Pay-per-use supercomputing for the masses
14 April 2008Tweet
Interactive Supercomputing (ISC) has launched a high performance computing (HPC) software service for engineers, scientists and analysts who have technical computing applications that will benefit from parallel processing, but don’t have access to high performance, multiprocessor hardware.
While on-demand HPC resources have been available for some time from leading supercomputing research labs and HPC system vendors like IBM, Sun and HP, they have been primarily hardware resources and have required users to acquire and integrate software. Star-P On-Demand is one of the first services to combine high performance, multiprocessor hardware and a parallel computing software platform in a single, on-demand solution, allowing the user to easily connect to a high performance system from their desktop.
Star-P On-Demand is a pay-per-use computing resource – based on the Star-P software platform – that delivers interactive parallel computing power to desktop computer users. Star-P enables faster prototyping and problem solving across a range of security, intelligence, manufacturing, energy, biomedical, financial and scientific research applications. Star-P On-Demand enables engineers, scientists and analysts that use desktop technical computing tools to take advantage of parallel computing by making the technology easy, affordable and available on-demand.
The service – which offers Star-P running on multiprocessor Linux clusters from partner Tsunamic Technologies – is accessible online from anywhere, anytime. With no hardware investment or server software to install, users can get up and running quickly to build, test and benchmark their applications for enterprise deployment.
‘Researchers in pharmaceuticals, finance, government, manufacturing and other industries have a need for high performance computing solutions for algorithm refinement, testing of problems at scale and any number of critical functions in a project lifecycle. We have found that many of these projects are stalled waiting on currently unavailable HPC resources and tools. Star-P On-Demand provides a solution for these organisations,’ said Bill Blake, CEO of ISC.
The service is designed to help scientists, engineers and analysts in three ways: to use on-demand computing resources for project workflow improvements; use Star-P On-Demand to address overflow computing requirements or short-term computing needs without having to invest in new hardware and infrastructure and access hardware and software for long-term production needs when the parallel hardware and software tools are not available internally.
According to Jie Wu, research manager with the Technical Computing group at industry analyst firm IDC, the HPC market is growing strong thanks largely to new entrants coming in at the mid- to low-end of the market. ‘For those new to HPC as well as those on the lookout for migration, "ease of everything" is the number one requirement,’ she said. ‘Star-P On-Demand not only provides easy and flexible access for scientists, engineers and analysts to high-end computing resources without having to go through the lengthy learning curve, but it also eliminates the hefty investments in high-end computing for customers across the board.'
Star-P On-Demand enables researchers to code algorithms and models remotely on their desktops using popular high-level languages like Matlab, and Python, automatically transforming the programs to run on multicore and multiprocessor hardware. The Star-P software eliminates the need to re-program the applications in complex languages such as C, FORTRAN or MPI (message passing interface) in order to run on Tsunamic Technologies’ Linux cluster.Users can access Star-P On-Demand for a test or trial period using a 20 CPU-hour trial account.
They can then access the service in a pay-as-you-go mode for less then $3.00 U.S. per CPU-hour for further development, or they can purchase monthly packages of core hours at discount pricing.