Loongson comments and questions from a HPC friend
Daniel Clark
loongson stuff with. He does a lot of stuff with high performance
computing and supercomputers. Interesting tidbits:
* They were the first site a SiCortex (massively parallel mipsel) box
was deployed; they did a lot of compiler testing. He said that by far
the fastest was from Pathscale Group, now at Cray - MIPS compiler -
*really* good, at some point they were considering making it free
software, although I couldn't find anything online about this. Project
lead was a "Fred Chou" who is amazing.
* He recommended also looking at LLVM for MIPS64; his impression was
that gcc wasn't that optimized for the platform, but that could have
changed in the interim.
* He had some interest in if there were any academic papers in English
on the architecture of existing or future loonson-based
supercomputers. They buy a lot of supercomputers, so IMHO it wouldn't
be out of the question for them to purchase a loongson based one at
some point, esp. if it had a performance per watt ratio similar to the
SiCortex, which they liked a lot (SiCortex was a victim of the US
recession).
Cheers,
--
\|/ Daniel JB Clark | Activist; Owner
FREEDOM -+-> INCLUDED ~ http://freedomincluded.com
/|\ Free Software respecting hardware
Daniel Clark
> was deployed; they did a lot of compiler testing. He said that by far
> the fastest was from Pathscale Group, now at Cray - MIPS compiler -
> *really* good, at some point they were considering making it free
> software, although I couldn't find anything online about this.
Ah, it Looks like "The Linux Fund" was trying to put together a bid,
but didn't beat Cray :-(
http://www.codestrom.com/wandering/pathscale/
yajin
About the papers, there are some papers about Loongson 3, which is not
available at market.
Weiwu Hu, Jian Wang, Xiang Gao, Yunji Chen, Qi Liu, Guojie Li
,“Godson-3: A Scalable Multicore RISC Processor With x86 Emulation”,
IEEE Micro, Mar./Apr. 2009, pp. 17-29
• Weiwu Hu, Jian Wang, Xiang Gao, Yunji Chen, “Micro-architecture of
Godson-3 Multi-Core Processor”, slides of Hot Chips Conference 2008
There are also some papers about loongson 2F, which is used in so many products.
Hu, W. W., Zhao, J. Y., Zhong S. Q., Yang X., Guidetti, E., and Wu C.
Implementing a 1GHz Four-Issue Out-of-Order Execution Microprocessor
in a Standard Cell ASIC Methodology. In Journal of Computer Science
and Technology, Vol. 22, No. 1, pp.1-14, 2007.
ZhangZu-Song Wei-Wu HuFu-Xin Li ,Microarchitecture of the Godson-2
Processor,Journal of Computer Science & Technology, 2005, 20 (2) :
p.243-249
Wei-Wu Hu, Jian Wang, Making Effective Decisions in Computer
Architects' Real-World: Lessons and Experiences with Godson-2
Processor Designs,Journal of Computer Science & Technology , 2008, 23
(4) : 620-632
Also there are one paper about open64 and loongson.
http://www.capsl.udel.edu/conferences/open64/2009/Papers/102-Open64onMIPS2.pdf
Adding the loongson backed to LLVM s really a good idea but it seems
no people are working on this.
Hope it helps.
yajin
2010/7/2 Daniel Clark <dcl...@pobox.com>:
> --
> You received this message because you are subscribed to the Google Groups "loongson-dev" group.
> To post to this group, send email to loongs...@googlegroups.com.
> To unsubscribe from this group, send email to loongson-dev...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/loongson-dev?hl=en.
>
>
Ling Kun
LLVM already have a mips backend
http://www.brunocardoso.cc/blog/?cat=3. But it does not work on
loongson now. and still have bugs.
The question is after make llvm work on loongson, then ?
LLVM likes a compiler, if no one test, update and use it frequently
on Loongson, then after a few months or years, some other guys need to
do this work again : )
yours,
Ling kun