The Community for Technology Leaders
Green Image
ISSN: 1556-6056
Yunlong Xu , Xi'an Jiaotong University, Xi'an
Rui Wang , Beihang University, Beijing
Nilanjan Goswami , University of Florida, Gainesville
Tao Li , University of Florida, Gainesville
Depei Qian , Xi'an Jiaotong University Beihang University, Xi'an Beijing
To make applications with dynamic data sharing among threads benefit from GPU acceleration, we propose a novel software transactional memory system for GPU architectures (GPU-STM). The major challenges include ensuring good scalability with respect to the massively multithreading of GPUs, and preventing livelocks caused by the SIMT execution paradigm of GPUs. To this end, we propose (1) a hierarchical validation technique and (2) an encounter-time lock-sorting mechanism to deal with the two challenges, respectively. Evaluation shows that GPU-STM outperforms coarse-grain locks on GPUs by up to 20x.
SIMD Processors, Multicore Processors, Parallel Programming, Run-time Environments
Yunlong Xu, Rui Wang, Nilanjan Goswami, Tao Li, Depei Qian, "Software Transactional Memory for GPU Architectures", IEEE Computer Architecture Letters, vol. , no. , pp. 0, 5555, doi:10.1109/L-CA.2013.4
80 ms
(Ver 3.3 (11022016))