used after the last crypto crash in 2018 for 350$ because I needed 10 bit colour output to a special 10 bit capable monitor for video editing and colour correction. I probably would have bought a 1080 ti instead with a much nicer cooler (for overclocking) which is a pretty similar card but that card was locked out of these features.
While I expected about $1600 for 4090, for the other two, whose GPU sizes are 9.5/16 and 7.5/16 of the top model, I expected prices proportional to their sizes, i.e. about $1000 and $700.
However, NVIDIA added $200 for both smaller models making them $1200 and $900.
It is debatable whether 4090 is worth $1600, but in comparison with it the two 4080 models are grossly overpriced, taking into account their GPU sizes and their 2/3 and 1/2 memory sizes.
They will price the 4080 at $1200 to push people over the line for the 4090. After all, the performance difference for "just" $400 is quite significant.
They are probably working towards cleaning out the 30 series stock given that crypto demand is dying off. It takes a lot to put 3060 prices on a marketing slide for 40 series.
That sort of make sense... I was thinking of upgrading when the 40 series came out, but looking at these prices, it makes the 30 series look a lot more affordable for what I need, even if it's less performant.
Not with even 50% of the FLOPS. The 4090 won't be the best perf/$, but considering how much perf/$ and perf/watt you can get these days, I think it's pretty hard to complain that these are available. If you don't play games, get a Ryzen 5 7600X for $300 with an iGPU more than capable of high-quality desktop computing.
You can also get a brand new laptop for $130 right now https://www.bestbuy.com/site/asus-14-0-laptop-intel-celeron-... , which is $57.37 in 1990 dollars. (according to https://www.usinflationcalculator.com/ ). The high-end is more expensive than ever, and the low-end is cheaper than ever. The Apple Lisa was $10k in 1983, which is about $30k today. You can get a Boxx Apex with multiple $10k Quadro GV100 GPUs and 8TB SSDs for that money today, or roughly a dozen high-end (but not TOTL) gaming PCs.
Yea this is insane. Screw Nvidia and wait for the second hand market to hit rock bottom (I would estimate this will start to happen in 1 to 3 months) and buy a used card for dirt cheap.
they're setting prices high (and delaying the launch of the mainstream and lower-tier cards) because they have huge stockpiles of Ampere chips they need to burn through. Launch the stuff at the top to get the sales from the whales (VR, enthusiast gaming, etc) who are willing to pay for performance and delay the lower cards that would compete with your ampere inventory.
It's the Turing strategy all over again - when you have a stockpile of older cards, you don't make your new cards too attractive. And yes, they also have a big TSMC allocation too - but they gotta get rid of the Ampere stuff first, the longer it sits the more it will have to be marked down, launching new cards that obsolete the inventory they're trying to sell would just make things worse.
AMD is going to be doing the same thing - they pushed back the midrange Navi 33 and will be launching only the high-end Navi 31 until that miner inventory burns through a bit. Similarly to NVIDIA, that likely implies they'll be launching at a high price, they'll undercut NVIDIA by $100 or $200 or something and take the margins but they're not gonna be the heroes of the $700 market either.
---
The EVGA thing is a tempest in a teapot though, the losses he's talking about are something that's happened in the last month (he supposedly decided to quit back at the start of the year) and not representative of the (large, 10x normal) margins that board partners have been making in recent years. I personally didn't see much evidence of "price caps" with partner's first-party storefronts selling 3080s at 2.50-3x FE MSRP either.
And yes, jensen is an asshole and being a partner is a low-margin business, everyone already knows that.
EVGA is losing money because of some of it's CEO's ridiculous prestige side-projects (custom motherboards, enthusiast monitors, video capture cards that turned out to be falsely advertised, pcie sound cards, etc) and generous warranty support (long and transferable with absurdly cheap extended warranties) coupled with a higher-than-average failure rate (because they contract out assembly) and a generally lower-than-industry margin (because they contract out assembly) and they're just being drama queens on the way out.
Especially when they've been have lower than expected revenue because of the crypto crash. You'd think that they would drop the price a bit to entice the people who've been waiting for GPU prices to drop to buy the new version.
450 watt TDP? I feel like a crazy person every time a new generation of GPU comes out and raises the bar on power consumption and heat generation. How is this ok?
The 3090 Ti series was already pushing 450W so this isn't new.[1] And it's because they clock these things incredibly high, well beyond the efficiency curve where it makes sense. Because that's what gaming customers expect, basically. On the datacenter cards they quadruple or octuple the memory bus width, and they drop the clocks substantially and hit iso-perf with way, way better power. But those high-bandwidth memory interfaces are expensive and gaming workloads often can't saturate them, leaving the compute cores starved. So they instead ramp the hell out of the clocks and power delivery to make up for that and pump data into the cores as fast as possible on a narrow bus. That takes a lot of power and power usage isn't linear. It's just the nature of the market these things target. High compute needs, but low-ish memory/bus needs.
This isn't necessarily a bad or losing strategy, BTW, it just is what it is. Data paths are often very hot and don't scale linearly in many dimensions, just like power usage doesn't. Using smaller bus widths and improving bus clock in return is a very legitimate strategy to improve overall performance, it's just one of those tough tradeoffs.
Rule of thumb: take any flagship GPU and undervolt it by 30%, saving 30% power/heat dissipation, and you'll retain +90% of the performance in practice. My 3080 is nominally 320/350W but in practice I just cliff it to about 280W and it's perfectly OK in everything.
[1] Some people might even be positively surprised, since a lot of the "leaks" (bullshit rumors) were posting astronomically ridiculous numbers like 800W+ for the 4090 Ti, etc.
> The 3090 Ti series was already pushing 450W so this isn't new.[1]
Every time this comes up, people forget that:
1) Flagship cards like the 3090 Ti and 4090 consume significantly more power than typical cards. You should not buy a flagship card if power is a concern.
2) You don't have to buy the flagship card. Product lineups extend all the way down to tiny single-fan cards that fit inside of small cases and silent HTPCs. It's a range of products and they all have different power consumption.
3) You don't have to run any card at full power. It's trivial to turn the power limit down to a much lower number if you want. You can drop 1/3 of the power and lose only ~10-15% of the performance in many cases due to non-linear scaling.
4) The industry has already been shipping 450W TDP cards and it's fine.
If power is an issue, buy a lower power card. The high TDP cards are an option but that doesn't mean that every card consumes 450W
Yeah, I thought those leaks were pretty insane, like are we really going to all need 1KW+ PSUs to run flagship GPUs? I picked up an 850W Seasonic Prime a few upgrades ago and even running benchmarks on an OC 12900KS + 3090 with 8 case fans it's totally fine. I was hoping to not have to upgrade PSUs for another few years.
For what it's worth, you can power limit them and I'd highly recommend it if you plan on running a few of these. In the past we've power limited RTX 3090s to 250W (100W lower than original) while losing negligible amount of performance.
Do you just limit using something like MSI Afterburner?
I was undervolting and underclocking my Titan XP for a long time (with a separate curve for gaming) but lost the config and have been lazy to go back. I mainly did this because it has a mediocre cooler (blower design that is not nearly as good as the current ones but also not terrible) and even at low usage it was producing a lot of heat and making some noise (the rest of my system I specifically designed to have super low noise).
