|
The effect of a large NRES is mainly due to the fact that the results file stores the load step, time, and cumulative-iteration numbers in an array dimensioned to the NRES value. As each solution is written to the file, the values in this array are updated, which requires that the array be read in, updated, and written back out. This is done to make it possible for another process to read the RST file at the same time as it is being generated, a very valuable feature. Setting NRES=1e6 with 2000 result sets, for example, would require a lot of I/O and, as a consequence, elapsed time. When the model is very small, this I/O can consume the majority of the time.
|
|
