RapidMind today delivered a revised software platform for helping
programmers write high-performance applications for multi-core processors as
painlessly as possible.
While traditional programs written using C++ work fine for most single-core
processors, multi-core chips pose more complex coding challenges. Moreover,
vendor tools lock the application into a specific hardware platform.
RapidMind Platform v2.0 targets those dilemmas to let developers write
software for chips such as the Cell Broadband Engine (Cell/B.E.) and
graphics processor units (GPUs).
The platform allows developers to program in C++ using existing compilers
and tools, and then “parallelizes,” or distributes, data across many cores to
boost the hardware performance. Also, applications created on RapidMind
are hardware-independent and can scale to future multi-core chips.
The problem RapidMind targets is not a new one, but it’s an issue that is
increasingly more imperative to address with the proliferation of multi-core
processors.
RapidMind President and CEO Ray DePaul, who helped bring the BlackBerry to
market for RIM years ago, said that while processor speeds progressed over
the last few decades, software “just went along for the ride.”
When the processors couldn’t get any faster due to power and heat
dissipation, chipmakers turned to multi-core processors as a way to amp
performance.
“There’s a wonderful opportunity to leverage these processors, but frankly
the industry has no idea how to take advantage of them,” DePaul said.
“Whether it’s gaming or high-performance computing, everybody is having
difficulty getting to those performance levels the multi-core processors
theoretically have.”
That’s where RapidMind hopes to make its way in the programming world.
RapidMind partnered with multiple providers of multi-core processing
platforms in creating the RapidMind Platform v2.0, including IBM, which
pre-installs the RapidMind platform on Cell blade servers for its Virtual
Loaner program.
The RapidMind platform v2.0 is available now, with a free developer edition
available for download.
The platform supports Windows XP Pro Vista, Windows Vista and certain Red
Hat, Fedora, Ubuntu and Yellow Dog flavors of Linux for the ATI x1X00,
Nvidia Quadro and GeForce CPUs and Cell/B.E. hardware.
DePaul said RapidMind expects to eventually adapt its platform to let
programmers write Intel and AMD multi-core chips; thanks to a $10 million
venture capital funding round led by Ventures West Capital Ltd., the company
expects to boost its R&D team to accelerate this mission.