What I worked on in week 11:
This week started with finishing the cleaner r.proj approach. I ran the full benchmark matrix for it after finishing the code implementation and testing if the output was correct. I benchmarked 14 scenarios covering three projections, both memory settings, and all interpolation methods, with every phase of the run timed. The new sizing approach within this new implementation cut the sizing time nicely and also helped bring speedup times for maps that didn’t do as great before. With the benchmarks done I opened it as draft PR #7807 with the full results table so it can be compared against PR #7627 directly.
On PR #7785, I got mentor feedback and addressed it. Some test cases produced the same results as the default on the test DEM, so I removed them. I also removed unnecessary overwrite flags. The old testsuite used to check category labels, so I added that back into the landform test. After that the PR was merged. Since the tests were now in main, I rebased PR #7783 over it and added the parallel identity tests that were waiting on this merge. These tests run the module with different thread counts, band sizes, and masks. Then they check that every run gives the same output as a single threaded run.
I also started updating my project wiki page for the final report. I cleaned up the abstract, scope, timeline, and PR log to match how things actually ended up.
What I’m doing in week 12:
I’ll be finishing the final report on the wiki page and sharing it with my mentors, and then submitting the final work product. While working on the wiki page I’ll be responding to reviews on the open PRs.
What’s blocking me? Nothing this week
Thanks,
Kaushik