2007-05-08

Ran those tests for Luke.

• Scaled the strong scaling case so it would have about 1M nodes per processor
• Scaled the weak scaling case so it would have about 2M nodes total
• Tested hmis and pmis on 32 processors, same for gauss-siedel solvers

2007-02-11

assumptions in NA

• you need a hihgly technical background in order to do and understand NA
• every porblem is unique and needs to be coded as such.
• speed of the code is the most important factor in an algorithm
• onputing the numbres is the most important factor in scientific computation.