I just used `nvidia-smi -pl 250` on linux, it's built into the driver itself, I'm sure there's equivalent tools (or maybe even nvidia-smi itself) for Windows as well.
I'll also add that Nvidia's drivers are fairly good at keeping wattages relatively low, at least for desktop cards. My 3070 will idle ~10 watts when decoding video or just accelerating my desktop environment. You only see that 300-watt TDP take effect when doing consecutive renders or local ML like Stable Diffusion, and you'll definitely want the extra power.
The fan curves though kinda suck for the default nvidia drivers - especially on the blower it will run them super low to try to make it silent (which is not very efficient heat wise) and then not start spinning it up until it is really under load. Whereas there is a level that is barely an increase in perceptible noise that cools the GPU significantly more and for many mid range loads will not end up with with the GPU temp (and the fan) going through lots of highs and lows.
Interesting - going to look into this - do you set custom fan curves at all? Wondering basically if I can use a command like this to set a ceiling on wattage and then a second command to set an increased lower limit on fan speed.
Nope, I don't touch anything else. This is essentially telling the GPU, don't use more than this power at all, it'll automagically figure out the best configuration and all that. I've ran those power-limited rigs for 24/7 for many weeks, no issues at all and reduced heat is a nice side effect.
There's literally a power slider now. So yeah just fire up Afterburner, take the power slider, and lower it as much as you want until performance falls off too much for you. I think you can drop all the way to 50% or something somewhat asburdly low (at which point you should probably have just bought a cheaper, lower-power card in the first place)
What you describe is accurate, the 3090 Ti produces a tremendous amount of heat under load in my experience and I would expect the same with these new cards.
Dennard Scaling/MOSFET Scaling is over and it's starting to really bite. Power-per-transistor still goes down, but density is going up faster. Meaning an equal-sized chip on an old node vs a new node... the power goes up on the newer chip.
Physics is telling you that you need to let the chip "shrink" when you shrink. If you keep piling on more transistors (by keeping the chip the same size) then the power goes up. That's how it works now. If you make the chip even bigger... it goes up a lot. And NVIDIA is increasing transistor count by 2.6x here.
Efficiency (perf/w) is still going up significantly, but the chip also pulls more power on top of being more efficient. If that's not acceptable for your use-case, then you'll have to accept smaller chips and slower generational progress. The 4070 and 4060 will still exist if you absolutely don't want to go above 200W. Or you can buy the bigger chips and underclock them (setting a power limit is like two clicks) and run them in the efficiency sweet spot.
But, everyone always complains about "NVIDIA won't make big chips, why are they selling small chips at a big-chip price" and now they've finally gone and done a big chip on a modern node, and people are still finding reasons to complain about it. This is what a high-density 600mm2 chip on TSMC N5P running at competitive clockrates looks like, it's a property of the node and not anything in particular that NVIDIA has done here.
AMD's chips are on the same node and will be pretty spicy too - rumors are around 400W, for a slightly smaller chip. Again, TDP being more or less a property of the chip size and the node[0], that's what you'd expect. For a given library and frequency and assuming "average" transistor activity: transistor count determines die size, and die size determines TDP. You need to improve performance-per-transistor and that's no longer easy.
[0] an oversimplification ofc but still
That's the whole point of DLSS/XeSS/Streamline/potentially a DirectX API, get more performance-per-transistor by adding an accelerator unit which "punches above its weight" in some applicable task and pushes the perf/t curve upwards. But, people have whined nonstop about that since day 1 because using inference is a conspiracy from Big GPU to sell more tensor cores, or something, I guess. Surely there is some obvious solution to TAAU sample weighting that doesn't need inference, and it's just that every academic and programmer in the field has agreed not to talk about it for the last 20 years, right?
I'm not sure what else to expect? Is it so crazy that they make the card even faster and bigger than before, and it uses more power? What else is the next generation of cards supposed to do, have the same performance but make them more energy efficient? Not sure how many people would buy cards that have the same performance but less TDP.
In a generation or two we're going to start bumping up against regular US outlet limits. Gamers might start setting up their rigs in the kitchen to use the 20A outlets.
It is. I remember when I bought my GTX 260. It was crazy how much power hungry it was at the time for a mid market card. I had to buy a drumroll 450W PSU. Years later, I upgraded to a GTX 960. At first I thought "I probably need an even bigger PSU now...". I was shocked and then very happy to learn that the GTX 960 consumed less than my 7 generations older GPU. They can do it right when they care.
The GTX 260 was build on a 65 nm process, the 960 on a 28 nm process. This is one and a half full nodes apart, on similar planar processes with a full node really was a doubling of density. While the transistor count increased by more than that (1.4B to 5.2B), leading to a larger die size, it can't be overstated how much the improved process and decreased Vth helped power usage in this era. It would have been a challenge to use more power across those processes.
Believe me, chip designers care more about power now than they did then. Back in the larger process nodes when you moved to the next node the voltage scaled also, and as power is proportional to voltage squared, power went down. It was the ultimate free lunch, smaller, faster, lower power. That mostly ended around the 20nm node. Now designers work very hard to make things more power efficient, or stay within a thermal budget.
> What else is the next generation of cards supposed to do, have the same performance but make them more energy efficient? Not sure how many people would buy cards that have the same performance but less TDP.
i mean, the option is there depending on what your requirements are. i recently went from a 980ti to a 3070, doubling performance while decreasing TDP... as someone who has the card for casual use (only really make use of it a few times a month), i actually did just look for the fastest card that ran less hot than my current one (i've come to value quiet a bit more)
I actually did same just couple months ago, 980ti->3070. Though TDP claimed to be 220 but depend on vendor it seem it can peak to 250. Also one thing i liked after upgrade that 3070 have fan off when you aren't have load on it like desktop apps.
Power is a non issue for most people buying these cards for gaming. I couldn't care less if my desktop draws 500W or 1000W while gaming. If I get 144 fps vs 100 fps, I'll gladly accept the extra power consumption.
Not "just" energy prices (which is important!) but even things like your home. Pulling >1KW through a single power socket (or 1.5-2 KW once you account for CPU, monitor, etc.) is non-trivial for a lot of people.
At that level you're really in the territory where people will trip breakers in their home. This is "you have to hire an electrician and rewire part of your house" level of power consumption, which doesn't seem sustainable? Or at least "don't turn the stove on at the same time", which seems like an awfully disruptive way to ask people to live.
Heck, at that level for homes/cities with older electric service at lower amperages, the computer is now a significant fraction of the total service capacity of your connection to the city grid.
What's the response then? "Sorry you're not living in a new build home, forget about gaming"?
For me the wall-power isn't a huge issue but now you have to think about air conditioning. 1-2KW dissipating in a closed room heats the room up fast. This is already a problem with the RTX 2000-series that I'm running now and it seems like it'll be even worse.
