Pages Menu
TwitterRssFacebook
Categories Menu

Posted by on May 28, 2012 in Hardware News |

SIMULIA Abaqus GPU Accelerated Direct Solver triples the performance of a Quad Core CPU

SIMULIA Abaqus GPU Accelerated Direct Solver triples the performance of a Quad Core CPU

From FireUser bring us an interesting comparative which explores the benefits of computers with configurations multi-socket/nodo offering professional video cards under the application of simulation: Direct Solver.

Recent versions of Abaqus is compatible with the standard API-accelerated computing (GPGPU) OpenCL, and by using the API in combination with a FirePro graphics card, a yield significantly higher than Quad microprocessor kernel (2.9x), and even greater than that of 2 quad-core microprocessors (2.2X) running in parallel in a configuration multisocket.

http://www.youtube.com/watch?v=nkc-8PA2yR4

Also quad socket configurations tested (16 cores) up to 12 sockets on multiple nodes (48 cores), in this configuration, the performance increases provided by the GPU are very few, because the work is shared between multiple CPU cores , while use is reduced GPU.

SIMULIA Abaqus GPU Accelerated Direct Solver triples the performance of a Quad Core CPU image 2

Link: Video: 2.9x acceleration of direct Abaqus solver using AMD FirePro + OpenCL (FireUser)

Tags: , , , , , , , , , , , ,