Allinea Software's debugging tools are now available on both the CUDA 6 platform and GE Intelligent Platforms’ systems.
Allinea DDT 4.2.1 has been tailored to offer full support for Nvidia CUDA 6, the latest release of the parallel computing platform and programming model.
Allinea DDT and Allinea MAP are also now certified to run on GE Intelligent Platforms’ systems, allowing developers to debug and profile code precisely for this environment and to ensure that every application performs at its best.
The CUDA 6 platform features unified memory and other new features that make parallel programming easier and enable software developers to dramatically decrease the time and effort required to accelerate their scientific, engineering, enterprise and other applications. The production release of the CUDA 6 Toolkit is available today as a free download on the CUDA website.
‘We worked closely with Allinea Software and PGI to provide world-class debugging capabilities for CUDA Fortran and OpenACC directives-based codes,’ said Duncan Poole, senior manager of Strategic Alliances at Nvidia.
‘Unified memory will be transformative for codes, as it removes the need to manually copy data between the host CPU and accelerator. The ability to debug with Allinea DDT from day one will make a huge difference to developers who need tools that can scale out to their biggest systems and most challenging bugs. With these tools available, users can quickly deploy CUDA 6 in production,’ said David Lecomber, CEO and founder of the software.
In addition to working with Nvidia and PGI Software, Allinea also worked closely with other major compiler vendors to ensure that Allinea DDT also supports their GPU capabilities.
However, High Performance Computing is also reaching out of its traditional setting in large compute clusters and into embedded systems used to run the sophisticated applications – such as modern signal processing applications in defence. This necessitates the need for debugging tools that can be used with embedded systems.
Peter Thompson, senior business development manager at GE Intelligent Platforms said: ‘High Performance Embedded Computing is seeing a rapid convergence with conventional HPC as HPEC faces similar challenges in energy efficiency, performance, heterogeneous architectures and scalability. There is a real need for development tools to provide insight into application performance and software defects that just can't be found with tools designed for single threaded development.’
‘We know that scientists and engineers rely on a new generation of advanced tools for their supercomputing applications,’ said Thompson. ‘These enable them to reduce the time it takes to develop applications, to optimise performance and to minimise both risk and cost.’
