Update source files for photon emitter, custom ptracer, and relevant examples #2
Conversation
- edit notebook to match python script with same name - update custom_ptracer.cpp based on updates to ptracer.cpp - add specific results files to .gitignore
andrewgait
changed the title
MPhys/Single_Emitter/single_emitter_test_new_change_nphotons.py
Outdated
Show resolved
Hide resolved
- move csv files into csv / test_data directories - edit scripts accordingly to match this - delete unneeded files
|
Everything is now updated, but the cluster is currently down for maintenance (Thursday 17th October). Once it is back tomorrow I hope to rerun the scripts on there to (a) get results for the updated version of mitsuba3, and (b) get log-log plots for scaling performance. Once these results and plots are created, this PR will be good to merge. I have also suggested some further investigative work of performance in the following tickets: |
|
Experiments on the cluster are now complete and this branch (which has got a bit ridiculous in terms of the number of files) is now going to be merged. I have obtained new figures compared to those detailed in the issue which I have written up into a new directory structure for experiments within the MPhys/Single_Emitter directory (hence the large number of extra files and a certain amount of duplication). Given that the MPhys directory is unlikely to ever be merged into the main branch of Mitsuba itself, I don't think there's any harm in leaving the dev branch like this, and we could also choose (eventually) to keep the MPhys directory off the main branch of this fork, to make it easier to do any PR of code that (eventually) could end up in the main branch of Mitsuba. The experiment writeup is at https://github.com/UoMResearchIT/mitsuba3-manchester/blob/1-update-to-include-manchester-photon-emitter-code/MPhys/Single_Emitter/performance_scaling_experiments/experiment_01_scaling_n_photons.md. |
Updated source files from previous Manchester fork of Mitsuba3 for photon emitter and custom ptracer to match recent changes in mitsuba3 and drjit.
TODO: