Record numerical error in benchmarks and cost + memory estimates for JAX functions #261
Conversation
This file contains hidden or 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
matt-graham
added
the
infrastructure
Contributor
|
This is great @matt-graham! Can we record mean absolute error of harmonic coefficients as well as max absolute error. |
Collaborator
Author
|
@jasonmcewen thanks - I've now updated to also record the mean absolute error. |
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Outdated suggestions cannot be applied.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.
Resolves #254
Updates benchmarks to record round-trip error for forward transforms - that is starting from a set of coefficients and computing a signal by performing inverse transform, computing maximum absolute difference of coefficients recovered using forward transform from signal compared to original reference coefficients.
Also updates benchmarking for JAX functions to use the static cost and memory analysis available for ahead of time compiled functions to get estimates of operation and memory costs, which should give estimates when running both on CPU and GPU, thus giving some way of tracking memory performance on GPU.
The benchmarks have also been updated to use the standard library
tracemallocmodule to try to trace memory usage when running on a CPU rather than the previous usage of thememory_profilerpackage which did give very reliable results. This only gives reasonable values for the NumPy methods (presumably as JAX's memory allocations are transparent to thetracemallocmodule) but as we don't have the cost / memory analysis statistics for NumPy this is still useful.