A Big Step Toward Exascale Application Programmability and Performance

Print Friendly, PDF & Email
On May 15, 2012, NVIDIA announced several innovative technologies aimed at improving the performance and energy efficiency of the Kepler GPUs, along with opening up a new class of applications and algorithms – creating a larger ecosystem of developers and applications for the GPU landscape.Two of these innovations in particular caught our attention.

The first is Hyper-Q—a technology that enables multiple CPU cores to simultaneously use the CUDA architecture cores on a single Kepler GPU. The idea here is to dramatically increases GPU utilization while slashing CPU idle times and advancing programmability. Hyper-Q is enabled in CUDA 5, and we can expect to see it used heavily with standard MPI codes. Hyper-Q offers great promise for dealing with concurrency issues as we scale to much larger systems.

The other technology innovation announced was Dynamic Parallelism. The concept here is to enable GPU threads to dynamically spawn additional new threads, allowing the GPU to ‘adapt dynamically’ to the data. This will greatly simplify parallel programming, enabling GPU acceleration of a much broader set of popular algorithms.

To see the NVIDIA news release, go here: NVIDIA Pioneers New Standard for High Performance Computing.

Be sure to watch the video panel discussion from, “Exascaling Your Apps” – a panel from the GPU Technology Conference moderated by the Exascale Report’s Mike Bernhardt.

Following that panel, we caught up with Steve Scott, CTO of NVIDIA’s Tesla Business Unit to discuss these innovative technologies and their implications for exascale.

Click here to listen to the audio podcast with NVIDIA’s Steve Scott.

For related stories, visit The Exascale Report Archives.

Be sure to watch the video panel discussion from, “Exascaling Your Apps” – a panel from the GPU Technology Conference moderated by the Exascale Report’s Mike Bernhardt.

Watch it now on YouTube