Nvidia GeForce GTX 1080 rumor mill just blew up.

sirplayalot said:
All types of 980tis selling for 450-530 on Reddit. It's so hard to not pull the trigger! Just a few more days...
Click to expand...

yeah I got the one in my sig for $450. Unless Pascal is really compelling (specifically for heat produced in my NCase) I don't see a need to upgrade. My upgrade will probably come when Gsync 1440p ultrawide monitors become cheaper.
 
So Tuesday at 9AM is the deep learning keynote by Jen Hsun. If a new Titan is being launched that's where it will happen.

Should at least get an update.
 
hehe

OH OH OH

I speak Chinese, and in the opening trailer there was a lady asking a question to her phone in Chinese and receiving an answer.

The question was "Who invented the mechanical calculator?"

The answer is Blaise **Pascal**.

It's happening.

It's happening alright.
Click to expand...
 
Loving this so far, Nvidia pushes the boundaries in so many ways. WTF has AMD ever done that comes close to all the work in AI, rendering, etc. that Nvidia does and continues to do.

AMD is such a useless company.
 
razor1 said:
just a joke ;)
Click to expand...
np, I hoped to get my post in before other posts, have now added the quote.
(it wasnt aimed at you :))

I just tried some VR pron to see what its about.
Amusing but needs the video stretching vertically to display on a normal monitor, and ability to zoom.
MPC-HC lets you do this with the numpad.
The 3D works very well if you set the zoom so the eye spacing is correct.
 
ah np!

Presentation notes


Cuda developers x4 over all

Cuda developers in automotive and hyperscale x10


nV SDK:

Gameworks:

Volumetric lighting

Voxel Accelerated AO

Hybrid frustum traced shadows


Designworks:

Adobe MDL

Iray


Compute Works:

CUDA 8

cuDnn 5

nvGraph

Index plugin Visualization of data quickly

VRworks:

Oculus Rift and HTC Vive

Unreal, Max Play and Unity


Driveworks:

Still working on it, but is available to test with. Early access has already started Q1 of next year is the release.


nV Jetpack

GIE: GPU inference engine coming soon in May (jetson tx1: 24 images/watt), CUDA is the most energy efficient approach for deep learning.

VR:

Going to be able to do design visualization
Going to places where we can't normal go

Photo-real is a necessity, we need more performance.

Iray Vr can do this takes many GPU's and time to do this for photo realism VR but can be done in real time now.

Iray Vr lite, can be used on any hardware types and already has integration into 3dsMax and Maya and Google cardboard box coming in June.

AI:

5 years ago Deep learning started.

Alpha Go, 1000 cpu's and 60 GPU's. Computers powered by deep learning can do more than humans can program for.

New Computing model:

Deep learning Object detection, DNN, Data HPC

No longer have to have different programs written to do different things and it gets better results.

Industry funding is high 5 billion

AI has become a platform

P100 is in volume production.

P100 samples are out and they are being used by OEMs Q1 2017 servers will be available.

Deep learning supercomputer DGX-1
170 TF
3200 watts, 8 GPU's, 7 tb sdd's, etc.
12x faster performance for deep learning from last year.

Pascal with recurrent neuronets:
Interconnect is very important
Capabilities:
Persistent RNN's, keeping everything in the GPU with less
Register file for pascal 14mb vs 8 mb in Maxwell
nV link helps with splitting work across GPU's. Creates a wider model with more processors (30x more)

TensorFlow, DGX-1, easy adaptability, performance is key.

DGX-1 $129k

Already getting colleges and research labs and medicine are targeted
 
Last edited:
Ooookooooooooooo

Gp100!!!!


Tesla like I expected

Seks
 
Holy shit I totally expected this and I'm still ecstatic
 
Amazing stuff!!! Tesla P100 is incredible.

I can't wait to see what they release for consumers.
 
Digital Viper-X- said:
are you planning on doing some deep learning ?! or some ai development?! :)
Click to expand...
Yeah, some friends of mine ran their training on my 980Ti remotely for the last few months, it's loads of fun and the speedup using Cuda and cudnn was absolutely insane

8 hours on dual cpu amazon server

45 minutes on 980Ti, at 50% gpu utilization :p
 
they aren't going to announce any consumer video card products?!?!?!?
 
They're saying a ~6x speedup in deep learning performance in a cluster (fp16) versus the older cards (I scaled the Pascal cluster down to 4 to be comparable).

There's only a 75% increase in fp16 compute performance over Titan X.

So this must be due to the new NVLINK interconnect plus the much faster memory bus? Guess Maxwell compute was truly bandwidth-limited...but we knew that already :D
 
Last edited:
I don' t think they will take about Pascal for gaming, but with everything they are talking about, looks like June looks likely for Pascal, GP104. Many of software packages are coming out at that time too.
 
This is going to be a real bummer if no consumer pascal. Not that I need a new card right now but it's definitely time for new cards to come out
 
He just held up the next-gen driving AI platform and spoke about "unannounced pascal GPUs".

Keynote isn't over yet, so I could be wrong, but it looks like no consumer pascal today.
 
schizo said:
He just held up the next-gen driving AI platform and spoke about "unannounced pascal GPUs".

Keynote isn't over yet, so I could be wrong, but it looks like no consumer pascal today.
Click to expand...
If they can produce the Tesla behemoths, surely they can make smaller variants. Although I believe he said the Tesla part is aiming at Q1 2017 for availbility (cloud sooner).
 
He did, and the tesla GPUs are all HBM2 also, which we know is unlikely this year.

So yeah, maybe. But not anytime soon.

Edit: Annnnnnd that's it, no consumer pascal. Sad panda.
 
Inside Pascal: NVIDIA’s Newest Computing Platform

Pascal also provides superior scheduling and overlapped load/store instructions to increase floating point utilization. The new SM scheduler architecture in GP100 improves upon the advances of the Maxwell scheduler and is even more intelligent, providing increased performance and reduced power consumption. Each warp scheduler (one per processing block) is capable of dispatching two warp instructions per clock.
Click to expand...
 
My theory? They couldn't get it cheap enough with HBM2 to even hit $999 for a Titan in 2016.
 
You must log in or register to reply here.
Back
Top