GLG: Visual Language for the Development of Parallel Programs in Cuda

Nowadays, parallel program development is paving path for providing an efficient solution to scientific problems. Particularly, the use of graphic cards has gained popularity since a large number of cores are available at an affordable price. However, the development of graphic card programs could be a challenge since the software developers must know the card architecture for its implementation. This paper proposes the GLG (Graphic Language for GPUs) visual language for developing programs in CUDA (Compute Unified Device Architecture). GLG programs could be created easily through icons interconnections validated by a grammar. In addition, a web GLG development environment is proposed that uses a client-server architecture to perform the execution of visual programs.