Implement Remainder for Decimal128 that handles when operands overflow #1175
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
For n % d remainder = n - (n div d) * d Signed-off-by: Navin Kumar <[email protected]>
Signed-off-by: Navin Kumar <[email protected]>
|
build |
NVnavkumar
requested review from
razajafri and
revans2
and removed request for
razajafri
revans2
previously approved these changes
May 26, 2023
abellina
reviewed
May 26, 2023
| @@ -58,6 +58,22 @@ JNIEXPORT jlongArray JNICALL Java_com_nvidia_spark_rapids_jni_DecimalUtils_divid | |||
| CATCH_STD(env, 0); | |||
| } | |||
|
|
|||
| JNIEXPORT jlongArray JNICALL Java_com_nvidia_spark_rapids_jni_DecimalUtils_remainder128(JNIEnv *env, jclass, | |||
There was a problem hiding this comment.
file needs copyright updated.
src/main/cpp/src/decimal_utils.cu
Outdated
| @@ -25,6 +25,9 @@ | |||
| #include <cmath> | |||
| #include <cstddef> | |||
|
|
|||
| #define D19_ UINT64_C(10000000000000000000) | |||
There was a problem hiding this comment.
I don't see this used in this PR.
Signed-off-by: Navin Kumar <[email protected]>
abellina
approved these changes
May 26, 2023
|
build |
revans2
approved these changes
May 26, 2023
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Outdated suggestions cannot be applied.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.
Part of a fix for NVIDIA/spark-rapids#8330.
This implements a native GPU implementation of the Remainder computation for 2 DECIMAL128 columns. This patterns the same algorithm that the JDK uses when computing the remainder for 2 BigDecimal values (which is the algorithm that Apache Spark relies on). This is used when trying the compute the remainder for 2 columns that would otherwise need to be upconverted to a higher precision than is supported by the Decimal128 format (higher than 38).
The algorithm is quite simple for
a % b(where a and b are 2 DECIMAL128 values/columns):r_i = a // b.b:a_int = r_i * b.a:r = a - a_int.This PR does this in a single custom kernel for optimization, and returns a result in the requested scale.
A corresponding PR in https://github.com/NVIDIA/spark-rapids/ will be created to implement the Spark plugin support for this.