Message to FPGA Manufacturers

Xilinx and Altera are a very large percentage of the FPGA market.  One way FPGA@Home could succeed is the same way that GPUs became involved in BOINC projects such as SETI@Home.  Nvidia wanted to demonstrate that GPUs via CUDA weren’t just for graphics but for algorithmic acceleration.  Given recent developments on the FPGA side of things, there papers, tools, and presentations on using FPGAs for algorithmic acceleration.  I plan to post some of these papers here as well but trust me they are everywhere!

Now all of these tools manufacturers claim to be wholesale football jerseys better than the other but it is hard to tell which one really is better unless you use it!  See an upcoming post on “My Tool is Better.”

Here is a quote from one of the people I have been corresponding with:Watch Full Movie Online Streaming Online and Download

Around 2007 Nvidia was trying to get into GPGPU computing and had CUDA. They approached Seti@home and asked to develop a CUDA version of their Multibeam science app. They also provided a card to Seti@home. Why Seti@home? Well they are the longest running volunteer computing project and by far the largest viagra 100mg cheap jordans online prix en pharmacie.

Later the optimizers took the existing app and made it even faster and updated it quite significantly to the point where it now runs using next to no CPU resources. Since BOINC added Nvida support as a coprocessor a number of other projects have also developed GPU based science applications. None of this would have happened unless Nvidia took the first step.

All the top hosts in all the projects that have GPU apps sport multi-GPU configurations. Why aren’t they using FPGA’s? Two reaons really. Firstly there isn’t any science app that actually runs on an FPGA. Secondly cheap football jerseys price. To get the price down they need to manufacture in volume, but you need to create the science app to drive the volumes up.

The FPGA manufacturers should consider a similar approach to Nvidia, where they provide some hardware and the development expertise to take an existing science app, such as the Seti@home Multibeam app and get it to work on their brand of FPGA. They could even come up with a “Seti@home FPGA coprocessor kit” that makes the hardware readily available to the hundreds of thousand of volunteers.

Extremely insightful!  Maybe cheap Oakleys sunglasses Xilinx a and/or Altera could get directly involved with the SETI@Home project.  I am cheap nfl jerseys sure a lot of their engineers contribute their PC time to SETI and similar projects already.  Why not ‘show off’ your hardware and tools?  Could make a great white paper for those avid white paper authors!  Hope you step up to this challenge!

 

Leave a Reply

Your email address will not be published. Required fields are marked *