RT Conference Proceedings
T1 Hardware support for scratchpad memory transactions on GPU architectures
A1 Villegas Fernández, Alejandro
A1 Asenjo-Plaza, Rafael
A1 González-Navarro, María Ángeles
A1 Plata-González, Óscar Guillermo
A1 Ubal, Rafael
A1 Kaeli, David
K1 Ordenadores - Equipo de entrada y salida - Congresos
AB Graphics Processing Units (GPUs) have become the accelerator of choice for data-parallel applications, enabling the execution of thousands of threads in a Single Instruction - Multiple Thread (SIMT) fashion. Using OpenCL terminology, GPUs offer a global memory space shared by all the threads in the GPU, as well as a low-latency local memory space shared by a subset of the threads. The latter is used as a scratchpad to improve the performance of the applications.We propose GPU-LocalTM, a hardware transactional memory (TM), as an alternative to data locking mechanisms in local memory. GPU-LocalTM allocates transactional metadata in the existing memory resources, minimizing the storage requirements for TM support. In addition, it ensures forward progress through an automatic serialization mechanism. In our experiments, GPU-LocalTM provides up to 100X speedup over serialized execution.
PB Springer
YR 2017
FD 2017-08-29
LK http://hdl.handle.net/10630/14425
UL http://hdl.handle.net/10630/14425
LA eng
NO This work has been supported by projects TIN2013-42253-P and TIN2016-80920-R, from the Spanish Government, P11-TIC8144 and P12-TIC1470, from Junta de Andalucía, and Universidad de Málaga, Campus de Excelencia Internacional, Andalucía Tech.
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 21 ene 2026