"Glad you want to game, now please evaluate your home's electrical and HVAC systems to ensure they are compatible with a giant box pumping 2KW of power and converting it directly to heat"
Even if power is expensive, it only uses that power when you are using it. The whole card is 450W, so maybe 3 hours of intense could be a dollar? The card isn’t going to be running 100% for any game I know. Compared to a slower 250W card, the premium is 1/2 that. With that it’s like you agree to pay an extra dollar for 6 hours of gaming. With expensive electricity. I feel like this is conservative and doesn’t seem to bad for people considering the most expensive consumer card on the market.
I’m in southern Sweden, which is seeing stupid electricity prices at the moment. My current daily electricity price tops out at 6.62 SEK per kW/h, so a 450W card would cost me a US dollar for a four hour gaming session.
Southern Norway crossed the 10NOK/kWh threshold earlier this year.
It is kind of mind boggling how this can happen for a country with almost infinite supply of hydro power, but here we are.
Of course, Norway being Norway, since the start of September we now get a 90% refund for everything above 0.7NOK (based on the average price of each 24 hours) so for most of us will survive.
Oh, and since most power plants are owned by the public this doesn't affect budgets as badly as it could have done.
(It is even more complicated but I don't have more time now. Someone please fill in if I got something really wrong. Also, yes, this is a wild tangent but thinking I know HN I guess it will the day a little less boring for someone : )
(although in this case I’m not blaming my government too much, beyond failing to anticipate Germany’s need for our excess energy, and for not setting a country-wide pricing model for electricity. With Putin’s current stupidity, a lot of the previous energy planning goes out of the window, and all we can do is hold on for the ride.)
> The most expensive electricity is Germany at $0.40 / kwh
That’s not true, my contract is for 0.41 (EUR, but that’s kinda the same right now) and I’ve seen far higher prices. Unless you are talking wholesale, which might be true but is useless for a consumer comparison.
Assuming you don't care about energy prices, climate change and fan noise, there is still thermal throttling that can lead to cycles of boost-throttle-boost-throttle that can result in uneven frame pacing.
Of course my assumption is that the system will be cooled adequately, avoiding any thermal throttling at sustained maximum load. That's true for the current hardware generation too.
I'm sure that for gaming, power is not an issue, but I will continue to lament that Intel and Nvidia do not give a shit about making power-efficient designs. They push the increased cost and headaches to consumers and OEMs who need to spend time and money cooling their computers.
Maybe that's the genius of their plan? The chip designers continue to put out better & hotter processors and expect someone else to figure out cooling. Thermal design failures are not attributed to Intel and Nvidia, instead to OEMs and consumers.
I don't understand. If you are unable to cool it properly, you can under-clock it, or run less intensive programs on it, or just get a last gen card.
Offering a more powerful GPU/CPU for people and use-cases that demand it is not a problem. Most games these days run well even on mid-tier cards of the 20 series, that's two generations ago. So people have all the choice in the world.
Just because a 1000 bhp LaFerrari exists in the market, doesn't mean you cannot buy a ~100 bhp Corolla. People who buy the Ferrari will need to take special care of their car to get the most out of it. That's a pleasure in itself for some.
Your car analogy is missing the fuel aspect. If I can get 1000HP with 50MPG, instead of 1000HP at 5MPG, I would buy a Ferrari.
At the high-end, Nvidia is making more powerful cards (HP) with the same fuel efficiency as before. Maybe they are making more efficient cards at the lower end but those don’t get attention on HN it seems.
> At the high-end, Nvidia is making more powerful cards (HP) with the same fuel efficiency as before.
No, perf/w is still climbing every generation. It'll climb hugely this generation too.
NVIDIA's marketing number is 2x perf at the same wattage as last gen, obviously needs to be validated by third party testing but they're shrinking two nodes at once here, it's a bigger node jump than Pascal was.
People have been primed to react negatively by a couple of twitter rumormongers like kopite7kimi. Yeah 4090 has a high TGP, but it's more like 400W TGP (450W TBP) and not the 800W TGP / 900W TBP that he was shouting from the rooftops about. There is still a huge gain in efficiency here, it's just also a very large (and expensive) chip on top of the higher efficiency.
But people don’t discard their incorrect mental frames when the information used to build them is falsified. Actually they just tend to retrench and dig deeper (“450W is still too much even if it is 2x perf/w!”). Ada is a power hog, “everybody knows it”.
People said they wanted a balls-out big chip on a competitive node, that's exactly what the 4090 is, it's highly efficient and very large, this is much closer to the cutting edge of what the tech can deliver than Samsung 8nm junk or little baby 450mm2 dies. Now people are complaining about the TDP and the cost, even if it's more efficient that's not good enough. Unfortunately the TDP is a matter of physics, thermal density is going up every generation and a big chip on a modern node pulls a lot of power.
There are lower models in the stack too and those will maintain that efficiency benefit. It's a larger-than-pascal node shrink, efficiency is going up a lot here.
Would that inherently limits the customers/market though? The fps above all crowd would be happy but I would bet there's addressable people who would want 4080 performance in a small case that's relatively silent?
Except Apple doesn't really have their cake and eat it too. Their most powerful GPUs can be more than 10x slower than similarly priced Nvidia cards, particularly for ML and rendering workloads. At least in Nvidia's case, I think they've got their ducks in a row. Their 7nm GPUs get better performance-per-watt than Apple's 5nm ones. 'Nuff said.
> And nvida is just milking performance by power draw more current (less efficient)
You'll be hard pressed to find a generation where power efficiency actually decreased. That didn't happen with the RTX 30xx series, for example. Even though power skyrocketed, performance increased by more than power did. Meaning efficiency still increased. That's been true for every generation I can think of, certainly all the more recent ones anyway. Maybe AMD had a few like that where they just re-rev'd the same architecture multiple times in a row.
Not NVidia, Intel. Intel has been unable to do a die shrink for years because they own their own fabs instead of using TSMC, GlobalFoundries, or Samsung to actually produce their chips. The new 4090s are supposed to be about 2x performance of a 3090 with the same power draw because of the smaller process (5nm TSMC vs 8nm Samsung).
It's good to remember that you don't have to buy the most powerful GPU model just because you can afford it.
Some people are probably in the target audience for the 4090. Others may prefer the 4080 models, which have a slightly lower TDP than the 3080 models but still get a nice performance boost from much higher clock rates.
I wonder when Nvidia start selling lower powered model like Intel "T" CPU. It's just underclocked/undervolted and a bit binned chip, but some consumer like it. EnergyStar also will like it.
They bought out exclusive access to the most efficient node, making it impossible for others rather than showing it is possible for them (though it presumably would have been possible if they outbid Apple at the time).
Running games that put my GPU (rtx 3080) usage at 90-100% doesn't even make my PC audible when having (open back) headphones on with ingame sound a normal levels. If you invest in good components and focus on quietness you can get great performance without much noise. In my case, without any watercooling.
With watercooling and a big rad, it can get even quieter.
yes, there are limits to how much heat can be removed by a dual-slot blower without making it extremely loud. open coolers can do a bit better, with the trade-off that they dump most of the heat inside your chassis rather than out the back.
