.. title: PAR Class 1, Mon 2021-01-25
.. slug: class01
.. date: 2021-01-24
.. tags: class
.. category: 
.. link: 
.. description: 
.. type: text

.. sectnum::
.. contents:: Table of contents
..


You
---

After we discuss the syllabus, tell the class about yourself for a minute or two.  What would you like to see in this course.


Material
--------

#. Read the `syllabus <../posts/syllabus.html>`_.

#. Read https://computing.llnl.gov/tutorials/parallel_comp/ for an intro to parallel computing.

#. Some points:

   a. Parallel computing is decades old; there were commercial machines in the 1980s.   I directed two PhD theses in parallel geometry then.  However, then, clocks speeds were increasing so serial machines were more interesting.
   #. Now: physics limits to processor speed.
   #. History of Nvidia.
      
      i. Curtis Priem had designed graphics HW for both IBM and Sun Microsystems.   (For awhile Sun was **THE** Unix workstation company.   They used open standards and had the best price / performance.)
      #. Nvidia designed gaming graphics accelerators...
      #. that just happened to be parallel coprocessors...
      #. that started to be used for nongraphics parallel processing because of their value.
      #. Nvidia noticed that and added more capability, e.g., double precision IEEE floats, to serve that market.
      #. Currently, some of the highest performance Nvidia boards cannot even do graphics because they don't have video out ports.
	 
   a. Intel CPUs vs Nvidia CUDA cores.
   #. Advantages and disadvantages of shared memory.
   #. OpenMP vs CUDA.
   #. Rate-limiting cost usually I/O not computation. 





Homework 1
----------

is online, due next Mon.