Predictions about Pascal?
Do you think it will be that big of a jump from the 700's? or 900's?
Who's gonna buy one, /g/?
>>51756264
I'll sell my 980ti once the top tier version of the pascal series is out to buy one.
>not manufacturing and releasing the volta series instead of the pascal series
>>51756264
>50% more tesselation power
>put jewworks into every big game
amd is done.
Go with AMD. Pascal will be shit for VR.
>>51756777
Source?
>>51756949
His ass. Maxwell sucks for it, but they're fixing that with Pascal.
That being said, you should still go for AMD. Better longevity.
>>51756264
just bought a 970 recently but I will sell it once they release a 1080 ti and I will buy that one.
http://www.gputechconf.com/
APRIL 4-7, 2016
SILICON VALLEY
Wouldn't buy anything now when it's just 4 months to GTC 2016
>>51757039
why?
>>51757088
want to play in 60+ fps on my 1440p 120hz monitor.
also because I can afford it.
The only difference it'll make is for MUH DEEP LEARNIN. This is also a graph for MUH DEEP LEARNIN
Ignore >>51757183 moron
http://www.geeks3d.com/forums/index.php?topic=4247.0
Pascal
-D__CUDA_ARCH__=600
-D__CUDA_ARCH__=610
-D__CUDA_ARCH__=620
Pascal's CUDA Compute Capability is enhanced over Maxwell
>>51757227
>troll / 10
again, this is a MUH DEEP LEARNIN graph and any performance gains applies to that only,
>>51756981
>fixing that with pascal
nice meme. Fuck Nvidia, you can't trust them for shit. Pascal ia going to be shit.
>>51757280
NOPE.AVI
It applies to other compute stuff also
>>51757363
>BUTTMAD AMDPOORFAG WITH HOUSEFIRE NO DRIVERS DETECTED
How close is the 1000 series to release? Does anybody have a clue or good estimate?
I'm running a single GTX 970 now and it's starting to drop below 60 fps in some of the newer games (e.g. getting more like 40 fps in heroes of the storm when the action picks up), so I'm looking for an upgrade relatively soon.
I would probably get a 980 or 980 Ti now (probably the latter to have some headroom for the future), but I'm wondering whether it would be better for price/performance to wait out for the GTX 1000 series.
(Alternatively, I could get a 980 Ti now, then sell it and upgrade when the 1000 series comes out.)
>>51757380
>Trusting novydia after the 3.5GB and fermi housefire fiascos.
How much did nvidia pay you?
>>51757415
6-9 months would be my guess from what I have seen and heard.
atleast 6 months, not sooner.
>>51757415
Q2 2016 for the 1080 and 1070 most likely.
>>51756669
good goy
>>51757452
Not to mention the drivers that literally killed laptop screens, or the 860M debacle.
>>51757470
>>51757475
That's long enough for me to buy a card now and then consider trading up on release, thanks.
>mixed precision
They don't even give a shit anymore do they? Meanwhile the fury X is shitting on ALL nvidia cards with 8+ TFLOPS of FP32 compute performance.
>>51757557
GTX 980 Ti & TITAN X is FASTER
8 TFLOPS but slower than the competition, just plain embarassing
It's not how much theoretical FLOPS you have, it's how you PERFORM
>>51757587
Is that FP32 compute performance? You know the performance required to ray trace quickly?
>>51757747
single precision if FP32, double is FP64, half is FP16
btw, nobody raytraces anymore, rasterization won.
>>51756981
>fixing that with pascal
That's not a guarantee. AMD's GCN cards are much better for VR than nvidia equivalent. I doubt this will change.
Most likely it will be another lie like "4GB" or "DX12" or "Async Compute"
>>51756264
>GFlops/W
>>51757803
rip
Raytracing gives so much more realistic results, it's a shame it's not researched more...
>>51757415
I've heard Q2 - Q4 I'm curious too, are there any prices yet shown? I have a GTX 770 and I'm saving my money to get a 1080 or whatever it's called by then.
http://www.overclock.net/t/1575638/wccftech-nano-fury-vs-titan-x-fable-legends-dx12-benchmark/110#post_24475280
>Saying that Multi-Engine (aka Async Compute) is the root of performance increases on Ashes between DX11 to DX12 on AMD is definitely not true. Most of the performance gains in AMDs case are due to CPU driver head reductions. Async is a modest perf increase relative to that.
>Async is a modest perf increase relative to that.
>2015
>Still posting async compute memes when it's been DEBUNKED by Oxide's own shitty devs
Nvidia fastest in DX11, DX12 and compute
>>51756264
huurrr who is ready for 7,5gb
>>51757986
Fable only uses 1 ACE. Older amd 7000 series has atleast 2. 290 series has 8. And so on.
There isn't any drawback from using a dynamic ACE scaling. (if cards support 8, use 8, if 2, use 2). The main reason for this limitation is Nvidia. The more async compute AMD uses, the more game performance AMD gets. The opposite is true for nvidia due to much slower contextual switching.
Also note, Fable runs Unreal engine, which by default disabled Async Compute. Unreal Engine and Nvidia are partners. So expect any unreal engine to not include proper async compute by default. The console games utilizing unreal engine utilizes thirdparty async compute fix.
>>51757803
I don't understand, why are nvidia GPUs dogshit in luxmark?
>tfw bought a 980ti 2 months ago
Oh well
>>51758160
Are you a moron? That is Oxide dev talking about Ashes, not Fabled
Stop trying to be a liar, your async compute memes have been DEBUNKED by Oxide themselves
>>51758025
kek
btw pascal should have up to 16 GB
>>51758392
You wont do anything with that memory. Hell, you don't ever need more than 4gb
>>51758392
Only the highest end Quadro and Tesla cards have 16GB, maybe a TITAN class card
Everything else will have less memory, 4-8GB
>>51758410
No game uses more than 3.5GB anyway.
>>51758289
Which part of my post is incorrect?
I think you're confused or just parroting without even understanding what Oxide is saying.
Re-read oxide dev's post. Even if I take what you believe Oxide devs concluded, this still means there could be more benefit from using async compute. A modest performance from spending only 4-5 days with small async compute usage.
The other point is, DX12 directly reduces overheads. this will directly improve AMD performances as we know AMD drivers have overheads. Nvidia's cards are already optimized so reduction in overheads won't help them much.
>>51757947
We just need more x86 core CPUs. CPUs always give the most accurate ray traced images and scale really well.
Maybe if AMD released a relatively cheap 64-core x86 cpu @ 2ghz then ray tracing might be done in real time (aka 60-120 fps) at 720p or greater.
Intel won't do that so we can only hope AMD will.
>>51758445
lol
>>51758190
It's opencl, opencl can't really be benchmarked unless all cards are benchmarked in the exact same circumstances.
>>51758410
that's not up to you, it's up to the devs. witcher 3 and fallout 4 are already maxing out my 4 gigs of vram. can't wait to max out 16 ;)
>>51758504
Fallout 4 is a horribly optimized game. ANY game can max out hardware if its programmed like shit
got 7790 with 1 GB of ram atm.
i'll have money to replace it at the end of february.
should i wait for pascal/greenland or just get something like 380x/390? i don't want to bother with selling it to purchase new one, if i do get one it's supposed to last ~2 years (or at least untill new architecture as i'll skip pascal/greenland then)
1080p, dont need everything on ultra, but i want 60 fps at decent settings.
>>51758584
Buy AMD. VR is going to be the hottest meme in the next few years. You wouldn't wanna miss out looking up Miku's ass would you?
>>51758584
What games do you need to play that 7790 can't play at medium/light?
TBQH, I'd wait for pascal/AMD offers then make the decision. Otherwise, a solid 390 would do you fine throughout DX12 era.
>>51758607
VR is just like 3d television, but actually worse.
>>51758625
depends on price of all that VR gear, up to lets say ~500 dollars i could get into it, but for example for 1500 not really. also depends on how good it'll be, didnt bother with 3d tv at all (part of it is that i dont see shit w/o my corrective glasses so anything that doesnt allow me to use them is out of question).
maybe there'll be more news about release dates, i assume greenland wont be here at least untill early Q3 so i could justify 390
>>51758637
Are you saying virtual reality isn't the future? VR is still primitive but it will only keep getting better. This means you are required to have an AMD card unless you want to vomit every few minutes from the input lag novydia cards cause.
>>51758559
It doesn't change the fact that there are instances today where more than 4GB of vram could be handy to have.
>>51758728
>Are you saying virtual reality isn't the future?
It's not the future unless you're training to pilot a spaceship into battle with other spaceships
>>51758814
better safe than sorry anon. Also VR is obviously going to be used to mainly look at pr0n.
>>51758885
If I want to look at porn I only need to look in one direction. Definetely never behind me where the camera, sound and lighting crew are standing, definetely never at the male
https://www.youtube.com/watch?v=wIl2-5f8NTo
Only thing VR is for
>>51758885
but I thought 3D was pig disgusting?
>>51758814
Eh, not really. I'm already using VR for primitive pass-through AR stuff and it really won't be long until most computer displays are HMDs. It will be trivially easy to do hand-tracked UIs and mixed reality in just a few years.
That being said you should be waiting for good mobile / wearable VR/AR hardware, and there's no point in buying a desktop for it. Desktop stuff will suffer from wireless transmission latency issues, since cable management is a game-breaker. And once you've got mobile VR, that's when VR will merge with AR and mixed reality will be the standard.
The thing about HMDs and hand/body tracked interfaces is that they are universally accessible and intuitive. You can literally poke at them with your hands and watch them in 3d space. Hands are the most natural user interface tool in existence, since everyone has a lifetime of training with them.
I wouldn't be surprised if HMDs were so small and powerful by 2060 that we build them directly into bionic eyeballs. Same thing with hand tracking sensors.
You should really check out this presentation on the subject:
https://www.youtube.com/watch?v=xAUG6prB2nA
It will be great in comparison to the 700 and 900 series because they will gimp them in drivers to make pascal look better.
>>51759111
>muh gimp meme
atleast nvidia doesnt kill your cards like amd does :)
>>51756264
The architecture is one thing. The big deal is the process. FinFETs will make it a worthwhile upgrade.
>>51758966
just stop it you fucking troll
>>51759194
what makes you think I'm trolling?
>>51758931
It is. VR is going to make H-games massively popular though.
>>51757363
>Fuck Nvidia, you can't trust them for shit.
i would rather buy a gimped gpu than a nuclear reactor rebrand any day of the week, shill retard
>>51759276
well shilled
>>51759146
except they did
three times
>>51756264
So does anyone know when they'll be available for purchase and how more powerful roughly they could be to the 900s? This time I don't want to play catch up and buy a lower range. I want whatever 1080TI or whatever they come up with off the bat. I want to make sure I'm saving up enough in advance
>>51759276
>i would rather buy a gimped gpu than a nuclear reactor
You know nvidia specializes in that right?
>>51758415
This is spot on, I'm a bit privy as to what's being planned. Rumors and word of mouth. Expect small Pascal to launch around May with customers actually getting them around June. 1080 should be about $500 and provide a 15% performance bump over the 980 ti equipped with 6GB, the 1070 will come in at $375 with 4GB and is slightly slower than the 980 to. About 3 months later expect the 1060 with 3GB of Ram priced around $240 and peformance between a 970 and 980, the next Titan will follow with the full 16GB and a performance gain of about 50% on the 980 ti, launch price should be about $1000 like normal. Then expect the big Pascal chip, the 1080 ti to be out for the holidays at about $750, peformance will be "40%" of a 980 ti, expect none reference designs to be faster, equipped with 8GB. All with HBM2.
However temps and power consumption are a problem right now. With what I've heard, expect even less overclocking headroom than Maxwell. But maybe this will change, I'd expect Volta to be the real game changer.
>>51759399
Summer 2016
1080ti with 16gb hbm2 memory 1999$
>>51759276
Good shill the goy.
>>51759369
>980 ti winning at an amd title
amd is so fucking incompetent
>>51759444
Yeah amd needs their own jewworks.
>>51759430
>1080ti with 16gb hbm2 memory 1999$
i wouldn't be surprised, we just found out today that zen and arctic islands haven't actually taped out yet. meanwhile nvidia taped out nearly 8 months ago, unless an act of god saves AMD we are in for both a cpu and gpu monopoly in 2016 and most of 2017.
>>51759444
Nice trips but at least our drivers don't kill your laptop.
>>51759430
I doubt it will be THAT much.
>>51759430
>1080ti with 16gb hbm2 memory 1999$
I'm wanting to go balls deep this time and shell out maybe 800€, but that would be too much
>>51756264
Looking at that chart, should I just skip Pascal, get a 980ti now, and then wait for volta?
>>51759479
>Nice trips but at least our drivers don't kill your laptop.
that was a manufacturing error, not a fault on nvidia's part
>>51759479
It already killed desktop gpus and it would've killed laptop gpus but amd is too incompetent and can't into efficiency so Nvidia owns 100% mobile discreet market.
>>51759505
Get a 390/970 and wait for the next generation senpai.
>>51759515
How much is nvidya paying you? Didyou quit your dayjob to shill nvidia all day? How do I sign up?
>>51759475
There needs to be a new image made to show how nvidia is gaming the DX12 games and changing Async Compute direction in their favor.
They are promoting limiting of async compute to only 1 engine or not using it at all. This works in favor of nvidia because they have 1 engine (and x amount of ques) but AMD can have 2 to 8 engine and more use only increases their performance.
>>51759570
when was the last time anyone saw an amd powered tablet?
>>51759606
>What are APUs
>>51759606
Nvidia is generally very good at everything they do. Servers, desktop, mobile.
AMD struggles to make any decent cpu and gpu.
>>51759633
I remember how hard trinity was shilled on /g/
They were desperate and the shilling immediately stopped a week after release from how incredibly shit it was
>>51759641
Your shekels are falling on the ground goy. Don't forget to pick them up.
>>51759686
AMD products are catered towards the lowest common denominator. When you have to drink ketchup soup and need to see up for computer you buy amd
>>51759728
*damage control intensifies*
http://www.anandtech.com/show/9815/amd-moves-pre-gcn-gpus-to-legacy
>So all pre-Kavari APUs: Llano, Trinity, and Richland, are now also legacy APUs
If you bought any AMD APUs, you're a moron in the first place
>>51759683
They really have improved since then though. Especially on the laptop side of things.
>>51759417
>performance between a 970 and 980 for $240 in less than a year
>next titan 50% more powerful than 980 ti
Yeah, no. This isn't happening unless the driver gimping meme is true. Pascal is literally an incremental improvement over Maxwell, not some kind of revolutionary new design.
>>51759818
My laptop A10-4600M is now a legacy APU?
fugg ;_;
I'm building a custom desktop next year so whatever.
Maxwell came out with a bunch of new tech like delta color compression, MFAA, VXGI, etc
What new tech will launch with Pascal?
Reminder: By supporting AMD products, you are indirectly supporting nVidiafags. When you keep AMD alive, you let the nvidiafags enjoy lower price due to competition. You should let AMD die and let the nVidia monopoly begin. This way next time, they'll have to pay $3000 for 3.5GB of 1070.
Proper way to live is to not buy any from nvidia or amd. Let AMD die, therefore nVidiafags suffer as a result. You still have your peace of mind to boot.
You should rather chop off your arm than let the arm rot and make it give you cancer.
>>51759342
>resultant energy draw still lower than AMDs
>B-BUT IT WENT UP FOR NVIDIA
It's hard to disprove a housefire with cinders, isn't it? :^)
>>51760092
due to AMD's investors being primarily saudi nations you are also indirectly supporting ISIS
>>51760199
>nations
nationals, rather
Who else is buying a new gpu next year just for oculus rift?
>>51760307
vr is a unicorn
>>51760334
Just like your hopes and dreams.
>>51760199
I thought that ISIS were fighting the Saudi government.
>>51760389
This. Now I'm confused as well.
>>51757803
Movies are still raytraced. Some of them even use global illumination.
Raytracing is dead for gaymen, though.
>>51760437
For now
>>51756669
Sell now, make mad dosh now, when Pascal is out , your 980 will lose value
>>51760389
>>51760423
not the saudi government, rich saudi businessmen
>>51756777
Lol, fucking kids and their shit understanding of tech.
VR isn't different and nvidia cards have no problems with it.
>>51757947
ray tracing by itself doesn't give more realistic results, only when it's fully physically based, and that isn't being done in realtime
>>51760994
Except nvidia cards have massive input lag in VR. Enjoy feeling like shit after a few minutes of using VR.
>>51758456
> CPUs always give the most accurate ray traced images and scale really well.
It's all IEEE 32 bit floats kid. There is no more accuracy in a GPU or CPU.
Intel is actually releasing a 72 core 288 thread processor called knight's landing.
>>51761053
>fully physically based
Everything is based in the physical world to some degree.
Have you ever seen a rendered sequence of images that matches a real life short movie in terms of quality without extremely heavy use of real-world textures? No?
>>51760455
For now? How do you think movie 3d is going to be rendered in the future?
The poster you applied to is 100% correct.
>>51761100
> Everything is based in the physical world to some degree.
That's not what physically based ray tracing means. It means that you will get a full simulation of photography.
I'm not sure where textures came into it, textures are more or less a solved problem.
>>51761059
That doesn't make any sense at all. You get a certain frame rate when running something. There isn't much more to it at the card and driver level.
>>51761119
He was talking about the gaymen part. Real time ray tracing can be done but it needs 1 or maybe 2 knights landing co-processors to do it. Which is right now, expensive as fuck.
>>51761148
>That's not what physically based ray tracing means. It means that you will get a full simulation of photography.
No, thats not what it means. It's a nebulous term and all it says is that the equation has SOME basis in our physical world as opposed to NPR or non photorealistic rendering which usually has no basis.
>I'm not sure where textures came into it, textures are more or less a solved problem.
what are you talking about?
>>51761206
What are you talking about is the question.
Someone said raytracing gives more realistic results. I'm saying it isn't that simple, ray tracing can give more realistic results based on how physically accurate it is.
You brought in textures and all sorts of nonsense.
And texture by the way are a solved problem, they are fast and analystically anti-aliased using techniques which are basically a 2 dimensional version of frequency clamping from signal theory.
We can keep going if you want to figure out where your rendering knowledge stops.
>>51756264
if it has full Vulkan and full DX12 support i'll consider purchasing NVIDIA again
>>51761302
>You brought in textures and all sorts of nonsense.
No, what I said is that without extremely heavy use of real world textures your rendered result will never EVER pass for photoreal, regardless of raytracing or scanline methods.
>And texture by the way are a solved problem, they are fast and analystically anti-aliased using techniques which are basically a 2 dimensional version of frequency clamping from signal theory.
how much of a lel are you? Keep up with the conversation you retard
>We can keep going if you want to figure out where your rendering knowledge stops.
I work in 3DCG, have for 3 years...in shader programming :)
>>51759146
Both AMD and NVIDIA have killed plenty of cards because of borked drivers
>>51761544
Name a single card NVIDIA has killed because of drivers
>>51761571
Their 196.75 and 320.18 are known card killers
More recently there was something involving Windows 10 and the driver it was installing for NVIDIA GPUs but idk if this was nvidia driver fault or just Windows 10 being the piece of shit it is
>>51761389
> No, what I said is that without extremely heavy use of real world textures your rendered result will never EVER pass for photoreal, regardless of raytracing or scanline methods.
You brought up textures out of nowhere, I'm still not sure why.
> how much of a lel are you? Keep up with the conversation you retard
You don't seem to have anything real or substantial to back up whatever it is you are trying to say.
> I work in 3DCG, have for 3 years...in shader programming :)
I'm sure that impresses most of the CS101 kids here, but the fact that you think that is impressive is how I know about where your level of knowledge is.
>>51761690
You're backpedalling hard m8. Go back to posting here in similar threads like this while I develop and ship actual product that people use ok :)
>>51761721
> You're backpedalling hard m8. Go back to posting here in similar threads like this while I develop and ship actual product that people use ok :)
You avoided saying anything, I'm assuming because you don't really know anything.
Raytracing produces realism when it used in physically based renders, the ability to collide lines with geometry doesn't automatically somehow guarantee good looking images.
Texture mapping is a by and large a solved problem.
There is no backpedaling here, only you avoiding any substance or explanation out of fear that someone on 4chan might show you that you don't know as much as you thought you did.
>>51761169
I am not an expert in VR, but from my understanding the motion to photons latency is different from frame rate broacasted to the display.
Even if you are running at a 90 fps or 1000fps or whatever lets say, the moment you move perform some action in a game it takes time for the frames corresponding to these actions to arrive.
So if your frame rate is 90 fps but the frames you are seeing are offset in time that can induce motion sickness if it goes over a certain threshold that you cannot tolerate.
>VR IS NOT A MEME I SWEAR GUYS I PROMISE VR IS THE FUTURE
2-3 years prior
>GUYS 3D IS NOT A MEME I SWEAR GUYS 3D IS THE FUTURE
>>51761571
>Name a single card NVIDIA has killed because of drivers
Are you this much of a newfag?
This place exploded back in the Thermi days when that driver, I don't remember the number, killed 400-series cards by disabling the fan.
>>51759241
>>51759194
This man speaks the truth. Anyone else who says otherwise is the same fagg who sawded the first airplane and went, "LOL WOOD FLYING THROUGH SKY".
>>51756264
Who the fuck gives a shit about nVidia when AMD is so much better
>>51759401
>2015
>still referencing Fermi
d a m a g e c o n t r o l
>>51765586(you)
>>51760307
>facebookulus rift
>not objectively superior vive
baka desu senpai
>>51757380
>Calling people 'poor' when they are dead tired of nVidia bullshitting them
Go fuck yourself fucking anti-consumer.
How can you even browse this fucking board and be anti-consumer
Jesus christ
>>51759641
AMD have plenty of good GPUs, only reason to get nVidia is for single card performance out of 980Ti if you can't afford 2x295.
Fuck 4k gaming currently nvidia gets stomped by AMD Fury X
>>51759641
>Nvidia is generally very good at everything they do. Servers, desktop, mobile.
Tegra is trash for anything other than gpu. Tegra as a cpu is unusable.
>>51759728
He says unironically, sitting atop his quad SLI house fires; actively getting paid to shill for the company that is burning down his house as he shitposts
Are all the 18% market-share dudes on this place?
>>51760307
I've already bought a 980 for that. I've also bought a i7-5930K and a HOTAS.
I'm just gonna play Elite Dangerous though, so i'm Ok. Might buy another 980 if i need it later.
>>51758410
>load everything into VRAM
>no more hits to the hard drive during gameplay
>>51756264
>Predictions about Pascal?
It's not gonna be nearly as big of a leap from Maxwell as people think.
>>51759369
Why the everloving fuck is the Titan X above the Fury X? The Fury has a better average and minimum. Same for the 980 and 390X.
VR is going to flop just like 3d tv's. nobody cares about that shit. all anyone wants is maximum fps for the $ and nvidia always delivers.
>>51762898
That is absolutely true for TVs that do lots of post processing but very unlikely to be any kind of a hardware issue with pc video cards. If there is any latency it would be in the drivers, although it is unlikely that there is much there now and it will be extremely unlikely that there will be much once dx12, vulkan, and actual headsets hit the market.
>>51766697
Have you tried VR?
Do AMD even have drivers for the Oculus?
Is it a beta driver clusterfuck like everything AMD?
>>51767021
>drivers for the Oculus
It's just treated like an ordinary monitor. The application is responsible for the stereoscopic rendering and lens distortion.
>>51767052
Oh WTF!
There's no VR things in graphic drivers?
>>51767116
Actually there is: https://developer.nvidia.com/virtual-reality-development
>>51761059
Nvidia cards run fine with VR already. I have a DK2 + 970 and I haven't had any problems with input lag or discomfort when it comes to VR.
>>51758559
therefore get a card with as much ram as you can afford, since you can't control the devs ;)
>>51767184
To be fair, that VR SLI thing is all but deprecated because of D3D12/Vulkan, and the multi-res shading is hardware-constrained.
>>51756264
60% boost in performance per watt
>tfw per watt
>>51760091
i want 10-bit color, but it ain't gonna happen. fuck more pixels, we need better pixels (unless you're in VR).
>>51759146
But they did. 196.32, 320.18 and I forgot the latest one with Windows 10 where the EDID gets borked. Also, the AMD version only fucked heavily-OC'ed cards cause normal and light-OC'ed cards still had hardware-level thermal protection as a failsafe.
>>51756777
>VR
Its a shitty gimmick m8, Sony is going to kill VR forever with its shitty gimped PS4 headset. Enjoy it never taking off and being a lost technology.
>2015
>people still literally unironically fighting the shitty duopoly cardwars
Good goyim, support the other company :^) it's not like they both have to exist in order to not be a monopoly and purposely hold back progress.
AMD is such a fucking shit company, fucking christ. They don't even give Jewtel any competition. We are all goyim cattle to the Jewtel/AMD duopoly and don't forget the NVIDIA/AMD Duopoly.
>>51769099
heres your reply.
>>51765615
source?
Anyone who brought a 980ti is a retard.
They're going from 28nm to 16nm, and using HBM memory.
>>51770556
true. 28nm TSMC for 4+ years is suffering.
>nvlink
knowing people will fall for this shit makes me want to kill myself
fuck every "game developer" who put jewworks into their shit game and jewvidia for still existing
fuck capitalism