I have bigger things to worry about personally as far as my energy consumption is concerned, but the main way this impacts me is that my wife and I probably need to get some electrical work done in order to support our devices being in the same room. Our two gaming computer towers theoretically have a maximum combined power draw of 1800 Watts, which is well above the 1450 W which I think is the maximum sustained (not burst) load for a 15 A residential circuit in the US. This is ignoring the few hundred extra watts for monitors, speakers, laptops, and a couple lamps.
My understanding is that the maximum should only be used for short periods of time, like for microwaves and hair dryers, rather than for sustained loads like a heater or computer. I think sustained loads are supposed to be 80% of the maximum load the breaker can handle.
NVIDIA has unleashed its next-gen GeForce RTX 4080 series graphics cards that come in 16 GB & 12 GB flavors at $1199 & $899 US pricing.
Welp it looks like I'm buying a Series X because fuck that noise. PC gamers get put through hell for the last two years because Nvidia preferred selling to Chinese crypto miners and this is our reward?
As a PC gamer who's slowly losing interest in favor of console gaming...
I don't mean to sound ignorant, but how is any GPU comparable to a console.
For $500, I can get: a Series X and a controller, ready to play out of the box. With GamePass (currently $1 first month), I get instant access to a large library of games.
A current GPU (20, 30, or 40 -series) costs more than that, and I also need to purchase the additional components to put it all together.
I built a PC in college because it used to be the most cost effective option. But that doesn't seem to be the case anymore. These 40-series prices will define the cost for the next generation of cards, and IMO its too much to justify for casual gaming.
The PC is for high end gaming. It's not cost effective (contrary to what you said I'm not sure it's ever been cost effective[1]) compared to consoles. It typically only gets close to consoles at the end of a console cycle because they have such a step-function and the pc has a far more granular cost/performance curve over time.
PC is really the only place where you get super high end framerates+resolution+graphics-fidelity combinations. It's only really cost effective if you upgrade slowly—a new cpu here, a new GPU there, amortizing the costs over the years. And the nvidia xx80/xx90 models are never the ones you'd buy if you are trying to be cost-effective. They are high end damn-the-money-just-give-me-all-the-graphics cards.
[1] maybe back in the era where the PS3 was $600, but even then a xbox 360 was cheaper and mostly just as good fidelity-wise.
It used to be that you could assume that you would need a PC of some kind, for work/email/homework whatever, so upgrading the pc so it could do games made sense.
Roughly the same time as the 360 era I built a PC for uni, and the extra cost to get console-beating parts (and adding in that PC games were cheaper or pirated) was way better value.
Now Im not sure what I would do in the same position, a laptop and console costs significantly less than a PC that even matches it, and unis mostly stopped having any focus on CAD
> A current GPU (20, 30, or 40 -series) costs more than that, and I also need to purchase the additional components to put it all together.
No it doesn't. An RTX 3060 brand new is under $500. So is a 6700 XT. If we're talking used, RTX 2060's are $200.
Now, are consoles cheaper if all you care about is playing games and nothing else? Sure. But that's always been true. Usually the pro-PC argument is either because you want the flexiblity of PC gaming (that is, your games last longer, you've got things like steam sales, mod support, you want better-than-console quality & performance, etc...), or you already were going to have a PC at which point take the money you would have spent on a console and just add it to the PC in the form of a GPU. In a world of increasingly fewer PCs as a household item, being replaced with laptops & phones, that's less true than it used to be. But ymmv.
> With GamePass (currently $1 first month), I get instant access to a large library of games.
Yeah, but as far as I know that requires making your Windows login a Microsoft account, and frankly I'd rather risk giving piracy malware. I already pay for Game Pass Ultimate but I've long since excised Microsoft's tentacles from my desktop.
Not sure what you're asking here? Consoles these days use midrange AMD GPUs. Getting a comparable rig is price wise perhaps a 100-200$ of price difference for (arguably) more flexibility.
Can you show me a build that you can make from off-the-shelf parts today for $700 that is comparable to the 6700XT in the Series X, without carrying over existing parts?
Don't forget console optimization, at least for triple-A games, makes the hardware punch well above its weight class compared to the PC version. Halo 5 on the Xbox One looks about as good as anything I run on my GTX 1070 despite it having the hardware equivalent of a GTX 750.
Kinda mostly not a thing anymore. DX12 / Vulkan closed up the last remaining notable gap here. Particularly as consoles became more PC-like in OS-level restrictions to combat piracy, and PCs added more low-level interfaces to improve performance.
Console optimizations these days is mostly just dialing in the settings & using liberal amounts of up-scaling (which the PC versions also often support, but is disabled by default)
For perspective, a PS5 is about the performance of a GTX 1070, give or take. So a 3080/4080 are far far better performance than that.
If you already have PC parts, you could potentially upgrade to a 3060Ti for $400 (minus whatever you can sell your current GPU for) and get better perf than a console.
Otherwise, you're right. Starting from scratch, consoles are a better value, unless you really really want the best performance.
Sure, but with console optimization the series x is roughly equivalent to a 3070 in real-world performance and I get the value-add of a controller, a UHD blu-ray drive, and compatibility with my library of 360 games. My GTX 1070 already plays 90% of my game library at the highest details and framerate on my 60fps TV. Also, never having to deal with Steam's screwed-up gamepad framework again is probably worth two hundred bucks in frustration alone.
NB: 4080 12GiB has 9720 cuda cores and 2.21/2.51 base/boost clock (in GHz), whereas the 8GiB version has 7680 and 2.31/2.61 (in GHz).
In other words, the name is utter bollocks because the 12GiB model has approximately 1.26 times the cuda cores of the 8GiB version for approx 0.96 times the core clock. Assuming no other bottleneck exists, the 12GiB version should be 1.21 times faster. This is supposed to be the xx70 and xx80 difference.
But then they'll also create a 4070 here. And then a 4070 ti, and 4080 ti, and then.... But that's another complaint. The 4080 is a different problem entirely - not enough names. Somehow they have too many names but also not enough so they can only differentiate via GB.
Their Performance section shows that the 12GB is 50%-150% faster than the 3080 Ti (so better than 3090 Ti!), and it also has more cuda cores and memory than the 3070.
Yea but that's how new tech works. If the price of every new generation of hardware scaled linearily with the increase in performance we'd be needing mortgages for a new PC/GPU.
That's how new tech worked in the Moore's Law era. Times are changing and you're observing the evidence of that.
And yes moore's law notionally died a long time ago, but it's been able to be hidden in various ways... and things are getting past the point where it can be hidden anymore.
Node costs are way way up these days, TSMC is expensive as hell and everyone else is at least 1 node behind. Samsung 5nm is worse than TSMC N7 for instance, and TSMC N7 is 5+ years old at this point.
Samsung 8nm (a 10+ node) used for Ampere was cheap as hell, that's why they did it, same for Turing, 12FFN is a rebranded 16nm non-plus with a larger reticle. People booed the low-cost offering and wanted something on a competitive, performant, efficient node... and now NVIDIA hopped to a very advanced customized N5P node (4N - not to be confused with TSMC N4, it is not N4, it is N5P with a small additional optical shrink etc) and people are back to whining about the cost. If they keep cost and power down by making the die size smaller... people whine they're stonewalling progress / selling small chips at big chip prices.
