What happend to nvidia? Why are they so bad at drivers now?

>aaymdrone faggot mad his company only releases budget rebrands

>>59985237
DX12

>>59985269
so its Microsofts fault?

File: opfaeg.png (15KB, 316x296px) Image search: [Google]

15KB, 316x296px

>>59985263
>Nvidia faggot that doesn't realize Nvidia is going full retard by following a no competition tactic

Top jej
>>59985237

File: 1480745341285.jpg (70KB, 389x357px) Image search: [Google]

70KB, 389x357px

>bought a 1060 without doing much research
>my soul when intense regret

>>59985275
No, it's just NVIDIA's current arch is unsuitable for any implementation of async compute due to lack of hardware scheduling.

>>59985275
NVidia's DX11 performance was all smoke and mirrors. They just patched shit into the drivers. Now they can't do that with DX12. Also, the async compute thing.

>>59985338
whats this async compute thing?

>>59985349
amdrone marketing buzzword

>>59985423
like fastsync?

what's wrong with nvidia drivers? i haven't had any issue with running games?

explain?

>>59985237
>Why are they so bad at drivers now?
The hardware is lacking a hardware scheduler, and they are using the driver to compensate for this. This of course isn't as good.

>>59985473
is that such a bad thing and why did nvidia removed it?

>>59985555
It is for DX12/Vulkan. They removed it for the power saving meme, that's also the reason they butchered compute.

>>59985555
Because Fermis was gigantic housefires.

>async compute
it housefire, that why nvidia remove it

>bad drivers
kek anon, the majority of people who report "bad drivers" are dumb fuckers like this guy
http://www.overclock.net/t/1627611/awful-aliasing-tearing-with-new-1080ti

>>59985611
This isn't English

>>59985349
It lets the GPU quickly switch from one thing to another without a huge performance hit. That's the ultra simplified version at least

AMD had hardware for it back in 2011 with the introduction of the 7970. Nvidia only added it to hardware in Pascal last year, but half arsed it so it doesn't work so well. It was pretty irrelevant 5 years ago, but both the PS4 and Xbox supporting it, and DX12 addeding support on PC. So now quite a few games take advantage of it.

>>59985349
It's like advanced pipelining.
>>59985423
Fuck off.

>>59985429

No, it let's your GPU use compute and graphics at the same time.

Nvidia only let's you run graphics OR compute.

>>59985813
>Nvidia only let's you run graphics OR compute.
The compute is gimped anyway.

>>59985813
Computing is for CPUs

amd shill: the thread

Good luck with you rebrandeon and wait vega (R)

>>59985237
Nvidia drivers are still better senpai

It's just AMD has rebraned 75% of their GPU line up over 5 years so their drivers are super stable at this point.

>>59985349
https://www.youtube.com/watch?v=v3dUhep0rBs

>>59985870
newfag

>>59985870
nVidia shill: the post

>>59985305
Yeah that made zero sense.

File: 1310429820535.jpg (13KB, 300x300px) Image search: [Google]

13KB, 300x300px

>>59985870
>rebrandeon and wait vega (R)

Looks like op got BTFO

>>59985847

Serial compute is. Parallel compute on GPUs humiliates CPUs.

>>59985955
im not an amd shill or nvidia shill im neutral

>>59985971
I don't care

>>59985903
That's a nice explanation.

>>59985823

On their GeForce shit tier cards. On their Quaddro it's actually competitive with AMD.

Which is why Nvidia does not want to put compute on their GeForce cards, because they will lose so much damn money.

>>59985901
Funny you should say that, because Polaris got released a month after Pascal.

>>59985998
>Quaddro
Pascal Quadros are gimped too. Well, besides GP100 one.

>>59986000
poolaris only offers mid-range budget cards though

>>59986015
(you)

>>59986015
Because it's half-Fiji and Fiji is a bottleneckfest.

@59986028
kys poojeet

>>59985269
>>59985331
>>59985338
>>59985423
>>59985473

Async is only useful when ur hardware is Inherently inefficient.
Nvidia figured out how to make their dx11 driver more efficient than using a hardware scheduler.

It's a genius move on AMDs part though
Remove hardware bottlenecks
And force driver optimization on the Devs

im a jew and i love nvidia and im hating amd because theyre like the nazis

>>59986056
Maybe AMD are just the better jews.

>>59986055
>Nvidia figured out how to make their dx11 driver more efficient than using a hardware scheduler.
That's a bandaid, especially considering modern engines are leveraging compute. Kys ramesh.

>>59986055
This
>>59986070
They had to manually patch each game.

>>59986084
>They had to manually patch each game.
And it still doesn't help when the game is compute-heavy. So hardware scheduler will be back and we'll get our second coming of Thermi.

>>59986055

Holy fuck. Who told you that? And why would they hate you that much?

It's inefficient to offload tasks from the GPU to the CPU then back to the GPU like Nvidia does. Versus having your GPU handle everything natively before being displayed.

>>59986104
>when the game is compute-heavy
>compute-heavy
amdrones everyone
buzzwords out the ass like retarded macfags

>>59986118
Are you braindead?

>>59986118

It doesn't even need to be compute heavy. Ashes of the Singularity uses a small amount of asynchronous compute and it took Nvidia engineers almost a full year of driver development on Pascal to get it in the same ballpark as AMD.

>>59985237
They're good enough to stay at the top. They sure as fuck won't improve for the hell of it.

>>59986176

Direct X 12 looks better than Direct X 11.

https://youtu.be/ZxsIOV2AjMc

>>59985338
'the async compute thing'
lmao
that shows me how you have no fucking clue what it is, or what it does, but you meme about it anyway because you watched a couple of adoredtv videos

nvidia gpu's are far more efficient than amd gcn shit, they barely benefit from async, unlike gcn which is a lot more focused on compute performance, which goes to waste in a lot of games.

>>59986176
>thread has actual discussion about hardware, with pros and cons of each vendor with people sharing experiences
>(You): Discussion on /g/! It's another shoa!

File: BF 1 dx11 vs dx12.jpg (118KB, 1279x477px) Image search: [Google]

118KB, 1279x477px

>>59986115
>inefficient to offload tasks from the GPU to the CPU then back to the GPU

It works fine because the nvidia DX11 driver uses every available thread.

Dx12 Bans driver multi threading
Hmmm I would wonder whyyy

File: thinkig.png (76KB, 448x448px) Image search: [Google]

76KB, 448x448px

>>59985237
>have to create an online account to use features local to my computer

File: amd.png (42KB, 653x726px) Image search: [Google]

42KB, 653x726px

>>59986346
DELET DIS DUMB NVIDIOT

>>59985237
I already regret my 1080ti since Vega is one month away, cheaper, and better.

File: 1485829114293.png (2MB, 2000x2000px) Image search: [Google]

2MB, 2000x2000px

>>59986724
>one month away, cheaper, and better.

>>59986763
Yeah I lied it's actually like half a month away

I own an gtx 1070, Amd guys let ask you, what would have been the Amd Alternative that would also let me only use a 650 watt power supply.

>>59986346
>the nvidia DX11 driver uses every available thread.
NVidia's DX11 drivers are held together with duct tape and silly putty.
>Dx12 Bans driver multi threading
You're a fucking retard. Look at the results from DX12 in Dolphin.

>>59986724
I wonder who could be behind this post...

>>59985237
OP is not a shill there has been windows 10 Nvidia driver issues for months.

>>59986803
An 1600/580 combo won't even use more than 335-400 watts under load.

>>59986824
>You're a fucking retard. Look at the results from DX12 in Dolphin.
BF1 uses DX12 as wrapper so results are shit.

>>59986834
Nice try OP

>>59986868
Figures, another shitty cherrypicked shill benchmark.

>>59986866
How exactly would a 580 compete with a 1070?

>>59985237
They don't care about gaymen anymore because they're making so much money selling Teslas in bulk for large companies to run CUDA workloads.

>>59986902
It can in amdrones mind

File: 14927070584951632212225.jpg (1005KB, 2560x1536px) Image search: [Google]

1005KB, 2560x1536px

>>59986879
LOL ok

>>59986928
Timestamp, tits and/or mantits or GTFO.

>>59985237
Their drivers are as good as always, but since Vulkan and DX12 push most of the responsibilities that used to belong to the driver onto the developer AMD no longer suffers from having shitty drivers.

File: 1492707644861-679734698.jpg (1MB, 2560x1536px) Image search: [Google]

1MB, 2560x1536px

>>59986941
Here you go nig

>>59986291
If they're so efficient that they don't need async, why do they suffer so hard under DX12 without their precious driver hacks?

>>59987165
Because they removed hardware scheduler to tame their housefires.

>>59987075
but this isnt even the gaming x edition

>>59985237
they've been putting out bunk drivers for a while now, they kinda suck.

File: Capture3.jpg (86KB, 1237x647px) Image search: [Google]

86KB, 1237x647px

>>59987165
>suffer

A bit exaggerated there m8y

>>59987267
>goyworks title with DX12 wrapper
Oy-vey, time to use that x64 tesselation.

>>59986346
look how minimal the loss is for those AMD cards are vs Nvidia
>defending massive loss in performance over new API

>>59987326
In the DF video the only game AMD wins in is Hitman which is an AMD sponsored game so by your logic it's disqualified as well.

File: 1477346282273.jpg (67KB, 800x600px) Image search: [Google]

67KB, 800x600px

>>59987197
I want them to add it back in and get some Fermi action going on.

So it /g/ for real that everyone that got a gtx 1070 has made a wrong decision?

>>59987349
They inevitably will. Like, they have no other choice. Duct taping more shit to their drivers won't help.

>>59986346
>Dx12 Bans driver multi threading
Except for the fact that it doesn't, AMD's DX12 driver is multi threaded.

Nvidia is a company that employs scummy business tactics, why would support much less give hundreds of your parents hard earned dollars to a company that just had to pay out a huge class action settlement because they flat out lied about the VRAM on a 970?
Nvidia owners are like abused housewives with Stockholm syndrome lol.

File: amada_shitposting.png (367KB, 1015x1072px) Image search: [Google]

367KB, 1015x1072px

>>59987416

>>59985237
Underpaid their driver devs, they were poached by AMD years ago
>>59985768
Nvidia had a hardware scheduler in Fermi, they removed it to tame their housefire

>>59987368
Not if you like crazy high fps numbers on just 1080p.

>>59987488
i like crazy high fps on 1080p

File: bc8.png (218KB, 592x385px) Image search: [Google]

218KB, 592x385px

>>59986346

>When you're being told a bald faced lie and know it's total bullshit

File: 64321659.jpg (112KB, 573x572px) Image search: [Google]

112KB, 573x572px

>>59986346
>Dx12 Bans driver multi threading

>>59986824
>>59987685
>>59987707
The API uses more threads
The driver doesn't
It's AMDs way to fuc over nvidia.
It's revenge for Game works killing AMD performance.

>>59988653
>The API uses more threads
>The driver doesn't
>It's AMDs way to fuc over nvidia.
>It's revenge for Game works killing AMD performance

If this is true then I love AMD. Sweet sweet revenge.

>>59989013
Nvidia has been doing shadey shit for a while.
Like having games over tesselate flat objects to lower AMD performance.

However, Nvidia has a lot of money and manpower.
They will most likely be able to counter dx12s shortcomings with newer Hardware.

>>59989160
>Like having games over tesselate flat objects to lower AMD performance.
And at ridiculously high levels like x64 or shit like that, way beyond the point it's even noticeable. It reached a point where AMD had to roll driver updates that limited tesselation to sane levels (the "AMD optimized tesselation" option) in order not to have their performance completely destroyed by goyworks.

>>59988653
>It's AMDs way to fuc over nvidia.
Microsoft is in control of DX12 last time I checked. Don't blame AMD for Nvidia's failings.

>>59989160
>They will most likely be able to counter dx12s shortcomings with Fermi housefires
Fixed.

Other than bad SLI support what bugs you?

>>59989160
>Nvidia has been doing shadey shit for a while.
>Like having games over tesselate flat objects to lower AMD performance.
Like this? https://youtube.com/watch?v=IYL07c74Jr4

>They will most likely be able to counter dx12s shortcomings with newer Hardware.
I guess in the end everybody wins.

>itt retards fall for normie yt b8

>>59985451

Yeah, this. I keep hearing the AMD shills repeat this same driver line but I haven't had driver issues since 2012 (when I bought an XFX 7970 which black screened every 10 minutes in games)

Phaggot op, I have a 1060 and a fake windows 7 with no updates and my card works fine, issues are with w10 , cause you have to be deep inside ms anal cavity aka up to date windows

>>59985847
Maybe in the 1990s

>>59989615
you can't blame AMD for XFX making cards out of oily rags stuffed into graphics card shaped paint cans

>>59990876
I have a xfx gtx260 still working

>>59989218
>Microsoft is in control of DX12
Dx12 shares 90% of it's coding with AMD's mantle.

>>59989615
I never believe anyone who says "I've never had an driver issue"
100% of the time it's bullshit.

>>59991169
>Dx12 shares 90% of it's coding with AMD's mantle.
That's Vulkan (which is also better than DX12).

nvidia drivers are pure shit

part of the reason i returned my exploding evga 1070

>>59991182
>That's Vulkan
No broski
Dx12 and vulkan are the same fucking thing
Mantle,dx12, and vulkan are the same API

>>59991921
OpenGL and DX11 are the same thing
OpenCL and CUDA are the same thing

>>59986163
And it still looks bad, runs bad and is played by nobody. What did DX12 achieve here?
What did nvidia even achieve here by optimizing for it? Nobody fuckin' plays it!

>NVIDIA will support the DX12 API on all the DX11-class GPUs it has shipped; these belong to the Fermi, Kepler and Maxwell architectural families.

https://blogs.nvidia.com/blog/2014/03/20/directx-12/

I'm still waiting for DX12 support for my GTX460.

>>59993206
Why. It's not like it is going to help you. The only benefit you might get from DX12 is lowered CPU overhead... but do you really have to worry about CPU bottleneck when your graphics card is that weak?

>>59985324
why intense regret? Mine runs fine in my laptop...

>>59993395
I mean, its not like theres an rx 480 laptop

>>59985305
>its nvidias fault they have no competition

>>59985237
http://blog.sec-consult.com/2017/04/application-whitelisting-application.html

>>59985237
>NoVideo GimpFarce Experience drivers which slow your card down over time and spies on you in the meantime

or

>AYYMD Housefire poos with drivers so bad that they take ages to actually get the basic performance the card where intended to get but doesn't matter anyways because all the new stuff they're supposed to release doesn't exist or is a rebrand

>>59993362
Why not. It's not that weak compared to console APU and Fermi got a hardware scheduler according to this thread.

Nvidia basically dictates OpenGL and then provides extensions that make it even faster. They support shit before anyone else and everything is tested on Nvidia first.

If you use Linux and game you basically can't buy anything else because AMD drivers crash.

>>59994134
>slow your card down over time and spies on you in the meantime
that's not all, they literally open you up to arbitrary remote file dropping.
>>59994048

>>59994154
What the fuck are you smoking dude. It gets shat on by console APUs, and DX12 is never going to change that.

DX12 does exactly 2 things. First of all it gives the game engine developer the ability to low level optimize for specific graphics hardware. They are NOT going to optimize for Fermi because it's obsolete and almost nobody uses it and optimizations for other architectures - even cards with different CUDA core count don't help you.

The hardware scheduler will help you by taking load off CPU which is normally used for scheduling... but this is useless, because your card is so weak you have to pair it with a completely garbage CPU to get any advantage from that. And who the fuck is going to optimize code for Fermi scheduler?

>>59994183
Ah, okay. They need announcement for no DX12 support for those architectures. I wanted to hear those reasons from responsible vendor instead of 4chan.が

>>59986916
this
nvidia got on the meme learning hype train hard

File: rms-7.jpg (136KB, 900x675px) Image search: [Google]

136KB, 900x675px

>>59986055
>scheduler
With all the talk about this lately both with regard to asynchronous compute and the CPU overhead of nVidia's drivers doing software scheduling, is there some good explanation anywhere about just what is being scheduled?

What's the unit of scheduling? Is it entire draw calls, or is it individual GPU threads, or is it instructions, or something completely else; and also, how does AMD's hardware scheduler solve the problem, and how does nVidia's software scheduling do it?

Is this information actually available, or is it the usual "it's GPU stuff, so we won't tell anything concrete at all about it" and /g/ is just throwing buzzwords around as usual?

>>59986055
>And force driver optimization on the Devs
Devs often appreciate this, mind you, since they'll then be in control instead of having to second-guess what the "intelligent" driver is going to do.

>>59991921
>Mantle,dx12, and vulkan are the same API
They have the same purpose, but they are very far from "the same".

>>59992910

DX12 is superior in every way.

https://youtu.be/9bvpyvoNXcw

>>59994309

No one really understands it but this guy explains it perfectly if you can deal with listening to him.

https://youtu.be/XOGIDMJThto

>>59986118
Kill yourself, retard.

>>59986055
>Nvidia figured out how to make their dx11 driver more efficient than using a hardware scheduler.
It was all fucking hacked together patches. The only reason NVidia "won" is they had more people to throw at the driver development team.

>>59994402
Please stop linking that game
It runs like garbage on BOTH DX11 and DX12.

I'd be willing to bet you've never even played it.

>>59986055
>And force driver optimization on the Devs
It's not a driver optimization problem in the first place, dumbass. It's something that should be developers responsibility in the first place. Just because NVidia did the devs homework for them for awhile doesn't mean it was going to last forever. Now game devs need to do their fucking jobs. Oh no, the horror.

>>59994533

Please eat my ass hole. Direct X 12 gives you a vastly better experience. Which you would see watching the comparison video.

>>59994585
DX12 experience is still shit. First of all because the game is bad, and secondly because the game runs bad on anything that isn't FuryX.
If there is an argument against DX12 ever being used, it's that game. Because it just shows that DX12 games are poorly optimized pieces of garbage. There isn't a single DX12 game that runs as good as DX11 or even Vulkan games. They all run like shit and don't even look good.

So somehow Microsoft managed to develop an API which reduces performance and makes games look worse. I can't figure it out, but good job MS.

>>59994421
To the extent that that video isn't outright wrong, it's just handwaving and no concrete information.

It makes it out as if asynchronous compute has to do with parallel processing of shader and DMA command queues, which is outright wrong, since whatever it is that async compute does, it is at least related to how compute work is scheduled onto the shader cores, not the GPU's DMA engine. And that's about the only concrete claim it made at all.

>>59994635
Welcome to every new API. Devs don't know how to use it, so it has worse performance, then it gets better as devs get the hang of it.

>>59994681
Vulkan is a new API and it doesn't appear to be anywhere near as bad so your point is moot.

>>59985237
They always were, but you were too busy falling for FUD about AMD's drivers to notice.

