The computer market has gone through significant changes over the past decade and GPU market leader NVIDIA has been pursuing new opportunities in mobile processors for the smartphone and tablet markets, addressing the challenges of cloud computing while also providing innovative solutions in the high-performance computing (HPC) space. However, if you ask NVIDIA how it views itself, the company will tell you that computer graphics remain at the center of everything it does, with the focus now being on making graphics performance more efficient.
For example, while the mobile computing market may be changing, demand for graphics continues to grow. More discrete graphics were sold into the notebook market in 2011 than in 2010. And the launch later this year of a reimagined Windows - designed for touch and mobility while offering support for office productivity applications and remaining accessible at all times - will provide strong market opportunities for solutions that deliver the best of all worlds, low-power and high performance, both on the client side and in the cloud.
Kepler platform
Nowhere is NVIDIA's focus on efficiency more evident than with its new Kepler platform, which was launched in March 2012. When NVIDIA announced its latest GeForce GTX 680 and 690 desktop GPUs, the company did not focus solely on having the best performance in the market, it was equally important to stress the efficiency of having the best per watt performance of any GPU on the market.
The Kepler platform was developed specifically with efficiency in mind, as NVIDIA realized that performance was not the only thing a demanding computer graphics market craved. Gamers wanted quiet systems and smaller system footprints, while mobile users wanted a longer battery life and cutting-edge graphics in smaller system footprints.
With Kepler, NVIDIA completely redesigned the streaming multiprocessor for optimal performance per watt. The company also added a GPU Boost feature that dynamically increases clock speed to improve performance within the card's power budget. And the platform's new streaming multiprocessor block (SMX) delivers twice the performance per watt of previous-generation products.
SMX uses an ultra-wide design with 192 CUDA cores and the new desktop GeForce GTX 680 has 1536 cores across the chip. What that means is that one GTX can deliver the same performance that three GTX 580s provided only one year ago. The GTX 680 also only draws 195 watts of power, compared to 244 watts on the GeForce GTX 580.
Compared with the closest competitive product, the GeForce GTX 680 GPU is more than 300% faster in DirectX 11 tessellation performance and up to 43% faster in cutting-edge games while consuming 28% less power.
Moreover, the Kepler architecture defines NVIDIA's GPU building blocks and will be the basis not just for a single desktop chip but for a family of designs that can be implemented in high performance desktop PCs, mobile solutions, workstations and data center solutions, as well as supercomputers.
Going mobile
On the notebook side, NVIDIA knows success is all about efficiency and the company is excited for the mobile designs its Kepler-based GeForce 600M series will make possible. The most important consideration for notebooks is battery life, which is why performance per watt determines how good a mobile GPU is. Compared to the 500M series, the 600M series is twice as efficient, resulting in faster frame rates and increased detail levels on chips that draw less power, which in turn leads to increased battery life, as well as reduced heat and noise output.
During Computex, visitors will for the first time be able to see thin and stylish notebooks running the NVIDIA GeForce 680M, with the performance matching what was only seen in fairly large and thick gaming notebooks a year and a half ago.
GeForce 680M on 13.3-inch Ultrabooks
The big question for NVIDIA on Ultrabooks was originally whether it could fit a GPU, physically and thermally, into an Ultrabook. At the original Kepler launch NVIDIA answered this question by highlighting a 15-inch Acer Aspire Timeline Ultra M3 Ultrabook featuring a Sandy Bridge processor and GeForce GPU onboard.
So now, with the launch of the GeForce 680M, the dialogue has transitioned to NVIDIA's ability to support increasingly smaller Ultrabook form factors with its GPUs. OEM partners at Computex are looking forward to showing off top-of-the-line dual core Ivy Bridge Ultrabooks with NVIDIA 680M graphics in form factors ranging from 15 inches and 14 inches, all the way down to 13.3 inches.
NVIDIA foresees the option of using discrete graphics in Ultrabooks as helping drive the development of the market and the company expects no negative headwind from growth in the Ultrabook market. Currently, Ultrabook volumes are modest due to the fact price points are somewhat high. Current prices are still in the US$800-900 range. However, when Ultrabooks hit US$599-699 price points, the market will take off.
At that point the Ultrabook market is expected to behave just like the overall notebook market, where discrete graphics are used to help vendors differentiate or provide market segmentation. In the Ultrabook market, vendors are also expected to have a very strong affinity for discrete graphics and the efficient Kepler architecture allows for the use of discrete GPUs in a variety of notebook sizes. Moreover, customers want the benefits of discrete graphics. Demand for discrete graphics in notebooks continues to grow and is especially strong in high-growth emerging markets. The attach rate for discrete graphics in notebooks is 80% in China and between 60-70% in Russia.
Kepler in the Cloud
Leading up to Computex, NVIDIA introduced a couple of new technologies - GRID GPUs and the NVIDIA VGX platform - that accelerate cloud computing using the computing capabilities of the GPU. NVIDIA's cloud GPU technologies are based on the company's new Kepler GPU architecture and designed for use in large-scale data centers.
NVIDIA GRID GPUs minimize power consumption by simultaneously encoding up to eight game streams, allowing providers to scale their service and offer support to millions of concurrent gamers. The processors feature two GPUs, each with its own encoder, and have 3,072 CUDA technology cores and 4.7 teraflops of 3D shader performance. This enables providers to render highly complex games in the cloud and encode them on the GPU, rather than the CPU, allowing servers to simultaneously run more game streams. Server power-consumption per game stream is reduced to about one-half that of previous implementations, an important metric for data centers.
The streaming technology also reduces server latency to as little as 10 milliseconds - less than one-tenth the blink of an eye - by capturing and encoding a game frame in a single pass. The GeForce GRID platform uses fast-frame capture, concurrent rendering and single-pass encoding to achieve ultra-fast game streaming.
NVIDIA also recently unveiled its NVIDIA VGX platform that enables IT departments to deliver a virtualized desktop with the graphics and GPU computing performance of a PC or workstation to employees using any connected device.
NVIDIA VGX enables remote workers to access a GPU-accelerated desktop similar to a traditional local PC and the capabilities extend to professionals needing to use 3D design and simulation tools. The VGX platform addresses key challenges faced by global enterprises, which are under constant pressure both to control operating costs and to use IT as a competitive edge that allows their workforces to achieve greater productivity and deliver new products faster. Delivering virtualized desktops can also minimize the security risks inherent in sharing critical data and intellectual property with an increasingly internationalized workforce.
This technology can allow client devices, such as an iPad or an Android tablet, to run a high-end application such as Autodesk in a virtualized environment.
While cloud gaming and cloud virtualization is not expected to replace NVIDIA's core business in the near future, the company's interest in the segment is part of its strategy to expand and enhance its product coverage to where it sees opportunities. The cloud is a trend that is definitely occurring, but there is a gap today in terms of computer graphics in the cloud and that is a big void that NVIDIA realizes it can fill and NVIDIA will continue investing in the area moving forward.
Microsoft RT and Tegra
By and large the big thing at Computex Taipei for NVIDIA will be efficient computing and the gearing up of Microsoft RT is a perfect example of the market moving toward platforms that demand more performance per watt.
The new platform will be optimized for hardware that has great battery life, while also being designed for mobility and optimized for touch. NVIDIA's Tegra processor is going to be the heart of a number of these new systems as it offers the best of all worlds. NVIDIA is the only provider in the world who can provide a quad core solution, can differentiate on content and has a long history optimizing its products for Microsoft.
The launch of Windows RT will also be a transformative moment in the computing industry. Having full office productivity running on an always-on, always-connected and long-battery life device represents mobility meeting the PC for the first time. The launch of Windows 8 represents as large an inflection point for Microsoft as the launch of Windows 95. And as the success of the iPad has shown, the appetite for these mobile devices is pretty high.
Computex Taipei
So this year at Computex Taipei, the theme for NVIDIA is Efficiency. It all revolves around the computing graphics of the new Kepler platform, delivering the best performance per watt with the GeForce GTX 600 series to enable efficient and quiet gaming solutions on the desktop; delivering gaming performance with longer battery life in increasingly compact Ultrabook form factors with the GeForce 680M on the mobile front; and delivering the best of all worlds with industry-leading quad-core performance on the Windows RT supporting Tegra platform.