Stories
Slash Boxes
Comments

SoylentNews is people

Submission Preview

No link to story available

China's New Supercomputer Uses a 260-Core Chip

Accepted submission by takyon at 2016-06-20 12:33:25
Hardware

[merge with anubi TOP500 story]

HPCWire received a report about Sunway TaihuLight [top500.org], the world's new #1 supercomputer system on the June 2016 TOP500 list [top500.org], in advance, and has some details about its architecture [hpcwire.com]. The system uses the native/homegrown SW26010 "manycore [wikipedia.org]" processor instead of Intel's similar Xeon Phi chips. Each SW26010 has 260 cores divided into four groups, reaches about 3 teraflops of peak floating point performance, and can access 8 GB of DDR3 memory. There are 40,960 of these chips, for a total of 10,649,600 cores. The system's efficiency is around 6.05 gigaflops per Watt, over three times more efficient than the Tianhe-2 supercomputer. Although the TOP500 and Green500 lists are due to merge, the Green500 list has not been published yet [top500.org]. As for what the system will be used for:

The system software includes Sunway Raise OS 2.0.5 based on Linux as the operating system. Dongarra's report also mentions basic compiler components, such as C/C++, and Fortran compilers, an automatic vectorization tool, and basic math libraries. Sunway OpenACC supports OpenACC 2.0.

The Chinese supercomputing leadership is targeting the new Sunway machine at four key areas: advanced manufacturing (CAE, CFD), earth system modeling and weather forecasting; life science, and big data analytics.

China has been called out in the past for putting hardware ahead of software development. China announced that is has (at least) three applications that are on the finalist list for the Gordon Bell Award, which will be announced at SC16. The accepted submissions include a fully-implicit nonhydrostatic dynamic solver for cloud-resolving atmospheric simulation; a highly effective global surface wave numerical simulation with ultra-high resolution; and a large scale phase-field simulation for coarsening dynamics based on Cahn-Hilliard equation with degenerated mobility. The report from Dongarra notes that all three applications have scaled to about 8 million cores, just under 80 percent of the total system.

The performance of the #500 system on the TOP500 list has risen from 206.3 to 285.9 teraflops. 94 systems now have an RMAX of over 1 petaflops, compared to 81 systems in November 2015.

More coverage of the list [nextplatform.com] and Sunway [nextplatform.com] is available at The Next Platform.


Original Submission