A Stencil framework to realize large-scale computations beyond device memory capacity on GPU supercomputers

Shimokawabe, Takashi*; Endo, Toshio*; Onodera, Naoyuki; Aoki, Takayuki*

Proceedings of 2017 IEEE International Conference on Cluster Computing (IEEE Cluster 2017) (Internet), p.525 - 529, 2017/09

Stencil-based applications such as CFD have succeeded in obtaining high performance on GPU supercomputers. The problem sizes of these applications are limited by the GPU device memory capacity, which is typically smaller than the host memory. On GPU supercomputers, a locality improvement technique using temporal blocking method with memory swapping between host and device enables large computation beyond the device memory capacity. Our high-productivity stencil framework automatically applies temporal blocking to boundary exchange required for stencil computation and supports automatic memory swapping provided by a MPI/CUDA wrapper library. The framework-based application for the airflow in an urban city maintains 80% performance even with the twice larger than the GPU memory capacity and have demonstrated good weak scalability on the TSUBAME 2.5 supercomputer.

Development of facility for in-situ observation during slow strain rate test for irradiated materials

Nakano, Junichi; Tsukada, Takashi; Tsuji, Hirokazu; Terakado, Shogo; Koya, Toshio; Endo, Shinya

JAERI-Tech 2003-092, 54 Pages, 2004/01

JAERI-Tech-2003-092.pdf:14.05MB

Irradiation assisted stress corrosion cracking (IASCC) is a degradation phenomenon caused by synergy of neutron radiation, aqueous environment and stress on in-core materials, and it is an important issue in accordance with increase of aged light water reactors. Isolating crack initiation stage from crack growth stage is very useful for the evaluation of the IASCC behavior. Hence facility for in-situ observation during slow strain rate test (SSRT) for irradiated material was developed. As performance demonstrations of the facility, tensile test with in-situ observation and SSRT without observation were carried out using unirradiated type 304 stainless steel in 561 K water at 9 MPa. The following were confirmed from the results. (1) Handling, observation and recording of specimen can be operated using manipulators in the hot cell. (2) In-situ observation can be performed in pressurized high temperature water and flat sheet type specimen is suitable for the in-situ observation. (3) Test condition can be kept constantly and data can be obtained automatically for long test period.

Structural analysis of Li-intercalated bilayer graphene on SiC studied by total-reflection high-energy positron diffraction

Endo, Yukihiro*; Mochizuki, Izumi*; Fukaya, Yuki; Takayama, Akari*; Hyodo, Toshio*; Hasegawa, Shuji*

no journal, , 

no abstracts in English