CUDA 4.1 makes it easier to accelerate scientific research with GPUs with key features including

• a redesigned Visual Profiler with automated performance analysis and expert guidance;
• a new LLVM-based compiler that generates up to 10% faster code; and
• 1000+ new imaging and signal processing functions in the NPP library.

The CuSparse library included with CUDA 4.1 has a new tridiagonal solver and 2x faster sparse matrix-vector multiplication using the ELL hybrid format, and the CuRand library included with CUDA 4.1 has two new random number generators.  The CUDA 4.1 toolkit also brings some great improvements to its debugging and performance analysis tools.

Sign up for a webinar to learn more about all the new features & high performance GPU-accelerated libraries!

CUDA 4.1 Toolkit 4.1 Feature Overview Webinar

• For Europe and The Americas: 10am (PST), Wednesday, Feb 1
• For Asia-Pacific and India:  10am (IST) Friday, Feb 3

Delivered by Acceleware’s Developers, who provide real world experience and examples, the training comprises classroom lectures and hands-on tutorials. Each student will be supplied with a laptop equipped with NVIDIA GPUs for the duration of the course. Small class sizes maximize learning and ensure a personal educational experience.

Register before January 13 and receive $250 off your course fee!
Enter promotional code AXTEB2012

HOOMD-blue 0.10.0 adds many new features. Highlights include:

• Added pair.dpdlj which uses the DPD thermostat and the Lennard-Jones potential. In previous versions, this could be accomplished by using two pair commands but at the cost of reduced performance.
• Additional example scripts are now present in the documentation. The example scripts are cross-linked to the commands that are used in them.
• Most dump commands now accept the form: dump.ext(filename=”filename.ext”) which immediately writes out filename.ext.
• Specify rigid bodies in XML input files
• Simulations that contain rigid body constraints applied to groups of particles in BDNVT, NVE, NVT, and NPT ensembles.
• Energy minimization of rigid bodies ( integrate.mode_minimize_rigid_fire )
• Existing commands are now rigid-body aware
• NVT integration using the Berendsen thermostat ( integrate.berendsen )
• Bonds, angles, dihedrals, and impropers can now be created and deleted with the python data access API.
• and more

HOOMD-blue 0.10.0 is available for download under an open source license. Check out the quick start tutorial to get started, or check out the full documentation to see everything it can do.

In this paper we investigate the use of distributed graphics processing unit (GPU)-based architectures to accelerate pipelined wavefront applications—a ubiquitous class of parallel algorithms used for the solution of a number of scientific and engineering applications. Specifically, we employ a recently developed port of the LU solver (from the NAS Parallel Benchmark suite) to investigate the performance of these algorithms on high-performance computing solutions from NVIDIA (Tesla C1060 and C2050) as well as on traditional clusters (AMD/InfiniBand and IBM BlueGene/P).

Benchmark results are presented for problem classes A to C and a recently developed performance model is used to provide projections for problem classes D and E, the latter of which represents a billion-cell problem. Our results demonstrate that while the theoretical performance of GPU solutions will far exceed those of many traditional technologies, the sustained application performance is currently comparable for scientific wavefront applications. Finally, a breakdown of the GPU solution is conducted, exposing PCIe overheads and decomposition constraints. A new k-blocking strategy is proposed to improve the future performance of this class of algorithm on GPU-based architectures.

(Pennycook, S.J., Hammond, S.D., Mudalige, G.R., Wright, S.A. and Jarvis, S.A.: “On the Acceleration of Wavefront Applications using Distributed Many-Core Architectures”,  The Computer Journal (in press) [DOI] [PREPRINT])

• A new LLVM based compiler
• Over 1000 additional image processing function in the NPP library
• A Visual profiler

There is also a new version of Parallel Nsight 2.1 RC2 with support for CUDA 4.1. To download and to find out more follow: http://bit.ly/sRpQvr