• Virtualization of the physical GPU devices. The virtual devices visible to the user map to a consistent set of physical devices, which accomplishes "user fencing" on shared systems and prevents users from accidentally trampling one another.
• Rotation of the virtual to physical mapping for each new process that requests a GPU resource. This provides a method for large parallel tasks to use common startup parameters and still use multiple device targets. I.e., when each new process calls for gpu0, the underlying physical device gets shifted allowing for the next process calling for gpu0 to get the next allocated physical device.
• Ensuring NUMA affinity for GPUs on systems that have multiple memory controllers. NUMA affinity can be mapped between CPU cores and GPU devices. This has been shown to have as much as 6-8% improvement in host to device memory bandwidth.
• Memory-scrubbing to wipe the user's GPU memory after use for security from subsequent users.

This is the official on-line repository for the code from the Graphics Gems series of books (from Academic Press). This series focusses on short to medium length pieces of code which perform a wide variety of computer graphics related tasks. All code here can be used without restrictions. The code distributions here contain all known bug fixes and enhancements. We also provide errata listings for the text of each book.

The gems can be viewed by category, by book, or by author. Gems code can be accessed in a variety of ways: by viewing the code directly from these pages or by downloading the whole website or a book's entire code base.  

SIML ("Single-Instruction, Multiple-LINGO") is a library containing implementations of a fast SIMD algorithm for calculating the LINGO chemical similarity metric. This method, currently implemented for CUDA and OpenCL-supporting GPUs is up to two orders of magnitude faster when run on a GPU than existing commercial implementations of LINGO.

This repository is associated with the GPU Computing Gems, vol. 1 chapter "Large-Scale Chemical Informatics on GPUs" and the publication "SIML: A Fast SIMD Algorithm for Calculating LINGO Chemical Similarities on GPUs and CPUs" (http://pubs.acs.org/doi/abs/10.1021/ci100011z).

 This is a GPU implementation of the treecode/FMM for the Laplace kernel. It includes a CPU implementation with/without SSE, and two different translation operators for the FMM.

 For details please go to the google code page:

http://code.google.com/p/gemsfmm/

The team "Architecture, Géométrie, Perception, Image et Geste" (AGPIG) of Gipsa-lab laboratory has built a platform "Perception Visuelle et Qualité d'Image" with an eye tracker and a platform for the study of  "Adéquation Algorithme Architecture", including computers with graphic boards. The AGPIG team works on visual attention on both static and dynamic (video) images. Those works are based on a set of processing modelling the first steps of the human vision system. In chapter 30 of the book GPU Computing Gems (Emerald Edition) edited by Wen-mei W. Hwu., we present the spatio-temporal visual saliency model implemented on GPU that mimics the human vision system all the way from the retina to the visual cortex. The model uses a saliency map to determine where the source of attention lies within the input scene, which may further be used to initiate other tasks. Here, the CUDA source code for the static pathway of the visual saliency model is linked.