Helper-thread of a task can hide the memory access time of irregular data on the chip muhi-core processor （CMP）. For constructing a compiler that effectively supports the helper-thread of a task in the multi-core sc...Helper-thread of a task can hide the memory access time of irregular data on the chip muhi-core processor （CMP）. For constructing a compiler that effectively supports the helper-thread of a task in the multi-core scenario based on the last level shared cache, this paper studies its performance stable condi- tions. Unfortunately, there is no existing model that allows extensive investigation of the impact of stable conditions, we present the base of pre-computation that is formalized by our degraded task-pair 〈 T, T＇ 〉 with the helper-thread, and its stable conditions are analyzed. Finally, a novel performance model and a constructing method of pre-computation based on our positive degraded task-pair are proposed. The efficient results are shown by our experiments. If we further exploit memory level parallelism （MLP） for our task-pair, the task-pair 〈 T, T＇ 〉 can reach better performance.展开更多
EDF R＆D is developing a new calculation scheme based on the transport-Simplified Pn （SPn） approach. The lattice code used is the deterministic code APOLLO2, developed at CEA. The core code is the code COCAGNE, deve...EDF R＆D is developing a new calculation scheme based on the transport-Simplified Pn （SPn） approach. The lattice code used is the deterministic code APOLLO2, developed at CEA. The core code is the code COCAGNE, developed at EDF R＆D. The latter can take advantage of a microscopic depletion solver expected to improve the treatment of spectral history effects. However, the direct use of the microscopic depletion solver is computationally very intensive because very small evolution steps （typically 100 MWd/t） are needed to reach a good accuracy, which is not always compatible with industrial applications. In order to reduce the calculation time associated with the use of the microscopic depletion solver, a predictor-corrector scheme has been implemented within COCAGNE. It enables the use of larger evolution steps, up to 1000 MWd/t. Tests show that the predictor-corrector procedure gives fairly accurate results while significantly reducing the calculation time.展开更多
文摘Helper-thread of a task can hide the memory access time of irregular data on the chip muhi-core processor （CMP）. For constructing a compiler that effectively supports the helper-thread of a task in the multi-core scenario based on the last level shared cache, this paper studies its performance stable condi- tions. Unfortunately, there is no existing model that allows extensive investigation of the impact of stable conditions, we present the base of pre-computation that is formalized by our degraded task-pair 〈 T, T＇ 〉 with the helper-thread, and its stable conditions are analyzed. Finally, a novel performance model and a constructing method of pre-computation based on our positive degraded task-pair are proposed. The efficient results are shown by our experiments. If we further exploit memory level parallelism （MLP） for our task-pair, the task-pair 〈 T, T＇ 〉 can reach better performance.
文摘EDF R＆D is developing a new calculation scheme based on the transport-Simplified Pn （SPn） approach. The lattice code used is the deterministic code APOLLO2, developed at CEA. The core code is the code COCAGNE, developed at EDF R＆D. The latter can take advantage of a microscopic depletion solver expected to improve the treatment of spectral history effects. However, the direct use of the microscopic depletion solver is computationally very intensive because very small evolution steps （typically 100 MWd/t） are needed to reach a good accuracy, which is not always compatible with industrial applications. In order to reduce the calculation time associated with the use of the microscopic depletion solver, a predictor-corrector scheme has been implemented within COCAGNE. It enables the use of larger evolution steps, up to 1000 MWd/t. Tests show that the predictor-corrector procedure gives fairly accurate results while significantly reducing the calculation time.
