Jump to content

Talk:Sunway (processor)

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

resource

[edit]

http://www.nytimes.com/2011/12/06/science/china-scrambles-for-high-tech-dominance.html?pagewanted=2&ref=science ... excerpt ...

Then, this October, another Chinese supercomputer, the Sunway Bluelight MPP, broke the petaflop barrier — a quadrillion calculations per second — putting it among the world’s 20 fastest computers. This machine proved even more surprising in the West. Not only was it based on a Chinese-made microprocessor, but it also achieved a significant advance in low-power operation. That might indicate the Chinese now have a significant lead in “performance per watt” — a measure of energy-efficient computing that will prove crucial to reaching the next generation of so-called exascale supercomputers, which are computers that will be a thousand times faster than the world’s fastest today, and which are scheduled to arrive by the end of this decade. “This is what Chinese companies need to do,” said Hu Weiwu, a professor at the Chinese Academy of Sciences who is the chief designer of another Chinese family of microprocessor chips. “We can send a spaceship to space. We can design high-performance computers.”

141.218.36.43 (talk) 21:52, 6 December 2011 (UTC)[reply]

Sources?

[edit]

Shouldn't there be kernel patches availiable since the Sunway Blue Light obviously runs Linux? --77.185.160.108 (talk) 02:47, 14 May 2013 (UTC)[reply]


News since 2011 ?

[edit]

So here we are, almost at the end of 2015. Any news? A quick web search doesn't find any. — Preceding unsigned comment added by 80.229.247.139 (talk) 22:35, 15 December 2015 (UTC)[reply]

Chip command testing.

[edit]

I recalled that the 80286 could be crashed if you asked for a 16-bit piece of data at location $ffff in the DS. The Tom chip in the Atari Jaguar. I know swap AX,AX was the NOP command but with these modern CPUs, do people still test for unexpected code? If you exchange the PC with the PC, what happens? Someone would need more hands on but with little if any assembly-language being used, the felide constructors don't allow for tests - Assembly language is required. Maybe others have come across bugs in newer chips? — Preceding unsigned comment added by 213.106.56.145 (talk) 15:34, 26 June 2016 (UTC)[reply]

Change of article name to "Sunway"

[edit]

I've changed the name of this to "Sunway" throughout, per recent publications, particularly http://engine.scichina.com/publisher/scp/journal/SCIS/59/7/10.1007/s11432-016-5588-7?slug=abstract , which I think makes clear that "Sunway" is the Sunway/ShenWei team's own preferred English-language version of the name. -- The Anome (talk) 11:40, 29 June 2016 (UTC)[reply]

FYI: Info on now #1?

[edit]

I think they are talking about the then upcoming #1 ("Tianhe-2A", not to be confused with now number #2 Tianhe-2) while names are different (e.g. "Matrix2000" and "GPDSP"): http://www.nextplatform.com/2015/07/15/inside-chinas-next-generation-dsp-supercomputer-accelerator/

This could be something different (or rumours) with "Each of the optimized GPDSP cores (with both scalar and vector units, dedicated vector memory, and VLIW capabilities) are integrated into one “supernode”" as I hadn't seen VLIW mentioned before.. comp.arch (talk) 14:56, 14 July 2016 (UTC)[reply]