Replace bigfile mpi #71
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- From mpi4py API and not using bigfile API at all.
- All custom codes are integrated in `abstractsnapshot.py` as new functions and in `get_data()` - Each rank read the data in segments using `mpi_file_handler.Read_at()` - So each bigfile blob should not be accessed by more than on rank as long as `comm.Get_size < num of blob files`, for astrid it is ~ 1300. - Test modules re added in `test` directory which runs on a copy of a bigfile `header` in `example_bigfile`
…spectra into replace-bigfile-mpi
- As to why we use COMM_SELF not COMM_WORLD in `MPI.File.Open()`
…spectra into replace-bigfile-mpi
Since it is Bcast communication, it is better to laod the needed block headers first. Otherwise, we would run into deadlock issue on rank != root
- prelaod all the needed block headers - some tests to check the particle load continuity across segments and ranks
…spectra into replace-bigfile-mpi
- This test can still only be run on a real snapshot not the example uploaded here
- Use non MPI file handle if `mpi4py` is not installed - Even in non-mpi mode, we can't match the data read by bigfile. So either : - There is an stupid bug I cannot think of - The snpashot data is not recorded as we think
…spectra into replace-bigfile-mpi
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Not Ready yet!
I have written bunch of tests in
tests/test_mpi_io.py(it shouldn't pass on GitHub yet because I don't have snapshot test here). The tests pass when running one of the ASTRID snapshots, but the spectra I get do not match what I would get withbigfile. I noticed even when I set MPI_SIZE=1, we don't get similar result, So:ToDO:
AbstractSnapshot.get_data()from this code andbigfile, they size of the output is ~ 1% larger when usingbigfile. So, I'm going to work on this a bit more to see what's going on.