>>59985429
Fastsync is a fucking miracle for people that play cs go and overwatch. Of you compare it to vsync you can legitimately feel the difference in latency. Considering its strictly in software and costs nothing its a cool technology. Fuck gsync tax forever though

>>59994160
js for desktop applications was a mistake

>>59994849
server side js started us down this dark path

>>59994309
>What's the unit of scheduling? Is it entire draw calls, or is it individual GPU threads, or is it instructions, or something completely else; and also, how does AMD's hardware scheduler solve the problem, and how does nVidia's software scheduling do it?
I'm surprised no one can ever answer this just about anywhere.

As far as I can understand, it's instructions from the graphics API to the graphics driver/GPU.
It's just like instructions to the CPU, say.. like an AVX instruction.
Well think of like RISC vs CISC.
AMD GPUs can natively handle much more complex instructions, along with async compute and much more done on the GPU itself without going back and forth between the CPU and GPU to handle filling the GPU with work. The GPU can fill itself with work better.

With Nvidia GPUs, it has to emulate a lot of more complex instructions with a software layer.
The analogy would be that instead of something like AVX being one native instruction, it would break it down into separate functions.
But that's not the best analogy, because that's more akin to what you'd see in optimized shader compiles. What it really ends up being different is what happens under the hood on a graphics API like OpenGL and DX11 and how Nvidia has to intercept those calls and change them in a way that's optimized for the inferior way their arch works.

>>59995144
you're fucking retarded kill yourself

>>59995356
You don't have to take it so personally. It's not your fault that Nvidia's drivers and architecture is so bad, Anon. Not unless you work for Huang.

>>59995377
Post technical sources of your retarded claims or kill yourself.
No, don't link me poojeettv vids.

>>59995396
It's closed source proprietary. Don't know exactly what it does.
We just know that AMD's hardware scheduler is more sophisticated, whereas Nvidia's DX11 driver does a lot more work that it splits up into multithreaded.

https://forums.anandtech.com/threads/nvidia-vs-amds-driver-approach-in-dx11.2502940/ here you can read others discussing it, but it is just conjecture. I said "as far as I can understand" as I can't see their actual code and Nvidia has never said anything public about it, dumbfuck.
But it's the reasonable explanation as to why GCN does so much better when receiving a stream of instructions with low graphics api/driver overhead, whereas Nvidia's always has a lot of software overhead even when you move to DX12/Vulkan.

>>59995144
biggest difference between nVidia and AMD is that AMD is VLIW, this is why they suck at DX9-DX11 drivers, because VLIW optimization is ass.

>>59995488
GCN is not VLIW. It's non-VLIW vector SIMD. That was one of the big changes with GCN.

>>59995488
I'm pretty sure you meant that Nvidia switched back to VLIW from SIMD/SIMT with Keplar and still uses VLIW for their arch today. Whereas AMD moved away from it.

And this harks back to the reason why Fermi never got a DX12 Driver: It would have outperformed Keplar.
It's actually funny how they both swapped at the same time.

>>59995396
Unlike AMD, Nvidia does not publicly document their GPU instruction set. Even their Linux driver is proprietary garbage.

They are a highly secretive jewish masonry illuminati organization. This is a fact.

>>59995571
>>59995535
Well fuck me silly and call me Sully.

>>59994009
Its certainly possible, my 480 under volts like a champ

>>59986055
>Nvidia figured out how to make their dx11 driver more efficient than using a hardware scheduler.
THIS JUST IN, NVIDIA LITERALLY MADE OF MAGIC THAT CAN MAKE SOFTWARE IMPLEMENTATIONS RUN FASTER THAN HARDWARE.

Or perhaps their scheduler was trash that ate nearly half the power budget of the whole chip and didn't manage to give them any appreciable performance increase.

>>59995649
AMD has done this same thing for almost a decade now.
They don't put enough CUs, and instead have to overvolt and overclock to get 30% more performance at 200% the power usage.
Polaris was very very efficient around 950-1050MHz. But Pascal was a bigger jump in performance and forced them to clock to 1266-1340 instead. And now up to 1450 with the refresh.

