NVIDIA GTC 2017 Keynote: What you need to know and the Tesla V100

2
Jensen Huang GTC 2017
Jensen Huang GTC 2017

NVIDIA GTC 2017 comes on the heels of a blockbuster earnings release from NVIDIA, buoyed in large part by its data center group. It is well known that AI, deep learning, and VR are the current-generation killer applications. NVIDIA is currently leading in all three areas, and at GTC 2017 we expected to hear more about the future from NVIDIA’s CEO.

Talking Compute Scaling

The first point is that Moore’s Law is dead. For those unaccustomed to these presentations, NVIDIA’s chief competition is Intel so the company takes a shot at Intel each show.

NVIDIA GTC 2017 End Of Road For General Purpose Processors
NVIDIA GTC 2017 End Of Road For General Purpose Processors

One of the big reasons is that through CUDA and GPU architectures, extracting parallelism leads to enormous performance gains. Each generation of GPU, we see performance gains backing up these claims.

NVIDIA GTC 2017 Rise Of GPU Computing
NVIDIA GTC 2017 Rise Of GPU Computing

Although we do not yet see GPUs running general purpose software, next-generation compute heavy workloads are running on GPUs.

Project Holodeck

NVIDIA is on a major push to get their products into design collaboration tools and virtual reality. Project Holodeck is one such project to enable that.

NVIDIA GTC 2017 Announcing Project Holodeck
NVIDIA GTC 2017 Announcing Project Holodeck

Our highlight from the demo was seeing the Koenigsegg walkaround.

NVIDIA GTC 2017 Holodeck Koenigsegg Demo
NVIDIA GTC 2017 Holodeck Koenigsegg Demo

If you have seen VR showroom demos, this is an amazing capability. You can even use capabilities you would not have in a physical showroom:

NVIDIA GTC 2017 Holodeck Koenigsegg Parts Demo
NVIDIA GTC 2017 Holodeck Koenigsegg Parts Demo

This is a very cool demo.

The Era of Machine Learning

If you want the big revenue driver for NVIDIA over the past few years, it is machine learning/ AI.

NVIDIA GTC 2017 Era Of Machine Learning
NVIDIA GTC 2017 Era Of Machine Learning

The keynote continued showing the amazing pace of modern AI and highlighting some of the new techniques and capabilities.

NVIDIA GTC 2017 Big Bang Of AI
NVIDIA GTC 2017 Big Bang Of AI

Here is the view of the growth of the modern AI/ deep learning field using a few market facts.

NVIDIA GTC 2017 Big Bang Of Modern AI Students And Investment
NVIDIA GTC 2017 Big Bang Of Modern AI Students And Investment

How does NVIDIA power this? NVIDIA “supports every major framework”, is available in systems from every major OEM, and is available on every major cloud.

GTX 2017 Powering AI Revolution
GTX 2017 Powering AI Revolution

In terms of scale, NVIDIA Inception is an 18-month-old program to help deep learning startups and now has 1300 deep learning startups.

NVIDIA GTX 2017 18 Month Old NVIDIA Inception 1300 Companies
NVIDIA GTX 2017 18 Month Old NVIDIA Inception 1300 Companies

Beyond startups, NVIDIA is working in the enterprise space.

NVIDIA AI and SAP for Enterprise

A very cool application NVIDIA showed off with its SAP collaboration was the ability to recognize brand impact in video.

NVIDIA GTC 2017 SAP AI Brand Impact
NVIDIA GTC 2017 SAP AI Brand Impact

This can be used by the advertising industry to evaluate the amount of exposure a brand got from a given sponsorship placement.

NVIDIA GTC 2017 SAP AI Brand Impact Scope
NVIDIA GTC 2017 SAP AI Brand Impact Scope

One can see that the partnership is looking to expand well beyond this brand impact application.

Introducing NVIDIA Tesla V100

The setup for the new chip is that models are getting bigger.

NVIDIA GTC 2017 Model Complexity Exploding
NVIDIA GTC 2017 Model Complexity Exploding

Here is the shot for the new NVIDIA Tesla V100 introduction. The NVIDIA Tesla V100 is the successor one year later to the Pascal-based P100.

