Prof. John Hennesy on the Future of Microprocessor Architecture 3

Posted by igor Wed, 30 Nov 2005 23:55:00 GMT

Prof. John Hennessy, erstwhile Director of the Computer System Laboratory, now President of Stanford University, visited the University of Edinburgh to receive a Honorary Degree and give a public lecture on the Future of Microprocessor Architecture.

The lecture was very well structured and thrilling, especially the part about Thread Level Parallelism and a preview of SUNs upcoming Niagara Architecture. After having spent quite some time with a famous book written by Prof. Hennesy, namely Computer Architecture: A quantitative approach, I was really anxious to see this talk.

Prof. Hennessy led the RISC (Reduced Instruction Set Computer) team at Stanford in the early 1980s and was a cofounder, in 1984, of MIPS Computer Systems, now MIPS Technologies. He is a recipient of the 2000 John Von Neumann Medal, the 2000 ASEE R. Lamme Medal, the 2001 Eckert Mauchly Award and the 2001 Seymour Cray Award. He is a member of the National Academy of Engineering and the National Academy of Sciences, and he is a fellow of the American Academy of Arts and Sciences, the Association for Computing Machinery, and the Institute of Electrical and Electronics Engineers.

So all in all this guy has done quite some stuff for computer science!

Trackbacks

Use the following link to trackback from your own site:
http://blog.solaris.bytelabs.org/articles/trackback/25

Comments

Leave a response

  1. Jakob Praher 5 days later:
    Wow thats great. The niagara is one of the most interesting processor designs I have seen lately. They have threading support as a first class citizen. SUN is going to open up the sources for the chips design (much like the power pc organisation). What was his essence? There is a very nice blog article[http://blogs.sun.com/roller/page/Gregp?entry=don_t_become_moore_confused] by Greg Papadopoulos. cu soon --Jakob
  2. _igor_ 6 days later:
    The basic message was that we are at the end of the raod of ILP (Instruction Level Parallelism). He also said that he doesn't believe that industry will solve this misery and that once again the academics will have to solve this problem since it is not easy to tackle. One also needs to be careful about the niagara - it is a good processor for the server market, but its floating point support is crap. Another factor one needs to consider is the amount of money SUN spent on the development of that processor and how much they can sell of them. Hennesy made a valid point that what just may hurt SUN is that they will not be able to sell enough of their processors which will weaken their market position. We have all seen companies with good ideas (Xerox, DEC etc.) fail because they were either too early with their ideas or the competition just had a better marketing approach. All in all we have a very interesting situation on the processor market and it is thrilling to see how the future will evolve!
  3. Jakob 6 days later:
    hi Igor, interesting point. I think that he/you have valid points, especially regarding market position and market dominance. It is true that it is merely a server processor. But what i think was the problem for DEC was the start of commercialization of software and that in that time that meant to concentrate on one ISA (intel). now i don't know, if you look at todays interenet infrastructure, you will see that most of them work on open source things and it is not that big problem any more to switch ISA then it is/was with binary software. And what I like about the blog of Papadopoulos is what he says by "don't let Moore's law fool you". The number of transistors keeps on doubling every 18 months, but that means that by 2010 we migth not have concepts like "motherboards" and "pci" cards anymore, but have everything integrated on one chip, except for a high speed serial bus which connects to io. Given the current strucutre of academia and looking at the last innvoations I don't really know where the next things will come from. It would be nice if they come from academia, but I think that it is becoming more and more difficult for the academia to come up with a design that meets the current complexities need. I mean sun has worked with stanford to produce the thing they called cmt - chip multithreading. So perhaps we see a more intermingled cooperationn. I am interested in the OpenSPARC.net community. I don't know what sun is really wanting to make here. Maybe we see again more diverse hardware platforms?
Comments