With the 7970, it was holy shit.. so absurdly efficient down at 0.9v or so. Instead it was at what, 1.178v in order to be 30% more powerful than Fermi instead of around equal to it?
But if you dropped it down, it'd use almost half the power, almost ~1/3rd the power of Fermi, for performance that matched Fermi.

>>59995711
>THIS JUST IN, NVIDIA LITERALLY MADE OF MAGIC THAT CAN MAKE SOFTWARE IMPLEMENTATIONS RUN FASTER THAN HARDWARE.
That's not true, you idiot.
It runs better than AMD's DX11/OpenGL, but far far far worse than their DX12/Vulkan.

They made a software implementation that makes software implementations slightly less shitty. But then it simply uses more overhead when something could more efficiently speak more directly to the hardware.

>>59985324

I almost did this, I have a gtx 760 and it runs pretty well, Mass Effect Andromeda was released and planned on upgrading to a 1060, once I did my research it wasn't worth it, you could see a better improvement with a 1080 but those are really expensive for me.

>Starting with the Kepler architecture Nvidia has been moving some control logic off of the chip and into kernel instructions which are determined by the assembler. This makes sense since it cuts down on die space and power usage, plus the assembler has access to the whole program and can make more globally optimal decisions about things like scheduling and other control aspects. The op codes are already pretty densely packed so Nvidia added a new type of op which is a pure control code. On Kepler there is 1 control instruction for every 7 operational instructions. Maxwell added additional control capabilities and so has 1 control for every 3 instructions.

>>59995356
You're the one that should kill yourself. >>59995144 was right.

File: SAD.jpg (67KB, 600x396px) Image search: [Google]

67KB, 600x396px

>>59985237
Because the consumer has proven that they will buy nvidia on brand name alone time and time again. Even when they cost roughly the same
and are about same performance.
See>>59985324
Nvidia can be shitty because the consumer will let them.


>>59986055
>>59986346
>>59988653
>>59991169
>>59991921
>The shill is this misinformed

File: Intel.jpg (599KB, 1312x1740px) Image search: [Google]

599KB, 1312x1740px

>>59993362
>Why.
Because they said they would

>>59994157
Get out

>>59996199
>Even when they cost roughly the same

not everyone is an amerifat

>>59996780
AMD GPUs are the better value in pound, AUD, and euro as well.
You foreigners are just stupid to understand how to comparison shop, what sites to look at. Thus you get ripped off and stay poor.

File: 1467143491548.jpg (106KB, 633x568px) Image search: [Google]

106KB, 633x568px

>>59985237
>all these buttblasted AMD fanboys mad that Ryzen is trash for gaming and AMD only releases rebrands of budget cards

Best place to get drivers for an gtx 1070?

>>59996863
The driver cd :^)

>>59996834
i-i-ii-ntel iii-i-s t-t-t--ttt-he b-b-b-bb-b-est

>Intel fags stuck in 4 cores forever la la land.

Progress is a bitch.

>>59997179
Making updated wafers is hard when you fire all your engineers and give all your cuck earnings to Feminist Frequency instead.

>>59995638
That being said, though, the Nouveau project has reverse-engineered it.

>>59995571
>I'm pretty sure you meant that Nvidia switched back to VLIW from SIMD/SIMT with Keplar and still uses VLIW for their arch today.
That is not true. nVidia hasn't used VLIW since their pre-tesla architectures. They did make some significant changes from Fermi to Kepler, but going VLIW is not one of them.

>>59995356
Buttblasted Ndrone detected.

File: consider-suicide-2.jpg (21KB, 460x276px) Image search: [Google]

21KB, 460x276px

>>59995789
What are you even reading into >>59995144 to say it was either right or wrong? The post is saying nothing concrete, just waving words around.

>>59998964
>The post is saying nothing concrete, just waving words around.
Learn how to read.