NVIDIA GTC 2017 Tesla Volta V100 12nm TSMC
NVIDIA GTC 2017 Tesla Volta V100 12nm TSMC

The new chip is clearly targeted at servers and HPC workloads. One of the key points is memory with a 20MB “huge” register file, 16MB cache, 16GB HBM2 providing 900GB/s. Just to give an idea in terms of performance improvement using the new Tensor Core:

NVIDIA Pascal V Volta One Year Later
NVIDIA Pascal v. Volta One Year Later

Tensor Core is the new compute construct for Volta. It essentially accelerates matrix processing in Volta. Here is the comparison between the two:

NVIDIA Tensorcore Pascal V Volta
NVIDIA Tensor Core Pascal v. Volta

Here is an example of the K80 to the P100 to the V100 performance improvement in popular frameworks:

NVIDIA GTC 2017 Volta V100 Performance Over P100 And K80
NVIDIA GTC 2017 Volta V100 Performance Over P100 And K80

Here are the Amber (molecular) and Googlenet benchmarks for generations:

NVIDIA GTC 2017 Amber And Googlenet
NVIDIA GTC 2017 Amber And Googlenet

The NVLINK servers are aimed at HPC markets where NVIDIA is battling the Intel Xeon Phi x200 series. We covered Intel Xeon Phi Updates at SC16 and how Intel was taking a lot of market share. These NVLINK parts are the high-end parts so we expect NVIDIA to push these to market as fast as possible.

These will be part of the new DGX-1 V with 8x NVLINK Volta GPUs.

NVIDIA DGX 1 V 149k
NVIDIA DGX 1 V 149k

NVIDIA will be shipping in Q3 (OEM partners in Q4) and will upgrade systems bought today to Volta when it comes out.

For hyperscalers, the HGX-1 will bring 8x NVIDIA Volta based Tesla V100 GPUs for their clouds.

NVIDIA HGX 1 With V100 For Hyperscalers
NVIDIA HGX 1 With V100 For Hyperscalers

Finally, for developers, a 4x Volta GPU NVIDIA DGX Station:

NVIDIA DGX Station
NVIDIA DGX Station

This will also be available in Q3 and is even water cooled. One item you can notice in the NVIDIA DGX Station is that it will include NVLink not just PCIe 3.0 to connect its GPUs.

Inferencing Speedup Volta v. Skylake Claims

In terms of inferencing support, Volta is expected to be a major speedup.

NVIDIA GTC 2017 Inferencing Broadwell K80 Skylake P100 V100
NVIDIA GTC 2017 Inferencing Broadwell K80 Skylake P100 V100

The inferencing 150w FHHL single slot card is set for adding in commodity PCIe nodes.

NVIDIA Tesla V100 FHHL
NVIDIA Tesla V100 FHHL

The claim was made that you can move 500 CPU based inferencing servers down to 33 nodes with new Volta Tesla cards.

NVIDIA V100 33 Nodes Versus 500 Nodes CPU
NVIDIA V100 33 Nodes Versus 500 Nodes CPU

We still have some time before we expect Volta to hit the mainstream GPU market.

NVIDIA Cloud! Containerzied Deep Learning

NVIDIA just announced containerized GPU accelerated deep learning platform.

NVIDIA Cloud Registry
NVIDIA Cloud Registry

NVIDIA will provide a registry with the popular frameworks, datasets and pre-trained models. You can then run workloads locally or in the cloud using their web interface:

NVIDIA GPU Cloud Web Interface
NVIDIA GPU Cloud Web Interface

At STH, we have been using nvidia-docker for some time. We even have Monero and ZCash cryptocurrency mining instances using nvidia-docker.

 

Final Words

If you have been noticing more GPU content on STH, that is for a good reason. We are going to be covering the space increasingly in the future. If you are looking for the key driver of today’s coolest workloads, this is it.

2 COMMENTS

  1. Isn’t V100 a recycled product name ? Wasn’t the V100 the chip powering the last generation of 3DFX graphics cards (3DFX 3 3000 etc …) before it went under and Nvidia acquired it ?

LEAVE A REPLY

Please enter your comment!
Please enter your name here