Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
* Back port spark-specific murmur32 hash code from cudf. * Run pre-commit to format files. We were behind a bit. * Update pre-commit config to 16.0.1 to match cudf. