The results are divided into two sections, first the results for the symmetric model problem, then the results for the non-symmetric model problem. There is also a distinction between results from PETSc and the results from hypre. Note that some of the scaling studies were done with a fixed total problem size as the number of processors grew, and some were done with a fixed problem size per processor. It will be made clear in each case which scaling study is being represented. All of the tests used one processor per node on the linux clusters, allowing for full utilization of the memory bandwidth and the processor interconnect by that cpu.