Background on VLSI, Supercomputing, and other engines of growth in computer hardware

Much of the following topics appear in my Computer Systems undergraduate course (CS 3650), as part of the computer architecture component. Since ther is no obvious place for it in the graduate program (since the Computr Architecture course is almost never taught), I'm putting the material here.

VLSI -- How are CPUs built, and when will Moore's Law end?

OVERVIEW:

Semiconductor Fabrication: Photolithography (with a broader overview of semiconductors here)
- And one more source, in still greater detail, of the photolithography process from which our VLSI chips are built: Semiconductor Photolithography - The Basic Process

SPECULATION ABOUT FUTURE CHIPS:

NEWS (9/30/16): Innovation to keep Moore's Law alive, says TSMC co-CEO (Over 300 engineers involved in R&D for 3 nm process: approximately 8 to 15 silicon atoms across)
NEWS (9/20/16): Nichia, Taiwan LED chips makers to start UV-C LED production (UV-C: See subdivisions of the electromagnetic spectrum.)
NEWS (9/19/16): Globalfoundries unveils plan for 7nm FinFET manufacturing (7 nm is approximately 18 or 35 silicon atoms laid end-to-end.)
NEWS (4/25/16): Competition in advanced process technologies market to heat up (from digitimes.com) (10 nm is approximately 25 or 50 silicon atoms laid end-to-end.)
NEWS (4/6/16): Samsung now producing 10 nm RAM chips: (from digitimes.com) (10 nm is approximately 25 or 50 silicon atoms laid end-to-end.)
NEWS (2/16/15): Samsung announces mass production of 14nm FinFET mobile chips (Samsung 8-core smartphones to be faster and lower power) (from digitimes.com)

AND MOTIVATION ABOUT WHAT IS DRIVING THIS:

Motivation for multi-core CPUs: Limits of CPU Power Trends at beginning of the millenium (from here)
A semi-recent article: the future of CPUs for smartphones, and the issue of power: Page 1 of Part 2
I also liked: Page 2 of part 1, talking about the race between Intel, and some fabrication foundries (manufacturing the designs of other companies) such as TSMC, GlobalFoundries, and Samsung. They all want to fabricate with the smallest line-width. Intel and others are working to develop a 10 nm process for the next generation: 10 nm is approximately 25 or 50 silicon atoms laid end-to-end (depending on which measure of the silicon radius is used).
A semi-recent article: the future of CPUs for smartphones, and the issue of power: Page 1 of Part 2

What are the largest supercomputers today?

NEWS: Coral supercomputers (appearing from 2016--2018; see especially the table, lower down in this article)

TOP500 Supercomputer Sites
- Lists of Top 500 supercomputers in the world, with Top 10 on first page.
- Some recent blogs from the TOP500 site:
  - Translating HPC Speak (an irreverent view of High Performance Computing)
  - What is the use of quantum computing without pictures or conversations? (What to do after Moore's Law ends?)
  - How To Kill A Supercomputer: Dirty Power, Cosmic Rays, and Bad Solder

NEWS:

Intel Announces Knights Mill: A Xeon Phi For Deep Learning (see also Coral supercomputers, below)

Deep Learning on NVIDIA GPUs

DeepMind Beats Human Champion at Game of Go (in 2015)

"[The deep learning algorithm of] AlphaGo doesn't actually use that much hardware in play, but we needed a lot of hardware to train it and do all the different versions and have them play each other in tournaments on the cloud. That takes quite a lot of hardware to do efficiently, so we couldn't have done it in this time frame without those resources."