WIP --- Temporal reduction profile

[burlen]

time each stage in the app.
this may need work/cleanup before merge
this info is already captured by the profiler.

[burlen]

Fastest

Average

Slowest

Takeaway: The temporal reduction is much faster on the GPU. I/O is slower, and has a lot more variability when GPU is used. Timing captures everything within execute of each stage

[burlen]

varying steps per request (1 reduce thread, 1 writer thread)

varying steps per request (4 reduce thread, 2 writer thread)

[burlen]

round 2 steps per request

I redid the tests this time going to larger steps per request. The same patterns appear.

varying steps per request (1 reduce thread, 1 writer thread)

varying steps per request (4 reduce thread, 2 writer thread)

[burlen]

single node w. MPI

[burlen]

new vs old

[burlen]

[burlen]

[burlen]