CUDA WEEK IN REVIEW GPU-Accelerated Science

WELCOME TO CUDA: WEEK IN REVIEW News and resources for the worldwide GPU and parallel programming community.

CUDA PRO TIP

Pointer aliasing is an important topic to understand when considering optimizations for your C/C++ code. In this tip we describe a simple way to alter your code so aliasing does not harm CUDA application performance. Read about it on the Parallel Forall blog.

CUDA SPOTLIGHT
GPU-Accelerated Science This week's Spotlight is on Dr. Michela Taufer of the University of Delaware. Michela heads the Global Computing Lab (GCLab), which focuses on high-performance computing and its application to the sciences. Michela comments: "My team's work is all about rethinking application algorithms to fit on the GPU architecture in order to get the most out of the GPU's computing power, while preserving the scientific accuracy of the simulations. This has resulted in many exciting achievements!" Read the Spotlight.
CUDA NEWS
CUDA 6.5 Available Now The CUDA 6.5 Production Release is now available to the public. Highlights include support for 64-bit ARM-based systems and Microsoft Visual Studio 2013 (VC12). Download CUDA 6.5. GPU Technology Conference 2015: Call for Submissions The Call for Submissions for the GPU Technology Conference (March 17-20, 2015) is now open. If you use GPUs in your work, science or research, consider submitting a proposal to share your accomplishments with a global audience. Global Impact Award: Now Accepting Applications The NVIDIA Global Impact Award is an annual grant of $150,000 for groundbreaking work using NVIDIA technology to address social and humanitarian problems. Researchers, non-profits and universities anywhere in the world may apply. Nominations are due Oct. 31. CUDA Coding Challenge Launched in India The NVIDIA team in India has launched a CUDA Coding Challenge, an India-wide contest to encourage parallel programming innovation. The challenge is open through Aug. 31. New Video: GPUs and MATLAB  In a new video from MathWorks, Dan Doherty describes how MATLAB users can leverage GPUs to accelerate computationally-intensive applications in areas such as image processing, signal processing and computational finance. Dan discusses the GPU-enabled functionality in MATLAB and add-on toolboxes, and demonstrates how to integrate custom CUDA kernels into MATLAB.
UPCOMING GPU WEBINARS	back to the top
Aug. 26: CUDA 6.5 Overview and Features, U. Kapasi Sept. 17: CUDA 6.5 Performance Overview, J. Cohen Sept. 24: Convolutional Networks: ML for Computer Perception, Y. LeCun, Facebook & NYU Sept. 25: HOOMD-blue 1.0: MD on GPUs, J. Anderson, J. Glaser, University of Michigan Oct. 15: Essential CUDA Optimization Techniques, C. Mason, Acceleware
NEW ON THE BLOG	back to the top
Parallel Forall: 10 Ways CUDA 6.5 Improves Performance and Productivity, M. Harris Unified Memory: Now for CUDA Fortran Programmers, M. Harris CUDA Pro Tip: Optimize for Pointer Aliasing, J. Appleyard Accelerate R Applications with CUDA, P. Zhao Calling CUDA-accelerated Libraries from MATLAB: A CV Example, J.Knight, MathWorks NVIDIA: NYU Explores Frontier of Data Science, Joins New CUDA Centers, C. Cheij