Please continue to follow the general
advice on reading assignments from the
first week's assignment.
Reading assignment:
Lecture 6: Semiconductor
technology basics:
This lecture will be an overview of the basic concepts, structures,
and processes of semiconductor technology, for the benefit of those students
who may not already be familiar with it.
As usual, the text from Feyman's physics of computing course offers
a good elementary introduction to the subject:
-
Sections 7.1, "The Physics of Semiconductor Devices", and 7.3, "VLSI Circuit
Construction", in Anthony Hey (ed), Robin Allen (ed), and Richard Feynman,
Feynman
Lectures on Computation, Perseus Books, Sep. 1996. (In bookstore.)
The following very briefly describes ordinary CMOS logic circuits and reviews
the contributions to their energy dissipation.
This is a more detailed technical discussion of the physics of semiconductor
MOSFET transistors.
If you find you need more background reading on this subject than the above,
let me know and I can try to refer you to library books and other resources.
Also, if you already know this subject area and can recommend a good short
introductory article to distribute to the class, I would appreciate a suggestion.
Lectures 7+8:
Semiconductor technology scaling laws & limits:
More on Moore's law, in case anyone hasn't had enough yet. Seriously,
it's a pretty good article that I should have included in week 1.
-
Robert R. Schaller, "Moore's Law: Past, Present, and Future", IEEE Spectrum,
June 1997, pp. 53-59. Will be on reserve.
The following is my own analysis, based on the semiconductor industry's
roadmap, of the minimum energy dissipation per operation of CMOS circuits.
This is the primary factor determining how many raw computational operations
can be performed per second per unit surface area available for cooling;
that is, how much computational power can you cram into a box of a given
size.
You might also look at section 7.9 of the thesis - "Scaling SCRL to future
technology generations." You don't know what SCRL is yet, so ignore
the sentences about it and just focus on the ones about (ordinary) irreversible
CMOS.
This next article by Meindl is an excellent survey of the constraints
that come into play at many levels - from low-level physics to systems
issues - that may impact further scaling of semiconductor technology over
the next few decades.
-
James D. Meindl, "Low Power Microelectronics: Retrospect and Proposect,"
Proceedings
of the IEEE, 83(4):619-635, Apr. 1995. Will be on reserve.
Carver Mead's is another article along similar lines.
See first-hand for yourself what the international semiconductor industry
is planning to accomplish over the next 14 years:
Additional background material:
Nothing special for this week. If you start to feel lost on something,
and you want to fix the problem, ask me for more background reading material
and I'll try to help you out.
Written assignment: (due Mon. 1/31)
This is the same as the
first week's written assignment, except that it should be on the subject
of this week's lectures and reading material, and it's due next
Monday.
Also, there is one new option: You may respond to any comments I wrote
on your week 1 homework paper (which should be returned in class on Monday
1/24).