JolokiaC++: An Annotation Based Compiler Framework for GPGPUs

Author

Patel, Vibha ; Aggarwal, Sanjeev ; Karkare, Amey

Author_Institution

Dept. of Comput. Sci. & Eng., Indian Inst. of Technol., Kanpur, Kanpur, India

fYear

2014

Firstpage

1134

Lastpage

1141

Abstract

We present JolokiaC++, an annotation based compiler framework which generates high quality CUDA (Compute Unified Device Architecture) code for GPUs. Our contributions include: (1) developing explicit and implicit annotations with illustrations of their use in C++, (2) showing the utility of these annotations by providing comparison code snippets, which demonstrates the ease of programming and performance gains, (3) evaluating their effectiveness on kernels like Blacks holes, Matrix-Vector multiplication, Matrix-Matrix multiplication, Jacobi 1D & 2D, Heat 2D, Vector Addition and Convolution.

Keywords

C++ language; graphics processing units; matrix multiplication; parallel architectures; program compilers; vectors; Blacks holes; CUDA code; GPGPUs; Jacobi 1D; Jacobi 2D; JolokiaC++; annotation based compiler framework; code snippets; compute unified device architecture code; convolution; explicit annotation; heat 2D; implicit annotation; kernels; matrix-matrix multiplication; matrix-vector multiplication; performance gains; programming; vector addition; Arrays; Graphics processing units; Kernel; Optimization; Semantics; Vectors;

fLanguage

English

Publisher

ieee

Conference_Titel

High Performance Computing and Communications, 2014 IEEE 6th Intl Symp on Cyberspace Safety and Security, 2014 IEEE 11th Intl Conf on Embedded Software and Syst (HPCC,CSS,ICESS), 2014 IEEE Intl Conf on

Print_ISBN

978-1-4799-6122-1

Type

conf

DOI

10.1109/HPCC.2014.167

Filename

7056885