Social Impact of the GPU

Computer Game Therapy

Video games are fun. Fun motivates learners. Therefore, video games motivate learners. Therapists in Italy are using this logic to help people of all ages overcome certain speech and mental deficits. As newer computers are equipped with increasingly powerful graphic cards, video games and applications are becoming ever more sophisticated with greater realism and improved 3D imagery.

This innovation is a boon to the researchers behind Vi.Re.Dis, who are developing virtual reality-based therapies to assist patients in developing their visual memory, storing complex pathways and developing predictive thinking skills. Early work has shown the power of video game-based methodologies to improve cognitive performance and sustain motivation throughout the therapeutic process.

More on Rehabilitation with Video Games (Italian)

Preventing Potential Pandemics

One of the primary reasons the 2009 H1N1 influenza outbreak was so infectious and deadly was due to frequent virus mutations that rendered existing anti-influenza drugs ineffective. Quickly identifying new mutations and developing inhibitor drugs to minimize the spread of the deadly virus would be key to preventing future pandemics.

Researchers in the United Kingdom and Thailand ran a large number of advanced simulations using a small computing cluster equipped with NVIDIA GPUs. This allowed them to observe how a multitude of H1N1 mutations could cause changes in the chemical and biological structure and behavior of a key enzyme of the virus. Armed with this information, they were able to determine, for the first time, what made the H1N1 virus resistant to existing antiviral drugs.

Helping to Ace the Space Race

Rocketry poses complex aeronautical, materials and computational problems, and each launch can cost tens of millions of dollars. This makes it essential to run complex, accurate design simulations quickly. India, which operates one of the largest space research agencies in the world, uses NVIDIA GPUs in its most powerful supercomputer (named SAGA) to accelerate and improve the design and analysis of its satellite launch vehicles.

SAGA runs compute-intensive, detailed computational fluid dynamics simulations that optimize launch vehicle design for improved performance under a range of conditions – and does so 7-8 times faster than with a CPU-based system. Harnessing the power of GPUs enables India's space agency to reduce the time needed to create, simulate and verify new launch vehicle designs from weeks to days, while dramatically improving the overall quality and durability of designs.

Doing More with Less

The energy efficiency of computing products affects nearly everyone. CIOs want to increase the performance of their data centers while driving down energy costs — goals that are seemingly at odds. Researchers, scientists and engineers can face serious limitations in their work due to the power consumption of high performance computing systems. Meanwhile, people all over the world have experienced how limited battery life can pull the plug on their access, pleasure and productivity while using mobile devices.

GPUs, however, are inherently more energy efficient than other ways of computation, and NVIDIA has long been focused on making its GPUs the most energy-efficient processors in the market. Because they're optimized for performance per watt and throughput, rather than absolute performance, GPUs have grown from their computer-gaming roots to enhance everything from smartphones and tablets to medical imaging and oil exploration.

Super-Sized Sampling and Simulations

Whether exploring insect behavior or potential pharmaceuticals, GPUs enable massive sampling and simulations that advance the work of scientists. Biologists are using GPU-accelerated programs to better understand the complex and coordinated behaviors that result from social interactions among individuals, such as swarming locusts and cells within tissue. Thanks to the exceptional performance of GPUs, scientists can now cost-effectively analyze the workings of biological systems using simulations of life-like group sizes - such as millions of ants in a colony - over evolutionary timescales and in realistic physical environments.

In the field of molecular dynamics, the GPU has allowed scientific inquiry to move forward without being limited by the amount of sampling that can be done. Molecular dynamics uses complex models to create realistic descriptions of atomic interactions. In areas such as drug discovery and materials development, this can involve examining literally millions of variations of a molecule's three-dimensional shape and other characteristics. Projects like AMBER, which more than 10,000 scientists use for the simulation of biomolecules, use GPUs to dramatically speed up performance while adding significant new functionality.

Safer, Smarter Cars

Improving automotive safety is increasingly an IT challenge. NVIDIA is working with car manufacturers like Audi AG to develop new safety systems that can detect pedestrians, read speed limit signs, improve navigation and help avoid collisions. Using cameras to analyze a car's surroundings, identify potential hazards and alert the driver is a task that requires the massive parallel processing of data, which is the GPU's specialty.

One of the safety concepts in development with GPUs at the Volkswagen (Audi's parent company) Electronics Research Laboratory is a feature known as pedestrian detection, which can identify children or objects in low-light situations. Work is also being done on collision avoidance, which allows cars to detect an impending collision and help stop the vehicle or prepare the vehicle for impact. The tight integration of these systems with the vehicle, as well as their simple user-interface and realistic graphics, will enable safer, more intuitive driving by providing critical information that can be easily understood with a quick glance.

From Local Weather to
Deep Space

For local weathermen and astronomers alike, accurately understanding what’s going on “outside” – whether in the atmosphere or in deep space – requires sophisticated computational models and immense computing power. NVIDIA technology is helping meet these challenges by speeding up performance times, cranking up processing power and dialing down energy consumption.

For example, NVIDIA GPUs will play a key role in the computational power of the most ambitious astronomy project on the planet – the Square Kilometre Array. Designed to pierce the mysteries of the cosmos, this mammoth radio frequency telescope will be 10,000 times more powerful than any telescope currently in use. Closer to Earth, meteorologists at NASA’s Goddard Space Flight Center are using NVIDIA GPU technology to create more accurate global climate models by vastly improving the resolution of cloud system simulations.

More on Square Kilometre Array | More on Climate Modeling

Modernizing Air Traffic Control

The current design and management of airspace and traffic flows is conducted through manual, human-centered processes. Reducing congestion and delayed flights requires enormous computational resources as tens of thousands of daily flight paths through hundreds of airports must be analyzed.

NVIDIA is working with companies like Mosaic ATM and Optimal Synthesis to better automate air traffic management, so air traffic controllers can manage the complexities of airspace, varying weather conditions and airport closures in real time. By integrating NVIDIA GPU-powered computing resources with the current human decision-making process, future air traffic control systems will be able to manage more planes in the sky and on the tarmac while improving travel safety.

A Supercomputer in Every Scientist's Pot

Traditionally, supercomputers have been large and expensive, making them accessible to only a select few scientists. The GPU’s parallel processing capability divides complex computing tasks into thousands of smaller tasks that can run concurrently. As a result, scientists and researchers across the globe – from Belgium to Mexico – are seeing results in days instead of months, even minutes instead of days.

NVIDIA also has worked with HP to offer the “GPU Starter Kit” as an alternative to the large footprints and lofty price tags of high-performance computers. Pre-configured so researchers can hit the ground running, the kit includes a cluster of HP ProLiant servers accelerated by NVIDIA Tesla GPUs – the computational engine under the hoods of the world’s fastest supercomputers. It also includes a broad set of development tools at discounted prices. With the power of a GPU-based supercomputer in every scientist’s lab or office, the potential for breakthroughs is limited only by their imaginations.

More on Improving Access to Supercomputing | More on GPU Starter Kit

A 3D View Into the Body

One of the most intriguing uses of GPUs is its application to data sets that represent the human body. Divvied up into high-resolution, 3D slices by imaging devices such as CT, ultrasound and MRI scanners, the body amounts to gigabytes upon gigabytes of data that offers doctors and medical researchers an unprecedented view into the human form. Processed by GPU-powered algorithms, this immense amount of data can provide doctors a realistic 3D view of a beating heart within a patient's chest or the real-time brain activity of a stroke victim, facilitating faster, better diagnoses.

Medical students at the New York University School of Medicine wear NVIDIA 3D Vision glasses to dissect virtual cadavers projected onto screens, with different parts and systems of the body brightly colored for identification. Forensic scientists are experimenting with similar technologies for use during autopsies to unobtrusively determine the cause of death of individuals. The future promises animated, searchable anatomical maps that doctors, researchers and laypeople alike can use as easily as online street maps.

More on Virtual Anatomy | More on Virtual Autopsies

Exploring the Big Bang, Efficiently

Germany's Bielefeld University is delving deep into the physics of matter in the moments after the Big Bang using an NVIDIA GPU-based supercomputer that delivers 125 times the performance of its previous system with much better energy efficiency. While the world's largest particle accelerators, like the Large Hadron Collider, are used to study the state of early matter experimentally, scientists use the Bielefeld University supercomputer to research quarks and other extreme forms of matter through sophisticated simulations in 4D – that is, 3D views modeled over the course of time.

