Closed jeff-cohere closed 3 years ago
Merging #99 into master will not change coverage. The diff coverage is
n/a
.
@@ Coverage Diff @@
## master #99 +/- ##
=======================================
Coverage 84.41% 84.41%
=======================================
Files 45 45
Lines 10176 10176
=======================================
Hits 8590 8590
Misses 1586 1586
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 91166e6...66fca30. Read the comment docs.
@jeff-cohere Instead of generating ./tdyprof
from the C code, how about creating a script (say tools/tdyprof
) that reads tdycore_profile.csv
?
Sure, that's easy.
Okay, I've removed the script generation logic and placed tdyprof
into the tools/
directory. Documentation is here.
However, I think I need to handle the fact that different processes generate distinct timing entries for nproc > 1. That doesn't seem too difficult, though.
This was inspired by the pprof tool. We can expand upon it or not. Currently, it just prints out the top N most expensive timings in a columnated style. The script is self-contained, with the CSV data generated by PETSc embedded directly into it, so you can copy it wherever and it'll still work.
Here's what I get when I use it with the Richards demo (built against PETSc 3.14):