Tesla K20X GPUDo you need to take your supercomputer into a harsh environment?  Today’s rugged computers are capable of analyzing huge amounts or real-time sensor data.   This is achieved by harnessing the computational performance of today’s GPU’s.   There are exciting opportunities for using GPU’s to solve problems like image analysis in defense applications.  There is an ever growing amount of video data that needs to be analyzed and processed.  This sort of data lends itself to parallel processing that can be achieved with the massively parallel architecture of GPU’s.  Typically any math-based analysis problem, whether it is in fluid dynamics, medical imaging, computer vision, weather/climate or seismic exploration, can benefit from a GPU.  This computing power can now be moved beyond a typical data center and into harsh environments enabling new applications.  It’s possible to have this equipment mounted in an aircraft, ship, or vehicle, which would have been unheard of when this processing power would have filled a room.  

Is This Right for Your Application?

You may wonder if your application software can take advantage of a GPU.  Today’s latest application software is being written to be GPU aware.  Even consumer-based programs are starting to take advantage of the untapped computing power in computer system GPU’s.  Application-based analysis programs like MATLAB have support for GPU computing.  Many high-level programming languages like C++ or FORTRAN can support GPU computing with available libraries.  One of the keys to optimizing for GPU performance is to analyze where your software spends the most time.  These areas will offer the biggest performance improvement when the algorithm is optimized and parallel.  

Choosing the Right GPU for Your Rugged Application

We can help you understand the difference between the NVIDIA GeForce and Tesla series solutions for your application.  The GeForce can be a good choice for GPU computing.  The higher-end Tesla card should be considered if any of the following will be critical in your application.  

  • Faster double precision throughput
  • larger memory capacity/datasets
  • ECC (error correcting code) protection on memory
  • Rigorous testing and QA for the demanding HPC space
  • cluster-friendly features like monitoring and power-management features
  • GPU direct for fast memory transfers between GPUs, or to third-party network/storage solutions

Chassis Plans has experience in designing GPU computers for rugged conditions.  We can help you design your rugged super computer for wherever you want it to go.   ©2012 NVIDIA Corporation. All rights reserved. NVIDIA, the NVIDIA logo, Tesla, Kepler, and CUDA are trademarks and/or registered trademarks of NVIDIA Corporation. All company and product names are trademarks or registered trademarks of the respective owners with which they are associated. Features, pricing, availability, and specifications are all subject to change without notice.