GTC15 Keynote Live Blog

11:11 Have a great GTC everyone.

11:11 Jensen wraps up. Titan X, the world’s fastest GPU, DIGITIS DevBox GPU deep learning platform, Pascal – 10x Maxwell for deep learning, Drive PX deep learning platform for self-driving cars.

11:04 Jensen – I get excited every time I get an [Tesla] OTA

10:59 10-50 MPH in urban environments is the most challenging part of autonomous driving, but we know how to solve it and will be there in a few years

10:58 Elon – Tegra will be a real enabler for autonomous driving

10:55 2B cars on the road, about 100M manufactured each year, 20Y replacement cycle, we won’t have 100% self-driving or 100% electric cars for a long time

10:52 Self driving cars will be like elevators (elevators used to have human operators)

10:50 Applause welcomes Elon Musk on stage

10:49 Drive PX DevKit, $10,000, May 2015

10:47: AlexNet on Drive PX, 184 frames per second

10:34 Jensen shifts discussion to ADAS, I think his special desk must be coming on stage soon

10:31 Key Pascal Features

10:28: Coming next year, Pascal, mixed precision, 3D memory, NVLink, 10X faster than Maxwell for deep learning, *very rough estimates

10:21 DIGITS DEVBOX

10:19 DIGITS Deep GPU Training System for Data Scientists. Process data, configure DNN, monitor progress, visualize layers.

10:15 Walking through some automated image captioning results. Thanks to all those Amazon Mechanical Turks who helped caption the original training data and to Julie B. for countless hours of reviewing results to pick out the most interesting examples

10:12 Andrej Karpathy describes how ConvNets and Recurrent Nets can be joined together for automated image captioning.

10:07 Deep learning revolutionizing medical research, predicting toxicity of new drugs, understanding gene mutation to prevent disease, detecting mitosis in breast cancer cells, groundbreaking work

10:06 Very glad to see EyeEm listed as one of the start-ups doing GPU-accelerated deep learning.

10:04 Volumetric rendering being done by Nvidia IndeX

9:58 Mike Houston joins Jensen to help explain how deep neural networks, specifically AlexNet, are processed on a GPU. No rabbits were harmed in the processing (have to watch the video).

Titan X with cuDNN2

9:40: #gtc15 on Twitter. You don’t have to be here to know that the press tables are full

9:39: Titan X, $999. Clapping.

9:36 Titan X for Deep Learning, training AlexNet, 43 days on 16-core Xeon, 2.5 days on Titan X + cuDNN.

9:31 Epic Games trailer – kite, need to watch the video, words don’t capture the beauty, 100 square miles (about the size of Silicon Valley), running in realtime on a single Titan X

9:30 Titax X. World’s fastest GPU. 8B transistors, 3072 CUDA cores, 7 TFLOPS SP, 12GB memory, based on Maxwell

9:28 Rolling Titan X video

9:27 Part of our promise is access to the platform. By putting CUDA in every GPU we make it easy for every developer in the world to access

9:25 Our promise is to accelerate your code, striking a balance between ease of programming and speed.

9:24 54,000 GPU teraflops around the world today, 3 million CUDA downloads, 60,000 academic papers

9:15 4 things to talk about 1) A new GPU and deep learning, 2) A very fast box and deep learning 3) Roadmap reveal and deep learning 4) Self driving cars and deep learning

9:13 Jen-Hsun comes on stage. GTC is about developers, about sharing ideas, about being inspired

9:11 Opening video starts, “do you remember the future … its here”. Going to be a lot of discussion of deep learning.

9:09 Tune into the Livesteam to watch.

9:08 Packed house for the #gtc15 keynote. Announcer asking everyone to take their seats

Advertisements

About Marc Hamilton

Marc Hamilton – Vice President, Solutions Architecture and Engineering, NVIDIA. At NVIDIA, the Visual Computing Company, Marc leads the worldwide Solutions Architecture and Engineering team, responsible for working with NVIDIA’s customers and partners to deliver the world’s best end to end solutions for professional visualization and design, high performance computing, and big data analytics. Prior to NVIDIA, Marc worked in the Hyperscale Business Unit within HP’s Enterprise Group where he led the HPC team for the Americas region. Marc spent 16 years at Sun Microsystems in HPC and other sales and marketing executive management roles. Marc also worked at TRW developing HPC applications for the US aerospace and defense industry. He has published a number of technical articles and is the author of the book, “Software Development, Building Reliable Systems”. Marc holds a BS degree in Math and Computer Science from UCLA, an MS degree in Electrical Engineering from USC, and is a graduate of the UCLA Executive Management program.
This entry was posted in Uncategorized. Bookmark the permalink.