Finally we get a big chip on a competitive node, this is much closer to the edge of what the tech can deliver, and... people whine about the TDP and the cost.
Sadly, you don't get the competitive node, giant chip, and low prices all at the same time, that's not how the world works. Pick any two.
AMD is making some progress at shuffling the memory controller off to separate dies, which helps, but generally they're probably going to be fairly expensive too... they're launching the top of the stack first which is a similar Turing-style "capture the whales while the inventory of last-gen cards and miner inventory sells through" strategy. That likely implies decently high prices too.
There's nothing that can really be done about TDP either - thermal density is creeping up on these newer nodes too, and when you make a big chip at reasonable clocks the power gets high these days. AMD is expected to come in at >400W TBP on their products too.
I think typically the XX70 has been in line with the xx90 TI of the last generation, so this is in fact a better improvement over that. It's still dumb though to have two xx80's.
Compared to the 3090, the 4090 has about 60% more CUDA cores (16k vs 10k), runs at ~2.2GHz (up from 1.4GHz) and eats about an extra 100W of power.
Over the last couple weeks, it's been possible to get 3090's on sale for juuuust under $1k (I picked one up after watching the prices come down from like $3k over the last couple years). The 4090 is priced at $1500... (and for me at least would require a new power supply.)
RTX 3090 was $1500 at launch, and was a 50% boost over the previous gen at best.
RTX 4090 is $1700 an is asserted to be ~1.9x performance, when taken with salt grain.
You can buy a 3090 for $999 today at Best Buy. That's not a bad price per frame.
$1700 for the 4090, if 1.9x speed for raster, is not a bad price per frame either.
Both are stupid. Anyone spending $2k for video games graphics cards is wasting money in a stupid way: they are big, hot, shipped from far away, and turn into toxic landfill garbage in a matter of years. They mark the worst form of destructive consumerism that late-stage capitalism has produced, on par with leasing a new vehicle every 2 years: both are bad as products that turn into needless waste and both consume energy and produce unnecessary heat.
I am definitely going to buy a 4090 on October 12.
Intel has moved to DisplayPort 2.0, but their current product specifications, e.g. for NUC12SNK, say: "DisplayPort 2.0 (1.4 certified)".
But I assume that they will get a certification in the future.
For Thunderbolt, where it was Intel who did the certifications, many companies have sold motherboards or computers for months or years with non-certified Thunderbolt ports which worked OK, while waiting for the certification.
I’ve been working with NVIDIA GPU VMs of late with the intention of running GPU enabled TensorFlow and PyTorch. I find working with NVIDIA software incredibly frustrating. In particular, installation instructions for CUDA / cuDNN simply do not work. The biggest culprit is missing or incorrectly versioned shared object libraries. Forums are loaded with developers having similar problems never with any viable solution. I’ve had better luck with NVIDIA distributed docker containers. There at least I can make it as far as running GPU enabled TensorFlow sample code.
I was having this same issue until I just got into the habit of using a conda env. There are community conda repos that make getting all versions of needed libs much simpler (condaforge is one).
Then it’s just a matter of managing envs for various projects. Smooth sailing.
Nvidia realized they can charge whatever they want and are taking full advantage of that. Hopefully Intel, AMD and used cards are able to bring them down a peg.
I'll be on the sidelines waiting to see how this turns out.
When there was a monthly income from mining and selling ETH, with payback that could be counted in a few months, insane GPU prices made sense for miners (or combo gamer/miners.)
Without that, it's down to gamers with high discretionary income, and how well alternatives are priced. For example, if you can get an RTX 3080 or 3090 for less than $600, you might not consider the RTX 4080 or higher. If AMD releases competitive parts and undercuts their prices (I'd bet on this), these parts will likely take longer to sell, and drop in price.
Almost certainly, with Nvidia's overstock of RTX 3000 cards, they simply want the RTX 3000 to look good in comparison, and get sold, while they control the supply of RTX 4000 cards (perhaps only producing a relative trickle until the RTX 3000 are gone.) Then they could adjust prices to meet the market, and again the RTX 4000 "lower" prices will seem better (if only in comparison to the previous prices.)
(Disclaimer: I paid $800 for an AMD Radeon RX 6700 XT 12GB, and mined enough ETH to pay it off, and then stopped mining. I wanted it for gaming, but I did not think the price was reasonable.)
> Hopefully Intel, AMD and used cards are able to bring them down a peg.
Really? RX 6650 XTs are already in the same price range as RTX 3050s. Nothing has changed. Nothing will change in the short term unless AMD does something really bad for their business.
I think you should wait until late November for dust to settle(budget segment won't change that much because of it if you don't consider buying used). By then both Nvidia and AMD will release their cards(it is likely that Intel too, but Alchemist is a joke), mining cards will hit the market in large quantities and pricing for everything will change drastically.
Depending on how much you want to spend, a 1080ti, 2080ti and 3080ti are all good upgrades (roughly doubling in price with each step). All of them come with 11GB or more memory, making them very usable for not only gaming but also "average" ML (e.g. stable diffusion runs great on my 2080ti, and it's very usable for developing tensorflow models).
It is hard to tell since you don't provide exact numbers. I've seen SD generating 50 samples 512x512 image on 2080 Ti in 8 seconds, but it is almost certain that there is a faster version out there.
I just tried it, on my 2080ti I get a 50 sample 512x512 image in 7.0 to 7.5 seconds (using PLMS, DDIM or k_lms). Or at 256x256 with 8-image batches about 1.15 seconds per image.
A used 2080TI goes for around $500, with the remaining $500 I can get a computer with a 6x3.6GHz Ryzen 5 and 16GB RAM. Which is something completely different than a Macbook Air, so that's kind of an apples to oranges comparison, but I don't see the big price advantage. And presumably GP already has the "supporting platform", since they are upgrading, making the 2080TI upgrade half the price of a switch to an m1?
It is VASTLY slower than 2080Ti. There is no competition at all. I don't know exact performance numbers, but I would estimate that M1 takes more than a minute to generate one image.
UPD: asked a friend to generate image on m1 with diffusion bee. It took 4 minutes 30 seconds to generate one 25 sample image. It is not even close and much worse than I thought.
Could you provide the exact settings you use for generating the image? Settings matter a lot in terms of it/s. Also please share the exact it/s you're getting :)
Stable Diffusion is but 1 project. M1 gonna be a huge bottleneck for anything that takes time. That "Ultra" for $4k is about 5 times slower than a single 3090. It will be nearly 10 times slower than 4090.
I upgraded from a 970 to a 2080 a while back and have been very happy with it. Still plays modern games at good frame rates, though not with max settings.
Obviously your choice depends greatly on budget + availability of hardware, as well as your desired resolution, but I'd suggest the 2070 or 2080 (maybe 2080ti?) as a good starting point.
