Tag: GPU acceleration
A recent effort led by Cycle Computing based on the SHOC benchmark revealed equal performance between GPU-accelerated cloud and native hardware.
This week we talked to Dale Southard of NVIDIA about the challenges and benefits of providing access to GPU resources in a cloud enviornment. We touch upon issues related to performance, CUDA, future directions, and Dale’s own experiences in HPC and how they carried over to his work with GPU clouds.
MATLAB users with a taste for GPU computing now have a perfect reason to move up to the latest version. Release R2010b adds native GPGPU support that allows user to harness NVIDIA graphics processors for engineering and scientific computing. The new capability is provided within the Parallel Computing Toolbox and Distributed Computing Server.