Free Trial

Safari Books Online is a digital library providing on-demand subscription access to thousands of learning resources.

  • Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint

GPU-Based Computing

An approach to parallelism that has recently evolved is the use of graphics co-processors as accelerators for computation. This came about because as the requirements for fast and detailed graphical representations evolved, the hardware to implement them began to increasingly resemble hardware that could perform fast parallel floating-point or integer computation. Recently, there have been multiple efforts to export the ability to perform computation on graphics processing units (GPUs) to common programming languages. The most well-known of these are Compute Unified Device Architecture (CUDA) and Open Computing Language (OpenCL). CUDA is specific to Nvidia, whereas OpenCL is supported on GPUs from both Nvidia and ATI.

Although this approach utilizes many cores to perform computations in parallel, the details of the approach are quite different from all the other approaches discussed in this book. The most important consideration is that GPUs represent compute co-processors, and there are several constraints with co-processors.


  

You are currently reading a PREVIEW of this book.

                                                                                                                    

Get instant access to over $1 million worth of books and videos.

  

Start a Free 10-Day Trial


  
  • Safari Books Online
  • Create BookmarkCreate Bookmark
  • Create Note or TagCreate Note or Tag
  • PrintPrint