UO-OACISS/apex

Output Teuchos type output for task tree

khuck opened this issue · 2 comments

khuck commented

With the task tree output, generate a text report like this:

each line has name (short), time, percentage of parent, calls, min, max, stddev, <?> and remainders. Maybe output with fancy colors?

|   1st solve: 8.86567 - 41.2112% [19] {min=8.86556, max=8.86587, std dev=0.000101646} <2, 2, 0, 1, 0, 2, 0, 0, 0, 1>
|   |   foo: bar: 8.8655 - 99.9981% [19] {min=8.8654, max=8.86572, std dev=0.000108096} <2, 2, 0, 1, 2, 0, 0, 0, 0, 1>
|   |   |   something: Operation Op*x: 0.159448 - 1.79853% [19] {min=0.143321, max=0.169668, std dev=0.0098303} <2, 0, 0, 0, 0, 0, 1, 3, 1, 1>
|   |   |   |   cudaDeviceSynchronize: 0.152324 - 95.532% [133] {min=0.136254, max=0.162555, std dev=0.0097566} <2, 0, 0, 0, 0, 0, 1, 3, 1, 1>
|   |   |   |   cudaMemcpy: 0.00178795 - 1.12133% [57] {min=0.00176981, max=0.00180203, std dev=1.16672e-05} <1, 0, 1, 1, 0, 2, 0, 0, 1, 2>
|   |   |   |   cudaFree: 5.64397e-05 - 0.0353969% [19] {min=5.5727e-05, max=5.7749e-05, std dev=6.33534e-07} <2, 1, 0, 2, 2, 0, 0, 0, 0, 1>
|   |   |   |   Remainder: 0.00527973 - 3.31125%
khuck commented

mostly working, but need to sort the output by exclusive time in children.

khuck commented

done