@phdthesis{dollinger:tel-01251719, TITLE = {{A framework for efficient execution on GPU and CPU+GPU systems}}, AUTHOR = {Dollinger, Jean-Fran{\c c}ois}, URL = {https://inria.hal.science/tel-01251719}, SCHOOL = {{Universit{\'e} de Strasbourg}}, YEAR = {2015}, MONTH = Jul, KEYWORDS = { parallelization ; high performance computing ; GPGPU ; CUDA ; heterogeneous computing ; polyhedral model ; mathematical modeling ; execution time prediction ; performance prediction ; profiling ; runtime ; dynamic ; adaptive code selection ; multiversioning ; scheduling ; load balance between CPU and GPUs ; CPU + GPU ; CPU vs GPU ; energy saving ; virtual machine ; speculative parallelism ; compilation}, TYPE = {Theses}, PDF = {https://inria.hal.science/tel-01251719/file/thesis.pdf}, HAL_ID = {tel-01251719}, HAL_VERSION = {v1}, }