China aims to try again with petascale super from Longsoon chips

Print Friendly, PDF & Email

The Google nets turned up an article in the MIT Technology Review on the Dawning 6000, the planned successor to the Dawning 5000A super, installed in 2008 at the Shanghai Supercomputer Center (debuted at 11, currently at 19). While the 5000A was originally planned to be built out of an earlier Longsoon, but it didn’t measure up when the time came to build the system and it was build with AMD chips. The Longsoons (also called the Godson family) are a blast from the architectural past based on MIPS, in an eight core configuration

While the quad-core Loongson 3 could find applications in everything from desktop PCs to set-top boxes (the chip incorporates additional instructions designed specifically to speed up multimedia playback), an eight-core version will likely be need for the proposed petascale supercomputer. That version will incorporate four regular cores, along with four “GStera” coprocessors designed especially for mathematically intensive calculations. These coprocessors are especially significant because they are better at handling intensive mathematical calculations, including the LINPACK test, which uses linear algebra to benchmark the world’s fastest supercomputers, and to determine their ranking (and their owners’ bragging rights) in the Top 500 list of supercomputers.

The MIT guys called up our man Jack to get his read on the project; he wasn’t optimistic

[Prof. Jack] Dongarra cautions that it’s pointless to speculate about the performance of the forthcoming Dawning 6000 until benchmarks have been run, not least because the MIPS architecture is nonstandard in high-performance computing. “While I wish them well, I see a lot of challenges to making the whole system work, ” says Dongarra. These challenges include having to adapt the software that Dawning runs.

If all goes well, the new Dawning is expected online by the end of 2010. While the Longsoon family has a checkered past in HPC, the chips are installed in notebooks, netbooks, and consumer PCs aimed at low cost markets.


  1. SiCortex was MIPS. Debian GNU/Linux and others run fine on MIPS. The software’s already ported. *shrug*

  2. Well, porting (as in compiles and runs correctly) is different from running efficiently. MIPS isn’t even a processor family broken out by name any more in the Top500 (it disappeared after June 2004, when it had 3 systems), and the “other” category has 0.2% of the Top500. Hard to argue that there is a current base of effective MIPS application software in HPC. A knowledgeable person might be able to argue that because MIPS is still very popular in embedded applications that getting the software together for drivers and stuff will be less of a challenge, but that’s out of my depth.