17 comments

[ 2.9 ms ] story [ 49.6 ms ] thread
If you want a great introduction to, and appreciation of, SoCs — given the announcements yesterday and to understand what goes into something like the M1 a little better.
Here are the lecture topics [1] more in-depth:

  1 Basic Processor & Memory hierarchy
  2 Advanced Out-of-Order Processor
  3 Data-parallel processors
  4 Micro-controller introduction
  5 Multicore
  6 RISC-V core 
  7 Advanced Multicore
  8 Multicore programming 
  9 Graphics Processing Unit (GPU)
  10 Heterogeneous SoC
  11 GPU Programming 
  12 Application-Specific Instruction-Set Processor (ASIP) 
  13 PULP: Parallel Ultra-Low-Power Computing 
  14 Architecture in the Future - Wrap-up
[1] https://iis-people.ee.ethz.ch/~gmichi/asocd/classinfo/ASoCD_...
Lecture 7 seems to have gone missing. Lecture 13 is blank. Actually, I think lecture 14 is actually lecture 13 and so 14 has gone missing.
What's the recommended textbook for this course? All I see is "PH 5th ed. 2013, (HP 5th ed. 2013)" and I have no clue what that is.
Computer Organization and Design: The Hardware/Software Interface by David A. Patterson and John L. Hennessy, ISBN 978-0124077263.
Catch em all:

> wget -m -p -E -k -K -np https://iis-people.ee.ethz.ch/\~gmichi/

AKA wget --mirror --page-requisites --adjust-extension --convert-links --backup-converted --no-parent ...
is it possible to do the same with curl?
I ran it and its not very server friendly. I am basically DDoSing the server. On the other hand, it finished within a minute (216MB) ....
Meh, at least it downloads one file after the other and not in parallel.
The Internet tells me that the collective noun of flag is bunting. I think this applies here: a bunting of wget flags.
(comment deleted)
Does anyone have courseware on non-advanced System-on-Chip design?
That's a very good lectures!

However, there is one thing is regrettable to me in lectures that talks about the micro-architecture of superscalar processors: they never talk about the register bank micro-architecture. The focus is always on the micro-architecture of the compute units.

Knowing that the wire area of a conventional register bank grows by the square of the number of read/write ports, and that superscalar processors require an increasing number of read/write ports, the micro-architecture of the register bank quickly becomes critical with the number of issue.

But there is very little information on the right micro-architectural strategy in this regard. There is some information on GPU register banks, but very little about modern CPU designs.