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

Decimal128 upmerge #4056

Merged
merged 31 commits into from
Nov 8, 2021
Merged

Decimal128 upmerge #4056

merged 31 commits into from
Nov 8, 2021

Conversation

revans2
Copy link
Collaborator

@revans2 revans2 commented Nov 8, 2021

Upmerge to latest 21.12 code

gerashegalov and others added 28 commits November 1, 2021 08:59
- Replaces copy-current executions with a single line inside the for loop in ant
- Factors the common aggregator dependency out of all profiles

Closes NVIDIA#3975 

Signed-off-by: Gera Shegalov <[email protected]>
Closes NVIDIA#3470. Adds support to run RAPIDS Accelerator on top of Spark 3.2.1-SNAPSHOT

Signed-off-by: Gera Shegalov <[email protected]>
* Fix aggegator jar copy logic

1) Make dependency:get conditional on whether the jar is being created
   in the current build's package phase, fallback on the previously
   installed
2) dependency:get should receive jenkins/settings only on CI

3) propagate maven.repo.local if set

Closes NVIDIA#3987

Signed-off-by: Gera Shegalov <[email protected]>

* Prefer files in the build dirs

Signed-off-by: Gera Shegalov <[email protected]>
* Allow null character in regexp_replace pattern

Signed-off-by: Andy Grove <[email protected]>

* Add more null character syntax examples to test

Signed-off-by: Andy Grove <[email protected]>

* Rename isNullOrEmptyOrRegex to isSupportedStringReplacePattern and invert logic

Signed-off-by: Andy Grove <[email protected]>

* Simplify isSupportedStringReplacePattern implementation
* Add integration test for RLike with embedded null in input

Signed-off-by: Andy Grove <[email protected]>

* enable test and update docs

* fall back to CPU for null character in regular expression
…GPU (NVIDIA#3862)

* Implement fix and basic test

Signed-off-by: Andy Grove <[email protected]>

* improve test based on PR feedback

* check tags more consistently

* Add test that does not depend on CAST of array falling back to CPU

* add license header

* simplify test to use spark.rapids.sql.hashAgg.replaceMode

* Update comments

* revert plugin changes

* fix some regressions

* WIP temporarily allow ObjectHashAggregate/SortAggregate/Sort to allow binary input to see what other issues remain

* scalastyle

* add placeholders for BinaryType checks

* ps notes and type checks

* enable more tests

* remove redundant and untested type check

* add test for sort fallback to cpu with binary input

* test for SortExec with BinaryType

* revert changes to aggregate.scala

* remove ps note for SortExec BinaryType
* Add shim for Databricks 9.1

Signed-off-by: Jason Lowe <[email protected]>

* Commonize files between 311db and 312db

* 312db requires override on TernaryExpression

* Commonize more code

* Add acronym and treenode details to shim v2 docs

* Add shuffle manager docs

* Fix missing code in poms

* Upmerge fixes

* Remove unused VERSIONNAMES in 311db and 312db shims
* fixed pool size

Signed-off-by: Rong Ou <[email protected]>

* address review comment

Signed-off-by: Rong Ou <[email protected]>

* fix test

Signed-off-by: Rong Ou <[email protected]>

* really fix test

Signed-off-by: Rong Ou <[email protected]>

* review feedback

Signed-off-by: Rong Ou <[email protected]>

* more accurate pool size

Signed-off-by: Rong Ou <[email protected]>
* Stddev Windowing

Signed-off-by: Raza Jafri <[email protected]>

* addressed review comments

Signed-off-by: Raza Jafri <[email protected]>

* removed row limit for tests

Signed-off-by: Raza Jafri <[email protected]>

* added unbounded window

Signed-off-by: Raza Jafri <[email protected]>

* removed row limit from the range window test

Signed-off-by: Raza Jafri <[email protected]>

* addressed review comments

Signed-off-by: Raza Jafri <[email protected]>

* Reverting test changes

These changes were causing test_lead_lag_for_structs_with_arrays
test to fail. I am not sure why but seems to be happening only for
the case where int_gen is nested in the array_gen.

Signed-off-by: Raza Jafri <[email protected]>

* put lazy val back in for evaluation

Signed-off-by: Raza Jafri <[email protected]>

* replace Expr Check with aggNotReduction

Signed-off-by: Raza Jafri <[email protected]>

Co-authored-by: Raza Jafri <[email protected]>
…IDIA#4021)

Refactoring accidentally used the shimmed implementation class instead
of facade.

Need a separate issue to remove consolidate this code in a single class
as it's currently replicated a few times in the repo

Signed-off-by: Gera Shegalov <[email protected]>
* Allow changing yarn resource gpu name

Signed-off-by: Thomas Graves <[email protected]>

* fix initialize from task

* fix

* cleanup

* Doc updates

* update configs

* Add test

* Add test for throws
…pynb'[skip ci] (NVIDIA#4014)

* add required property:display_name

Signed-off-by: liyuan <[email protected]>

* add comma

Signed-off-by: liyuan <[email protected]>

* touch a new notebook file

Signed-off-by: liyuan <[email protected]>
* Update buildall script and the build doc

Signed-off-by: Gera Shegalov <[email protected]>

* Reviews

Signed-off-by: Gera Shegalov <[email protected]>
* Check for CPU cores and free memory to set TEST_PARALLEL

Signed-off-by: Chong Gao <[email protected]>
* Return batch with rows and no columns for empty resultExpressions in the aggregate

Signed-off-by: Alessandro Bellina <[email protected]>

* Add integration test for no resultExpression aggregate

* Cleanup on where the data types are coming from

* Remove some extra materialization that was happening in the reduction case
* Change the catalystConverter to be a `val`.

Signed-off-by: Firestarman <[email protected]>
* Enable some approx percentile tests

Signed-off-by: Andy Grove <[email protected]>

* Enable some approx percentile tests

Signed-off-by: Andy Grove <[email protected]>

* enable tests and add more tests

* improve null test

* add tests for byte input

* remove temp debug print

* Remove comment

Signed-off-by: Andy Grove <[email protected]>

* update documentation

* run approx percentile tests with and without AQE

Signed-off-by: Andy Grove <[email protected]>

* Add test for split CPU/GPU approx_percentile and implement fix

* scalastyle

* Revert fix for issue 3770

* address PR feedback
@revans2 revans2 added this to the Nov 1 - Nov 12 milestone Nov 8, 2021
@revans2 revans2 self-assigned this Nov 8, 2021
@revans2 revans2 merged commit f5e6288 into NVIDIA:decimal128 Nov 8, 2021
@revans2 revans2 deleted the decimal128_upmerge branch November 8, 2021 22:49
@sameerz sameerz added the task Work required that improves the product but is not user facing label Nov 9, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
task Work required that improves the product but is not user facing
Projects
None yet
Development

Successfully merging this pull request may close these issues.