It's over, Nvidia is finished. http://www.forbes.com/s

>>57827991
A thread died for your shit bait, please kill yourself

Google threw a penny at AMD meanwhile Nvidia stocks rose from around 60 to 90 dollars because everyone wants in on CUDA for neural networks

>>57828164
CUDA is the shit. I literally just bought a 6GB 1060 for deep learning. That being said, I'd love to see some competition in the GPGPU space, and this is literally the first I've heard of any. It's a good thing.

I wish AMD would get their shit together, we really need the competition

this means p. much nothing tho

even googles own framework for deep learning only runs on nvidia, how are they even gonna do this?

I think nvidia will rather get competition from those vendors doing NN/low-precision-specific ASICs than AMD at this point. Supposedly google has ASICs deployed that are specifically designed for TF and similar workloads

>>57828164
>>57828582
>he doesn't have processors designed specifically for training neural networks

>>57828582
I have significant doubts you do anything, let alone know anything about Deep Learning.

You just sit at your desktop, with your mid-range headphones and dac, and only listen to hipster pop trash. Sipping your extra sugary tea, shitposting on boards because you have the latest midrange Nvidia card purchased with birthday money from your family because you're actually a depressed poorfag with low willpower and rock bottom self esteem. But you dropped out oif college and everyone around you thinks you are super smart at 'da puters'.

>>57828582
>. That being said, I'd love to see some competition in the GPGPU space
You're kind of a moron, aren't you?

>>57828872
>>57828164
>this means p. much nothing tho
>Recently, Advanced Micro Devices clinched a deal with Google to supply its FirePro S9300 x2 GPU in the latter’s cloud platform. AMD’s stock price has shot by close to 30% since the announcement of this deal
Of course it's nothing my shill friend
>even googles own framework for deep learning only runs on nvidia, how are they even gonna do this?
Porting to OpenCL and/or Vulkan

File: D5E.jpg (83KB, 1258x818px) Image search: [Google]

83KB, 1258x818px

>>57830324
Please shit down my throat.

>>57828164
>>57828582
>>57828872
http://wccftech.com/amd-cuda-compilercompatibility-layer-announced-with-the-boltzmann-initiative/

>FP64 0.8 TFLOPs (1/16)

AYYMD IS FINISHED & BANKRUPT

>>57831750
>he thinks you need FP64 for deep learning

"lel"

>>57830324
10/10

I guess this is as good a thread as any for my question:

So I recently bought an amd freesync compatible monitor, and I was planning on buying a 1060 but now I feel like it'd be shame to let a feature go to waste. Is it worth getting a 480 for freesync?

>>57830324
Someone feels the need to feel intellectually superior.
Daddy didn't pay you enough attention?
Or maybe he paid too much attention.

>>57830324
This feels like a projection

>>57828872
I mean amd will be big soon, they are working on HBM cards now

>>57830336
>AMD stocks shot up 30%

so it rose by 2 bucks?

Nvidia rose by 40 dollars within the past 6 months, does that make them 20 times as good as AMD?

>>57832061
>Muh Async!
>Muh HBM!
>Muh moar cores!

Isn't it fucking hilarious how AMDrones always suck their dick for these "innovations", yet Intel/Nvidia still blow them the fuck out by just doing what they do?

>>57832061
>RX490
>HBM2
>can't beat GTX1080

>>57832023
Answer my fucking question

>>57832133

If you have a freesync monitor already just get the 480. Any differences between it and the 1060 are barely worth discussing.

>firepro

More like firestarterpro.

>>57832133
19 minutes doesn't entitle you an answer. Fuck you.

>>57827991
>poojeet company buys poojeet products

>>57832133
buy GTX1070 if you have enough money
if not then go for the RX480

also I really hope you bought a 144hz monitor and not a shitty eyecancer 60hz

>>57828164
hmm lets see..
someone will choose cuda because?oh wait it can do all sorts of paraller works
oh wait..
no never mind

also i think you dont really understand what is going on here...
we know since 2015 that amd made a de/compiler for cuda the boltzmann initiative
they practicly just fragments the serial work that cuda does into multithreaded ones
that is why they have won big contracts so far..they dont need to push opencl anymore since they can just beat cuda on its own house
>>57828872
trying to be a smartass
buying a gpu that doesnt have any core for this job only to brag about something he doesnt know

>>57830336
> Porting to OpenCL and/or Vulkan

that's a very significant effort tho, and so far there is no sign of that happening. I mean, I guess google could've been doing this all along in secrecy and soon we'll have a huge source drop that adds the functionality, but that seems like an odd way to go about it.