Transforming Your Enterprise Magazine

Current editions

Past editions

Putting petabytes to work for your company

Rapidly growing amounts of data represent a real opportunity in a growing number of markets. The opportunity lies in the actionable, mission-critical information that can be derived from that data using sophisticated high-performance computing (HPC) simulation, modeling and analysis applications.

Whether it is an oil and gas firm sifting through petabytes of seismic data in order to be the first to find new drilling sites or an investment management company conducting advanced risk modeling faster than the competition, the ability to process large amounts of data quickly can mean the difference between agile decision-making and getting one-upped by the competition.

• In biosciences, HPC promises quantum advances in genetic sequencing, chemistry research, drug discovery and molecular dynamics.
• Medical equipment manufacturers rely on HPC for CAT scans, magnetic resonance imaging (MRI) and ultrasound diagnostics.
• Digital content creation, electronic design automation and computer-aided design stand to reap huge benefits from increased computational capability.
• The public sector needs next-generation processing capacity to power its search and encryption projects.

The information is there for the taking. The challenge is developing the massive processing power required to put those petabytes to work. While the capacity of storage devices has grown exponentially over the past few years, processor speeds haven’t been able to keep pace, thanks to basic physical limitations.

“We hit a heat wall on processors that prevents us from increasing the clock cycles,” says Kent Koeninger, a Product and Technology Marketing Manager for HP’s Accelerator Program. “The heat grows exponentially: when you double the clock speed, you get four times the heat.”

The way to get around this roadblock is to offload computationally intensive sections of applications to co-processors that are designed to perform a subset of the functions of a general purpose processor, but run them at higher speed. Essentially, these application accelerators are specialized arithmetic sub-processing units added to an existing system.

To use accelerators, applications have to be specially written to take advantage of their capabilities. Think about one of the most common accelerators in use today: the graphical processing unit (GPU) on the video card inside your computer. Computer games are one good example of the type of application that requires such a high level of data processing that game developers write code that specifically takes advantage of the processing capabilities of the GPU.

Keeping it simple

“Using this technology, you can increase the speed of your processing tasks by ten to thirty times,” Koeninger says. “We’re talking about turning days into hours, hours into minutes.”

To enable customers to enjoy these benefits, HP has established the Accelerator Program, based on open standards and accelerator industry partnerships. The program’s goals are to:

• Make sure third-party accelerators will work with HP servers
• Provide services and support to customers as they adopt accelerator technologies

Part of HP’s Catalyst for HPC Innovation Program, the Accelerator Program’s team of experts conduct research and give customers recommendations on choosing the accelerator technology that makes the most sense for their requirements, whether it is a:

• GPU
• Application-specific integrated circuit (ASIC)
• Field-programmable gate array (FPGA)

HP works with the vendors of each of these technologies, and qualifies their products to work in their BladeSystem and rack-mount ProLiant servers. Recently HP has engineered faster, higher-power I/O slots, optimized interfaces and qualified a range of third-party accelerators for space, heat, electrical, BIOS and software compatibility. It also benchmarks results to make sure the right accelerator fits with the right server and application.

“HP has been working with the industry to improve programming languages for accelerators,” Koeninger says. “A few years ago, if you wanted to program a GPU, you had to write in OpenGL.”

The vendors HP works with have developed compilers that enable application developers to use C-like languages when programming accelerators. A more widely understood programming syntax will enable more applications to take advantage of the advanced performance accelerators offer.

“The HP Accelerator Program is a great example of innovation,” says Koeninger. “We provide confidence that third-party accelerators will function to their full potential with HP. Our goal is to act as a trusted partner for our customers.”

Related links