Prof Mori briefly explained the C’ (pronounced "C-dash" – it’s an acronym for "C-like Design Automation Shell") language which he has been developing in collaboration with Prof Uehara and some of their research students. He told us that C’ was Prof Uehara’s invention, and that they currently teach it to their graduate students.
C’ is based on the interpreted language Tcl and is "a CPU core design tool" aimed at designers of custom CPUs within embedded systems. Using C’ a designer can declare a set of registers and define a complete instruction set architecture. The output of the C’ design automation shell is a Spec C description. Prof Mori said that they have also been developing a tool to transform a C’ description into a Verilog HDL description, giving them a route to implementation, and that they are very interested in the prospect of transforming a C’ description into Handel-C. He added that they have not tried to write a compiler to generate netlists directly from C’; he felt that this would require a lot of effort and experience. According to a set of Powerpoint slides he gave us, it seems they have also been working on transforming a C’ description into NTT’s Structured Function description Language (SFL) [http://www.kecl.ntt.co.jp/parthenon/index_e.htm].
He said that there have been research publications on C’ in Japanese, but nothing yet in English.
As far as teaching is concerned, Prof Mori said that undergraduate students in his department currently undertake a third year course in which two hours per day per week for two weeks are spent designing a counter. He said he hoped that Handel-C might provide a way for the students to actually generate a netlist for this, which is something they don’t attempt at the moment.
He mentioned that as well as using standard Verilog-HDL tools they are also using the freely available "Persimon" tools from NTT. (I subsequently searched for information on the WWW about this, but couldn’t find any.)
He said that he had attended several seminars on Spec-C but was rather disillusioned with it because it is "not connected from top to bottom".
Quoting from a research publication he gave us entitled "Generating Spec C sources in C’" -written in Japanese but with the abstract translated into English - "Spec C is the most popular language [for hardware/software co-design] in Japan, because STOC (Spec C Technology Open Consortium) supports it". [STOC Home Page: http://www.specc.gr.jp/eng/index.html]
Prof Mori told us that another of his research projects is to do with fault-tolerant architectures for WSI (Wafer Scale Integration). They have designed a "stream-oriented" architecture that incorporates pipelining and a voting mechanism. He gave us a copy of a research publication written in English about it.