Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add vectorization support to Flux map function #4186

Closed
wolffcm opened this issue Oct 26, 2021 · 0 comments · Fixed by #4329
Closed

Add vectorization support to Flux map function #4186

wolffcm opened this issue Oct 26, 2021 · 0 comments · Fixed by #4329
Assignees

Comments

@wolffcm
Copy link

wolffcm commented Oct 26, 2021

With #4184 and #4185 complete, it should be possible to execute very simple vectorized functions from map in Flux.

The work of this issue will be to execute lambdas using the compiler package, passing the table chunk as input, by creating an instance of values.Value that is the table chunk as an object with vector fields. An example of this is here:
https://github.com/influxdata/flux/pull/3930/files#diff-ca52e5e5693fff638788b406cc69092210f978d482989588503142cf525b95f4R288

When the function has completed executing, the map transformation should convert the output value into an instance of flux.Table.

Add a new feature flag that will make map execute the vectorized version of a function if it exists.

DOD:

  • New feature flag added
  • map uses vectorized function if it exists and feature flag is enabled
  • There should be existing acceptance tests that demonstrate that this all works end-to-end. They might already exist. If they don't, add them
@wolffcm wolffcm added this to the Sprint 21-Q4-3 milestone Nov 1, 2021
@jsternberg jsternberg self-assigned this Dec 7, 2021
jsternberg added a commit that referenced this issue Dec 8, 2021
The columnar version of map will run map over an entire table chunk
rather than running it and then appending per row. This version of map
should be more efficient in the normal cases and be more amenable to
vectorized functions.

Related to #4186.
jsternberg added a commit that referenced this issue Dec 8, 2021
The columnar version of map will run map over an entire table chunk
rather than running it and then appending per row. This version of map
should be more efficient in the normal cases and be more amenable to
vectorized functions.

Related to #4186.
jsternberg added a commit that referenced this issue Dec 8, 2021
The columnar version of map will run map over an entire table chunk
rather than running it and then appending per row. This version of map
should be more efficient in the normal cases and be more amenable to
vectorized functions.

Related to #4186.
jsternberg added a commit that referenced this issue Dec 8, 2021
The columnar version of map will run map over an entire table chunk
rather than running it and then appending per row. This version of map
should be more efficient in the normal cases and be more amenable to
vectorized functions.

Related to #4186.
jsternberg added a commit that referenced this issue Dec 8, 2021
The columnar version of map will run map over an entire table chunk
rather than running it and then appending per row. This version of map
should be more efficient in the normal cases and be more amenable to
vectorized functions.

Related to #4186.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging a pull request may close this issue.

3 participants