NVIDIA Moving From Visual Computing To HPC

Vishal Dhupar
Managing Director, South Asia, NVIDIA
NVIDIA is consistently evolving its architecture to make GPU computing efficient and faster. Over the years, the company has made significant inroads into new solutions, which cater to high performance computing (HPC), deep learning and Cloud for its enterprise line of business. In the consumer space, it has not only widened its graphic card portfolio, but has also launched its own brand of gaming tablet and Android TV.

The latest Maxwell architecture was introduced earlier this year, leading to the most advanced graphics and gaming GPUs.

The most important feature of Maxwell is an all-new design for the Streaming Multiprocessor (SM) that dramatically improves power efficiency by more than 40% higher delivered performance per core, without compromising on the benefits of the previous generation Kepler architecture.

There are several improvements to control logic partitioning, workload balancing, clock-gating granularity, instruction scheduling, number of instructions issued per clock cycle, and many other enhancements.

Speaking to the Channeltimes, Vishal Dhupar, Managing Director, South Asia, NVIDIA Graphics, said: “NVIDIA has been in the forefront from the visual computing in Consumer and Enterprise sectors. In consumer space, we have the popular NVIDIA GeForce graphics cards that we sell for gamers, and the pedigree of the company comes from gaming. We also have the serious side of the business, which is our enterprise. Today, visual computing/ professional graphics has become a mainstream in last 20 years. Fundamentally, because of the huge impact of graphical imaginary in professional applications. Today, no physical products could be developed in paper, all moved to digital.”

GPU is a cash cow
For NVIDIA, the Moore’s law has played on both the semiconductor hardware and software. Its Quadro graphics card solutions for the workstations has been gaining momentum. Today, NVIDIA claims to have a lion’s share in both consumer and enterprise visual computing and has been consistently improving the power and performance of these cards to have better photorealistic rendering for 4K display. It has built- in display capabilities for 5k and 8k.

“We bring out our new architecture, which upgrades the power and performance of the cards every year. On the traditional business, cards are all about acceleration, power and performance. Hence, our traditional card business is in a linear momentum, where these cards are growing along with the applications, that has been going consistently, although we dominate the market in GPU,” he added.

Besides Visual Computing, in the enterprise side, the company has seen huge amount of acceleration for its compute solutions Tesla. With Tesla, from energy exploration to machine learning, data scientists can crunch through petabytes of data with Tesla accelerators, up to 10x faster than with CPUs. For computational scientists, Tesla accelerators deliver the horsepower needed to run bigger simulations faster than ever.

Getting strong with Tesla

Dhupar added, “Tesla has emerged as the next growth engine for us. Tesla has become a very strong compute engine as a traditional CPU is a serial based computer, where GPU is a parallel based computer. So Tesla has emerged as the compute solution of choice for the verticals like higher education research, scientific research, govt, defence and supercomputing. We have several scientific communities are using Tesla in scientific research, IIT Mumbai, 100 of universities are using, scientific govt of departments of the countries. It is now emerging as a big revenue bet for NVIDIA.We have been in Visual Computing for last 20 years, and now we have compute avenues going big way.”

The company has also revived its Tegra business, after seeing a poor uptake of embedded processors for mobile phone market, which has now re-launched, and now gone into automotive solution TX1 launched new version. It is not positioned as mobile processors, as the bulk of the mobile business is price conscious so it does not make business sense. It is used to power your infotainment and visual imaginary with in the cars.

Betting Big for Cloud 
The other big thing NVIDIA made foray into is Cloud. “One of the challenges, we have seen is with the traditional professional graphics cards, it is very difficult to run on the cloud, other than the traditional workstation fundamentally due to large data set and connectivity. The other initiative we have taken in recent times is we have partnered with virtualization players like VMware and Citrix, and created NVIDIA Grid Server solutions, it is currently positioned as captive data center based virtualized graphics solution. In India, we have positioned it typically for smaller companies, with having 50 users, while bulk of the market, we will continue to be Quadro workstation driven. Only certain profile of the companies can be the users with multi locations usage, IP security concerns and mobility,” Dupar explained.

The company is still at a nascent stage. The approach it is taking is direct and mapping those companies, who are looking for workstation refresh and working closely with the business owners and workstation OEMs.

Launch of Shield Tab and TV for gaming
In consumer space, from the beginning, NVIDIA has been looked at as a components suppliers in the PC business. In the past NVIDIA tried to test its fortunes in the smartphone business, with its Tegra processors. But it did not last for long. Therefore, the company decided to launch its own range of gaming devices. Today, Tegra is being developed for its two products as Tegra X 1 automotive solutions and Tegra based Shield tab.

In last couple of years, the company has made conscious decisions to get into the consumer business. In the gaming side, it has launched Android based gaming tab, ‘Shield’ and complemented it with our 4K interactive TV with its Tegra X1 processor. To make this business viable, the company has acquired talent from consumer companies like Sony and Philips. Presently, the devices are available and testing in US and Europe markets, but it will get launched in the Indian market in near future as well.

“It is very unusual for an engineering company like NVIDIA to make a foray into consumer brands. This move would not only allow us to enter and offer an alternative to Sony and Microsoft gaming console business,but at the same time, it also opens up the doors for $100 bn gaming market. It is high potential market for us. It can power NVIDIA as a consumer branded devices.

Channel business

In Channel, for its enterprise and channel business, the company has two main distributors for graphic cards: Leadtek and Rashi Peripherals. Moreover, it also works closely with the resellers and assemblers. The company is also looking to broaden the horizon of ISVs to move from desktop to workstations.

Leave a Response