Nvidia Sees The Future In Deep Learning

by Sohini Bagchi    Aug 30, 2016


Nvidia has been a leader in producing the technology behind high-quality graphics for years, but the company is now betting on a different future. With the rapid advances in self-driving vehicles, warehouse robots, diagnostic assistants, and speech and facial recognition, there’s plenty of reasons for companies to be excited about deep-learning-based artificial intelligence (AI). Nvidia is one such technology firm that is extremely bullish in this space.

In a recent conversation with CXOtoday, Vishal Dhupar, Managing Director-South Asia, Nvidia said that the company is moving ahead with its AI-focused hardware, software, and solutions for the enterprise.

As such data scientists in both industry and academia have been using graphics processing units (GPUs) for machine learning to make groundbreaking innovations across a variety of applications including image classification, video analytics, speech recognition and natural language processing or NLP. In particular, Deep Learning, the use of sophisticated, multi-level “deep” neural networks to create systems that can perform feature detection from massive amounts of unlabeled training data, is an area where Nvidia is significantly investing in recent quarters.


According to Dhupar, that’s because Nvidia’s GPUs have proved to be especially adept at the parallel processing needed for deep learning. The chipmaker in May unveiled its GeForce GTX 1080 and 1070 graphics processors based on its Pascal technology that were reportedly 10 times faster than the previous generation.

“While the idea of using GPUs in gaming was always popular, Nvidia believes GPUs can be used in various facets of like and work and this is where deep learning can play a pivotal role.” Dhupar calls this as “a new computing model, a fundamentally different approach to developing software.”

Read more: 3 Key Trends Shaping The Future Of AI

The combination of massive data, better algorithms and powerful GPUs led to a disruption in modern AI. In many cases, deep learning is now surpassing the capabilities of humans. Examples of this progress in the past year including Microsoft’s work on image recognition with the ImageNet database, Berkeley’s work on robotics, Baidu’s speech recognition services, and most recently Google DeepMind’s AlphaGo - reasons enough for Nvidia to bet big on deep learning.

And this has clearly reflected in its double digit year-over-year sales growth for three consecutive quarters. Analysts expect that trend to continue for the next two quarters, boosting its annual revenue of 22 per cent to about USD 6.1 billion. That represents a massive acceleration from its 7 per cent sales growth last year.

Dhupar explained how deep learning is impacting its businesses, primarily driving it in three areas:

High-end GPUs to accelerate deep learning

Earlier this year, AMD made a comeback in gaming GPUs with its low-priced Polaris cards, the RX-460, 470, and 480. The 4GB version of the 480, which costs USD 200, was touted as the cheapest dedicated GPU for “VR ready” PCs. Shortly after, Nvidia struck back with its Pascal-powered GTX 1060, 1070, and 1080 cards.

Read more: NVIDIA Unveils GeForce GTX 1060

The USD 250 GTX 1060, which had 6GB of RAM and ran roughly 15 per cent faster than the RX 460, giving gamers more - and better - choice. Nvidia then launched a 3GB version of the 1060 for USD 200. The company claims that the low-end card runs 10 per cent faster than the 8GB version of the RX 480, which costs USD 240.

Driving its data center revenue

While Nvidia is primarily known for its gaming GPUs, it’s also been making solid progress in data centers with its high-end Tesla GPUs. That’s because GPUs are more effective than stand-alone CPUs at machine learning and AI processes.

One of Nvidia’s key partners is IBM announced in May that it will start offering Nvidia’s Tesla M60 GPU accelerators to cloud-based enterprise clients. The Big Blue collaborated with Nvidia previously offered Nvidia’s Tesla K10 and K80 GPUs to its cloud-based clients, and claimed that pairing its own Power CPUs with the K80 enabled its Watson AI platform to answer questions at a much accelerated pace.

A boost to connected cars

Another smart move was looking at connected cars by working around its ARM-based Tegra mobile processors, which failed to gain much traction in the smartphone market.


Leading automakers like BMW, Audi, Honda, and Tesla started using Nvidia’s Tegra-powered VCM (visual computing platform) because it delivered better graphics, 3D navigation, and audio for infotainment systems than older processors. Nvidia is building upon those foundations with Drive PX, a Tegra X1-powered “supercomputer” which powers the ADAS (advanced driver assistance systems) that manage automatic braking and lane changes.

The new PX2 platform allows the automotive industry to use artificial intelligence to tackle the complexities inherent in autonomous driving. It utilizes deep learning on Nvidia’s most advanced GPUs to determine precisely where the car is and to compute a safe, comfortable trajectory. Already Volvo, Ford, and Audi, among others have installed Drive PX in select vehicles.

Read more: Four Technologies Shaping Cars Of The Future

“Nvidia’s GPU is central to advances in deep learning and supercomputing, believes Dhupar. We are leveraging these to create the brain of future autonomous vehicles that will be continuously alert, and eventually achieve superhuman levels of situational awareness,” Dhupar stated.

In such a scenario, autonomous cars will bring increased safety, more convenient mobility services and become a powerful force for a better future.

(Image courtesy: Nvidia)