hgpu.org: a year on the Internet
On October 25, 2010, a year ago, the hgpu.org site was launched.
hgpu.org is a resource for people who seek the possibility to accelerate
their computations by means of Graphics Processing Units (GPU).
The emergence of GPU which allow to perform parallel calculations over the last few years has been a prominent advance in acceleration of applications. General purpose GPU computing began with sophisticated graphics languages like Cg and OpenGL but introduction of high-level programming languages such as CUDA and OpenCL substantially
facilitated creation of codes for general purpose calculations on GPU and gave rise to huge number of
applications accelerated with GPU. Using of GPU provides order of magnitude performance gains over central processing units solutions in more cases with low energy consumption and low costs of the computations. Even consumer modern video cards consist of up to 1000 computing units with overall performance up to 5 TFlops.
Within the year hgpu.org has collected and systematized more than 4100 papers concerning programming, applications and hardware for high performance computing on GPU and related parallel architectures, since July 2011 links to video materials about GPGPU have become available for the site visitors.
For the people who want to test applications on different platforms the hgpu.org team design the two-node computing system with commodity NVIDIA and ATI graphics cards. The system can execute OpenCL applications delivered via the Web.
We hope that information and services presented on hgpu.org in the year have been found to be useful for GPGPU community. We welcome any kind of ideas, comments and propositions from GPGPU admirers.