Because GPUs use parallel processing to break down complex problems into many smaller tasks that run simultaneously, a few GPUs can perform certain tasks much faster than a lot of CPUs. Plus, GPUs are engineered for optimal performance per watt, so supercomputers like Bielefeld University's consume less energy for power and cooling, even while realizing incredible performance gains.

More on the Bielefeld University Supercomputer

Mapping the Genome

The field of genomics – the study of an organism's hereditary information – holds the key to personalized medicine and deeper insight into diseases. However, one of the primary hurdles the science faces is managing the enormous amount of data required to evaluate genetic sequences and accurately align many small sequences against entire genomes. GPUs are helping genomics researchers save dramatic amounts of time by using parallel processing to break down complex computing problems into many smaller tasks that run simultaneously.

Projects that once required supercomputers often can now be run on individual machines. By accelerating computations that used to take hours or days, GPUs can enable scientists to interact with their data in near real time. For example, using Life Technology's Ion Torrent solution equipped with GPUs, German and Chinese scientists were able to crack the genetic code behind a deadly strain of E. coli. The researchers sequenced the bacteria's genome in three hours, which helped doctors pinpoint the source of the outbreak, develop a test to detect the disease and mitigate its spread.

Moving Microsurgery Forward

Operating on a person's eye or brain is a delicate business. Conventional microsurgery is limited by the surgical microscope, which constrains a surgeon's field of view and depth perception. To overcome these limitations, a researcher at Johns Hopkins University is using GPU-accelerated technology to create 3D images of the microstructures and tissue planes beneath the viewable surface of the area under a surgeon's examination.

Called interventional Optical Coherence Tomography (OCT), the non-invasive technology provides surgeons with micrometer-level resolutions at ultra-high speeds – there's no waiting for images to be processed – making it highly suitable for guiding microsurgery. GPU acceleration is very cost-effective compared to the overall cost of a conventional OCT system and no optical modifications are required. With detailed 3D images provided in real time, surgeons have a more complete view of a situation and can intervene immediately and more accurately, diminishing surgical risk and improving outcomes.

GPU-Accelerated Cancer Treatment

Radiotherapy has proven to be an effective treatment for many types of cancer, the second-leading cause of death in the U.S. The treatment involves directing a precise yet lethal beam of radiation at a patient's tumor to kill cancerous cells, following a pre-designed treatment plan. This happens over the course of several weeks, making it vital that the treatment plan is adapted to the patient current anatomy, which can vary from day to day due to changes in a tumor's size and shape.

Current radiotherapy methods rely on treatment plans that are up to several weeks old. However, researchers at Moore's Cancer Center at the University of California in San Diego are using GPUs to re-optimize the treatment plan through a dynamic process called online adaptive radiotherapy, or ART. Using GPUs, the medical team can develop a new treatment plan based on patient's new geometry in a couple of minutes, so the radiation is still focused on the tumor itself – significantly reducing the amount of radiation administered to healthy tissues. The outcome of this research, which is entering clinical trials, could have a profound impact on cancer treatment in the U.S. and, potentially, worldwide.

Greening Urban Planning

Currently at 7 billion, the human population is expected to crest 9 billion by 2050, with most of the growth occurring in cities. As urban areas become ever more densely populated, reducing energy use and mitigating air pollution will be critical. For decades, urban planners have attempted to make cities more sustainable through green infrastructure projects, such as parks, alteration of building rooftops and the use of novel paving materials for streets and parking lots. However, understanding the complex interactions among these projects, the environment and urban microclimates on citywide scales is a complicated challenge.

To bring greater insight to this issue, computer scientists at the Universities of Utah and Minnesota are collaborating on developing large-scale simulations of urban environments using extremely fast and inexpensive modeling tools that run on GPUs. This includes an interactive and immersive virtual environment that examines the dynamic physical processes associated with energy use and pollutant dispersion in settings ranging from neighborhoods to cities to metropolitan areas. With a better understanding of these relationships, urban planners can design future projects and policies that optimize green infrastructures and energy conservation while minimizing air pollution in urban landscapes.