I upgraded my wife's 2015 workstation from 980ti to 2080sc of some sorts. The limit is pci bandwidth of the MB, anything later made no sense. My kids use that as gaming PC and the happiness index on the latest titles (valorant, halo) went way up. It cand compare to the 3080 that I have in the latest gaming PC, but it does great. Bonus is that I sold the 980 to someone for 70 dollars, I was going to trash it but apparently it is still worth something
I have a 980, too, and it's great. It plays most stuff on Steam without any problems. I'm only considering an upgrade because I want to play Flight Simulator and Cyberpunk on max graphics. I'll be getting a 3080-- they're fairly mainstream and a great value. I love ultra-quiet hardware so I'll be getting the Asus 3080 Noctua, which is the quietest version by a huge margin.
I think your local prices matter the most. Here in Hungary, right at the moment, looking at one specific large retailer, the 2060 makes the most sense, if we look at price/value. Compared to the 980, it offers <50% extra speed. It really depends on your circumstances, to judge whether that works for you or not.
I've considered this, but I'm not sure. Is there any concern about cards used for mining being too "beat up"? They say you shouldn't buy a used car from someone you know was using it for racing. Does that apply to graphics cards used for mining?
Apparently mining cards are ok because they suffer a constant load, sometimes run at a capped peak performance (?), vs games which can be very spikey.
Whether this is just misdirection thrown around by crypto miners looking at a hole of sunk assets or not, I could not say. Personally I'd just buy new, but I have also probably never spent over 300usd on a card.
My previous card (Radeon 7970HD) was used for BTC mining in a gaming PC and lasted until about 2018 (So about 6 years). My current GPU is a 1080Ti that used to be in a mining rig and shows no signs of damage at about 4 years of age.
I'm looking forward to buying a used crypto mining card when the current one dies.
The major concerns would be the thermal paste degrading or the fan dying. Both are easy and cheap to remedy, and are probably worth preventively fixing when you buy the card.
perhaps a 3080 , maybe 4080 12gb when the price drops a bit ? I'm envious I went form a 970 to 1660ti I'm not sure what my next move is Im usually more comfortable finance wise with 70s than 80s
10-20% faster isn't a lot of future proofing. Certainly not worth paying double.
You certainly don't need a 3090 for gaming in 4K. nVidia's claim of the 3090 being a "4K" card is marketing bullshit. If a game can't be run in 4K on a 3080, then the 10-20% performance increase of a 3090 isn't likely to change that.
Today is a gaming product announcement and I'm not sure that games have a need even beyond 12 GB today. I suspect that the 4090 price not budging much is telling as far as what ram amount demand has been focused on for games.
I assume they will soon have a professional card announcement that includes 48GB+ cards. Assuming that the high ram cards have improvements similar to this generational leap in the gaming market, they will be in high demand.
Sure, but you dont buy a high end GPU and then buy the best one next generation... I mean some do sure but a high end GPU like this you keep for a few years...
I believe part of the problem is they went with 24 GB gaming cards a couple years ago and the need for that didn't materialize. Hence, cold feet for launching with 48 GB because there still isn't a need for more than 12 GB and 48 GB impinges on their high margin professional market.
Seems they don't want to eat up market share of their "professional" cards with the consumer cards. Want above 24GB? Better put up money for the data center cards.
At least in the games I play usually texture detail and screen resolution are two independent settings (and if you play windowed it can be any arbitrary resolution), I am mostly a PC gamer though, not sure what console games do.
You can absolutely get away with 12gb, 10gb, even 8gb vram and play 4k high refresh rate.
But also doesn't account for poor programming. I remember a reddit post about God of War using 15gb of vram after a while of playing, probably a memory leak.
Having available vram just gives yourself headroom for newer games with higher quality textures, there was a game which name eludes me which shipped with 8k resolution textures for 4k gaming so it used like 10gb of vram? Or something along those lines.
But it can also be used for pre-caching stuff, like if you were playing WoW and the next zone was already precached so you don't need to load it from the disk.
I think that new tech Sony and MS were working on, direct storage access? Will play a bigger and better role than higher vram.
I only game at 1440p so I don't care for more than 8gb of vram.
Guess A6000 is in the middle of those two options, comes with 48GB of memory. Not so cheap if you're just using it for gaming, but if you're using your GPU for professional work, could be worth the price.
Does VRChat count as a game? In a very populated instance with 50+ people it can take up to about 20gb of vram.
Honestly everything VR takes up huge amounts of vram, it's not possible to run No Man's Sky at max settings with only 8gb vram, and you still might need to dial down some settings with 12gb vram.
This is with a Quest 2, I'm guessing headsets with higher resolution will be even more demanding.
> €5500 compared to €1200 for just double the memory at the price of less performance is not really worth it in my opinion.
It's not just double the memory, is faster memory as well. Although, less CUDA cores and also doesn't include latest NVENC encoder/decoder, nor AV1 encoder, so... Tradeoffs I guess :) For some people it will be worth it, for others no.
Well, yeah, broski. GPUs are just highly optimized matrix multipliers, so they're for any computation involving matrix multiplication...such computations arise in computer graphics, crypto mining, and machine learning (aka AI shit). And yes, memory size is important for some applications. Is that okay?
this simplification to “everything needs matrix multiply and GPUs are matrix multipliers” doesn’t really match the actual architecture.
the hardware for low-precision matrix multiplications in ML is mostly independent of the hardware that’s typically utilized for vertex and fragment shaders in computer graphics. It really only gets used for graphics if using DLSS. It’s not used at all for (integer) computations in cryptomining.
Are they for high-precision scientific simulations because it’s possible to multiply matrices on them? These cards are probably quite shit at double-precision matrix multiplications.
Is it though? I guess it is for machine learning. But do games require that much memory? Remember that GeForce is the consumer line. I would expect it to mainly target the gamer market.
Realistically speaking, the 3090 targets enthusiasts and hobbyist ML work. There’s not much need for a card of that size for gaming, but there’s definitely room for a card that can do ML and gaming.
I do much more inference than training, but a bit of both.
Speed is generally not a serious issue:
- I'm not trying to scale to millions of users on inference; it's a dev machine
- Training is a bit of fine-tuning; I'm not building big models from scratch.
- Worst-case speed has gone into "annoying" territory, but never into prohibitive. Worst case, I can grab a coffee while things run.
- It's not a lot of coffee, since I'm trying to visualize some interesting things about some data. The majority of that is data cleaning, visualization, data engineering, UI/UX, meetings, and a million other things. My GPU is sitting idle >>95% of the time.
I am constrained by RAM. In the era of GPT2 -> GPT3, Stable Diffusion, etc., RAM becomes a major bottleneck. There are cloud workflows, but they aren't a great fit, both due to data security / regulatory issues, to how we plan to do dev ops / deployment, as well as just to time for integration.
Dropping $8k on 4x 24GB GPUs would be a no-brainer if it let me work with bigger models.
I'm sure people can describe Better Ways to Do Things, but changing workflows, system architecture, etc. to those workflows would be obviously out-of-scope for what I'm trying to do. At that point, I'm better off working with smaller models. They're good enough.
As a streamer AV1 encode support is really exciting to me; when I can enable in OBS and stream to Twitch etc getting an Ada GPU would be high on my list of upgrades.
Does AV1 have much better compression than HEVC? I use NVENC for my entire desktop with virtual resolution of about 6k x 2.5k in real time and barely feel anything.
