[comp.edu] High Performance Computing course

jab@duke.cs.duke.edu (John A. Board) (10/14/87)

I plan to teach a course next spring with a title like "high performance
computing" as an excuse to learn some more about advanced computer
architecture in theory and in practice.  The course is at graduate / 
advanced undergraduate level; our standard (also grad / advanced undergrad)
computer architecture course will be a prerequisite.

I will certainly cover trendy topics like RISC designs and other pipelining
strategies, hardware and software aspects of vector processing, fast
arithmetic techniques, SIMD computing, and real and paperware MIMD computing.
As we have a BBN Butterfly and an INMOS Transputer system at Duke, I will
discuss shared-memory and message-based parallel processing systems.
I like to illustrate material with as many real examples as possible, so
lots of actual architectuers will be dissected.  I don't yet have a detailed
syllabus, and I hope for some input from the net to improve my initial ideas.

Questions: 
1) Who has given / received a course like this before?  Do you know of a
   better / different syllabus?  Any pet topics that should be covered?

2) Is there a reasonable textbook covering some fraction of this
   material?  Any comments on these books (I haven't seen any of them yet):
     Supercomputer Architecture, by Paul B. Schneck, Kluwer Publishers, 1987.
     High Performance Computer Architecture, by Stone, Addison-Wesley, 1987.

Please email any advice/suggestions, etc.  I will summarize for anyone
interested!  Thanks.

-- 
John Board                                   INET: jab@cs.duke.edu
Assistant Professor                          UUCP: ...!mcnc!duke!jab
Dept. Electrical Eng'g and Dept. Comp. Sci.  BITNET: DBOARD@TUCC
Duke University, Durham NC USA               AT+T: +1 (919) 684-3123

steve@nuchat.UUCP (Steve Nuchia) (10/15/87)

In article <10409@duke.cs.duke.edu>, jab@duke (John A. Board) writes:
> I plan to teach a course next spring with a title like "high performance
> computing" as an excuse to learn some more about advanced computer

> 2) Is there a reasonable textbook covering some fraction of this
>    material?  Any comments on these books (I haven't seen any of them yet):
>      Supercomputer Architecture, by Paul B. Schneck, Kluwer Publishers, 1987.
>      High Performance Computer Architecture, by Stone, Addison-Wesley, 1987.

Haven't seen those, but:

	The Architecture of High Performance Computers
	Roland N. Ibbett, Springer-Verlag 1982

		Good reference for the comparative anatomy of machines
		that held the high ground through history, with particular
		attention to the design tradeoffs and the historical context
		in which they were made.  Coverage through the CRAY-1.
		Recommended for students, particularly in conjunction with
		a more theoretical text.

	Computer Architecture and Parallel Processing
	Kai Hwang and Faye Briggs, McGraw Hill 1984

		I had a class from Professor Briggs and taught a lab
		for him.  He's a sharp guy, but the book really drags.
		I haven't managed to even get started good on it.  It
		seems to have good coverage of areas the authors are
		interrested in (Briggs is into caches, for instance)
		and more perfunctory (and dry!) treatment of other
		areas.  Not recomended for students until they need
		excrutiating detail - good reference for some topics.

-- 
Steve Nuchia	    | [...] but the machine would probably be allowed no mercy.
uunet!nuchat!steve  | In other words then, if a machine is expected to be
(713) 334 6720	    | infallible, it cannot be intelligent.  - Alan Turing, 1947