Project Assignment 13
Attention: There is no lecture in week 14 (11/19) and PA 13 is due on
a Monday (11/24), the same time as the RA!
Continue cross-platform I/O and communication benchmarking and prediction
-
Verify the 3 prediction algorithms and error calcuation.
-
Benchmarking MPI-IO collective operations with more parameter combinations (16
data distributions, including 2-D and 3-D, especially some fine-grained distribution).
Repeat the benchmarking and prediction for 2, 4, 8, and 16 processors. Also, create one mixed
prediction where the processor number is combined with the data distribution (2 2-D
distribution and 2 3-D distribution). Compute the standard deviation of errors for each
prediction result point (2 known data points, 3 known data points, etc).
-
Perform prediction on the broadcast results with higher number of processor
number configurations (1, 2, 4, 6, 8, ... 16). Re-measure the broadcast performance
without the ack messages.
-
Finish a preliminary draft of the term project report.
Requirements
- For all benchmarks, schedule processors on the same chasis at the HPC cluster
- Use the machinefile to schedule processors at the OS cluster in the same way as the HPC cluster does.
- Plot standard deviation on the result curve.