Quantlib GPU Codes
Scott Grauer-Gray, Will Killian, Robert Searles, Dr. John Cavazos
Email Scott Grauer-Gray at sgrauerg@gmail.com if you have any questions.
This contains codes for Black-Scholes, Monte-Carlo, Bonds, and Repo financial
applications which can be run on the CPU and GPU. All the codes are described
in the paper "Accelerating Financial Applications on the GPU" which was presented
at the Sixth Workshop on General Purpose Processing Using GPUs (GPGPU 6) in 2013 and
any work that uses these codes should cite that paper.
Link to paper: http://www.cis.udel.edu/~grauerg/accFinAppsGpu.pdf
See README-parameters.txt for info on the parameters in each application.
Directions (tested for Linux-based systems w/ gcc/g++ compiler, NVIDIA GPU, and CAPS Compiler for HMPP/OpenACC):
For CUDA/OpenCL codes (CUDA available for each application, OpenCL available for Black-Scholes
and Monte-Carlo):
1. Set the PATH and LD_LIBRARY_PATH environment variables to point to the appropriate
locations for CUDA/OpenCL.
2. Navigate to CUDA/OpenCL folder in target application.
3. Run Makefile (by using "make" command).
4. Application executable should be create and can be run.
For HMPP/OpenACC Codes:
1. Set the PATH and LD_LIBRARY_PATH environment variables to point to the appropriate
locations for CUDA/OpenCL.
2. Set the environment variables needed for the HMPP/OpenACC compilation environment.
3. Navigate to HMPP/OpenACC folder in target application.
4. Run Makefile (by using "make" command).
5. Application executable should be create and can be run.
For CPU/OpenMP Codes (assuming using g++ compiler):
1. Run Makefile (by using "make" command).
2. Application executable should be create and can be run.