The Portal 2 image comparison seems pretty dishonest to me.
You can clearly see the graphics details have been reduced in the RTX off version. Check how the outline of the portal or the gun look like for example, completely different textures.
I mean it’s getting there. Add this to a best in class desktop processor at 300W and add in all the other power consumption and you are close to the wattage of a hot plate, kettle or small space heater.
The 30xx cards needs a wide bus to feed it, but with the L2 cache on the GPU, you can get away with a smaller bus width (and faster memory on top, things shifted in the last couple of years and DDR6(X) memory is even faster now than it was).
Some of this also has to do with what kind of memory configurations you want to offer, as they are tied to the bus width. So if you widen the bus, you are looking at using more memory chips and having more memory. This pushes the price point upward, and you want it to tier somewhere in the middle of it all.
One "mistake" on the 30xx (Ampere) series were that the 3080 was much too powerful for its price point, and you couldn't really cram too much more power out above it due to its memory config. With this change, you can introduce a 4080Ti in between the 4080 16Gb and the 4090 as a mid-seasonal upgrade, and widen the bus a bit to satisfy it.
This looks like a smaller bump than Ampere. 3090 seems cheaper and better than 4080. As people have pointed out, what I'd really like is more / upgradeable RAM. NVidia is trying to not compete with their higher-end cards, but ML tools are increasingly consumer.
If it’s not visually distinguishable, it won’t matter. I hear recent versions of DLSS are very good at 4k despite earlier versions having some bad artifacts. I guess we’ll have to wait for the reviews :)
I guess it depends on your definition if distinguishable, but as someone who has dabbled with most ML frame interpolation techniques, this appears just as artifact ridden.
https://i.imgur.com/9U9HIpH.jpeg
You do realize that it’s about 350 euro in added 22% VAT, right? That makes it about $/€50 different. You expecting a massive discount or something? I’m not saying euro pricing isn’t usual higher, but everyone always forgets the tax.
The tax is literally the problem with quoting the price in Euros.
As far as the price, the equivalent card from the 3 series was the 3090 RTX, which released at $1499 msrp 2 years ago. Given the inflation over past few years, $100 increase is not an increase in constant dollars. This new top of the line card has twice the performance for effectively the same price.
Yes these top of the line cards are crazy expensive, but it doesn’t seem like this is actually worse than the 3 series pricing. You are talking about more than 6 Xbox’s worth of compute here for an ultra enthusiast market.
It's also a problem by USD oriented pricing. It tend to be done by US based company like Apple and Nvidia. I found new Samsung Galaxy price isn't looks so bad unlike them because they are Korean.
Yeah, this was remarkably irritating for me here in Austria - if you're going to change the language based on location, please let the user override it.
Luckily 'en-gb' is less likely to be overridden than 'en-us'.
With Ethereum's transition to PoS and the crypto crash, the only demand bottlenecks I can foresee are scalpers and hungry gamers. Indeed, we may see a repeat of what happened with the PS5, when finding a product from a store was impossible because scalpers had bought them all.
Could be. Here in Italy is still hard to buy a PS5. I was smart enough to pre-order mine and score other 2 consoles (for friends, not scalping), but after 2 years that's ridiculous. They're not even on display in shops.
This might sound crazy but would it make more sense for me to buy a Mac Studio instead of the new GTX40 series if I just want to dabble in some fun ML like stable diffusion? I don’t plan on training anything crazy on my own, even if I do, it’ll be fine tuning rather than scratch.
No, go pick up a 3090 Ti from Best Buy instead. It's $999 right now -- over half off, maybe more with this announcement -- and has 24GB of VRAM, and can chew through Stable Diffusion requests in seconds versus dozens of seconds or minutes for alternatives. And because it's an nvidia card, you'll be in a much better position to troubleshoot problems or whatever if you want to retune/retrain models or get your feet wet. Or, you know. Shove it into another PC and play a video game.
That said the Mac Studio is a pretty badass machine. Don't let me dissuade you from one if you just want it. But it's a worse choice if you just want to dabble in ML stuff, IMO. Staying with the well trodden road will help you more when you're new to it all.
Does regular RAM matter if I want to run stable diffusion? I only have 16GB RAM (but obviously upgradable). If I load the weights onto the VRAM, does it work SSD > RAM > VRAM or does it bypass the RAM?
Yes, Stable Diffusion performance will improve the more VRAM you give it. This is generally true of almost every ML model, because ML models are often limited by bandwidth. There is a constant need to shove more data into the accelerator, but PCIe is very slow compared to data in VRAM. If you constantly had to push data over PCIe, your GPU would run at 5% of its max performance, and spend 95% of its time waiting on PCIe transfers. Instead, you load a bunch of data over PCIE at once into VRAM, as much as possible, and then you begin working. So the more VRAM you have, the more data you have "in hand" -- so you can have a bigger batch size for your training iterations, for example. So one of the things various releases of Stable Diffusion have tried to optimize, for example, is VRAM usage, because that's the #1 limiting factor for most people.
Modern GPUs use virtual memory so they can see very large amounts of data and the GPU and OS will work together to page data in and out, like normal RAM. But if you want good performance, you need to work out of VRAM, so you need as much of it as possible, so you aren't paging data in and out. (More specifically to answer your question, yes it will page data from SSD into RAM and then into VRAM; this is because the data must pass through the memory controller in practice on most designs to go over PCIe, but if you are using fancier Tesla/Quadro cards, they can actually do P2P DMA transfers and go directly from SSD -> GPU, but this requires special care and support from all pieces of hardware inbetween.)
Note that in a funny twist of fate, the M1 series from Apple kind of has an advantage here. They use very high-bandwidth LPDDR5X for their designs, and the GPU and CPU share that. Which means that in theory, the Mac Studio for example could have astronomically large amounts of VRAM dedicated to a task like ML. So why not use a Mac? Because the hardware is only good as the software you run it on, and training/tuning ML software on Macs is simply nowhere near as well polished as Linux/Nvidia. That's all it comes down to. And I say "well polished" very liberally; in practice it's all a bit of a hodge-podge nightmare because the job is very hodge-podge...
The model checkpoint file (i.e. weights) being used is read from disk, rather than from system memory afaik. 16GB of system RAM is more than sufficient.
It's been a while since I've updated my GPU (I have an RTX 2060 XC Black), so I'm out of the loop: what's the gpu market like now? I'm assuming we're seeing less fewer GPUs being purchased for crypto. Are prices normal now?
MSRPs for a given market segment seem to have permanently increased significantly. This generation is expected to be available at this higher MSRP, however.
No, the prices are still pretty high (although not insane anymore). They're probably not going down.
On the other hand, these GPUs are very very good for gaming - with DLSS, most of modern games will run at high details at 4K and look like something that scifi shows dreamed of just 10 years back.
Was seriously considering buying a used A100, just because it had >16GB RAM and was built for model training. Considering this new 4090 has 24GB RAM and is probably faster and cheaper than a used V100, I'll be picking one up for sure.
I am mystified at how many comments here are shocked at the price; the 4090 exactly in line with recent NVIDIA pricing. The 4080s do seem a little strangely overpriced, but not by huge margins.
We’re not shocked that the prices are in line with their recent pricing strategy. We are shocked that they haven’t decided to change course to make up for the past lack of availability, crazy power consumption, and card prices soaring through the roof.
I am just hopeful that I will be able to buy my first new video card in over 10 years.
Last year I was bored in the hospital with little to do, so I decided it was time to build a new PC after upgrading my last build from 2009 for many years. I was completely out of touch with the parts buying experience and the shortages/price-hikes. I was genuinely baffled when I found out I could not buy a video card from any PC part store. I managed to snag a 2070 super secondhand for a not ridiculous price, and I was lucky for it.
I've waited a long time and I'm just excited that I might be able to get a 4070 some time in the new year.
Disappointed that VRAM isn't going up. May have to pick up server cards and rent colo for some of the things I'd like to do. Or I guess the 4090 TI is rumored to have 48GB so I'll wait and see.
I wait for inevitable price drop when first adopters have gotten theirs and miners don't prop up demand anymore... I just don't believe these price levels stay.
I’m on an 1070 right now and feels like it’s time for upgrading but then I need a new CPU too so a new mobo as well etc. Oh well but always exciting to get a new PC!
Hooray, can't wait to buy a 3090 once the prices fall even further than they already have and completely ignore these new, overpriced cards that draw half a kilowatt!
I'm more excited about the Hopper server-class cards — not because I'll be using them, but because now Stability can swap their A100s for H100s and train more Stable Diffusion models for me to mess around with!
For a while I was looking ahead to getting a used card from a miner but I am stuck on the horns of this dilemma: NVIDIA is very quick to roll out new features. It’s been a while since I’ve done AI work but there is software out there that won’t run on the two 10-series cards I used and I’m worried a 30-series might have trouble running models that come out a year from now.
It’s actually a 2x performance per watt increase. The 450w power consumption of the top card is the same as the previous generation’s top card, 3090ti.
Why decrease overall power consumption of the top of the line cards? PSUs and cooling can handle it and it’s for the customer that wants top performance. There are plenty of other cards in the line at lower total wattage with a good performance per watt.
What’s the point in upgrading? Maybe I’m just getting old, but todays games dont look much better or than what i played 5 years ago. And they have these kooky monetization strategies now. Looks like gamers are paying a premium to enjoy derivative or scummy games.
Everyone has different requirements for what their computer should do. If I want to play flight simulator at 4k at high framerates, or VR, etc there is still room for faster cards.
It's not for everyone, but that doesn't mean that games have remained the same for 5 years, or that there's no reason to upgrade. If you don't want to, go ahead, but there is definitely appetite for faster cards.
3090 stays on PCIe 4.0, it's rational decision. I wonder why Intel and AMD implements PCIe 5.0 slot and M/B manufacturers had only one 5.0 x16 slot. It's really useless to have (even in foressable future), add costs, and not flexible
New cars, new smartphones, new clothes, new GPU's... every year. Are we really in climate crisis of this neverending rat race? Or is it alarmism a new PR strategy that supports consumerism?
Just because they are updating GPUs every year doesnt mean you have to update your GPU every year. Same with cars, smartphones, clothes.. Just because someone else is in different spot of their update cycle doesnt mean everyones cycle should be the same and synchronized
Sure, I don't have to and I won't. But production, advertisement and hype is still there. Our sources are not infinite, right? Why aren't we focus more on quality?
I have had a hard time feeling like it would have been worthwhile to upgrade through all these generations from my current 1080. Am I just no longer an enthusiast? Am I getting old?
I'm running a 1080Ti I picked up used in early 2020 and threw on a kraken g12 with an AIO, it dropped temps down to a max of 61c and increased FPS on most games by 30%.
Not for me, but for Microsoft and all those that would want to play that game now and specially in the future when a KWh becomes more expensive than $1.
> Until the mid-1990s, Santa Clara, Calif.-based Intel internally named its processors after cartoons, dead musicians and other well-known monikers. Attorneys for the late rocker Frank Zappa helped put an end to that practice after reaching out to the company about a potential trademark violation, said Intel spokesman Bill Calder. The legal team for the world’s largest computer-chip maker subsequently ordered that any new product code names be based on “existing, non-trademarked places in North America that can be found on a map,” Calder said.
> After struggling to find a suitable or available name in Oregon, where much of the chip design team is based, he stumbled upon Haswell in his ZIP code database.
> “I didn’t know anything about Haswell other than that it’s a good name,” Hampsten said. “Another reason I liked it, it had a suffix — well — which I could use for various other names.”
Because that's the way they have named their architectures since... For as long as I can remember. Previous generations include: Ampere, Volta, Turing, Kepler, Fermi and more.
To be pedantic, it was units of temperature at first, with a well-chosen segue of Curie to the superset of other names.
This does mark a departure from a convention of only single surnames, and one could surmise as to why that decision was made. Though one might also consult a Falsehoods list, since 'Ada Lovelace' was never $first_name $last_name to begin with, having been derived from something like /(Augusta Ada|Ada Augusta) King \(née Byron\), Countess of Lovelace/
After moving to the Steam deck, I'm not sure I care about my PC GPU anymore.
The combination of the Steam Deck w/ Xbox Cloud Gaming, has changed how I will move forward with PC gaming.
I think in the future, I'd want to use steam deck w/ an external GPU when driving my large monitor, and then cloud streaming, SoC gpu for handheld mode.
> When the cloud providers electricity contract runs out the cloud service will cost 10x.
Location matters greatly here. There are tons of data centers in my area due to stable electricity prices due to a high nuclear base generation capacity. I paid the same ten years ago as I do now.
Ok, interesting in what country are you if you don't mind the question? I'm guessing US?
I think marginal prices are going to shuffle the board pretty harshly from now on though!
Energy does not create itself out of thin air. (except from the sun, and that takes millions of years to get stored as coal, oil and gas) so it's just a matter of time before the nuclear plants margins are going to break too.
I was for nuclear after I discovered uranium has 1.000.000.000x the energy contents per weight of batteries but now I realize without cheap marginal coal, oil and gas nuclear is too risky.
As I see it now the only solution is to reduce consumption, yeasterday voluntarily or tomorrow (today in EU) dragged by our hair.
That article makes no sense for someone who know how things really work. There is no physical way cloud gaming can be faster than local. Probably that game only shows the fire after the server replies.
Also note that nothing is preventing Optical Flow Acceleration [0] (and subsequently the DLSS 3.0 models that they claim are exclusive to the 40 series) from running on either 2/3 RTX cards. Just like RTX Voice and other gimmick "exclusives" I expect it to be available to older cards the moment they realize their backlog of 30 series cards aren't clearing as quickly as they thought.
They're competing against an over-bloated secondhand market, AMD, Intel, much better integrated GPUs, and comparatively cheaper consoles that maintain sky-high demand with subsidized games via new subscription programs. They're vastly overestimating their brand loyalty (think Microsoft v Sony after the 360) and EVGA's exit makes more sense now than ever.
[0] https://developer.nvidia.com/opticalflow-sdk