[main] . [history] . [methods] . [results] . [future work] . [files] . [who is dope] . [links]

-------------------------------------

DOPE FFT FUTURE_WORK

-------------------------------------

[future_work]

Future Work

There are several paths for expansion of this project. While we considered
the raw computational complexity of each algorithm in terms of
arithmetic/trigonometric operations, this is by no means a complete analysis.
Factors like memory bandwidth and general interaction with the memory hierarchy
(issues like temporal/spatial locality of cached array accesses and any
optimizations that could be performed in this area) must also be evaluated for
a complete picture of performance. Additionally, work could be done to extend
this analysis into the multiprocessor (shared or distributed-memory) regime.
Such work might center on data distribution, avoidance of false sharing and
unecessary invalidations/updates and efficient synchronization strategies.

[/future_work]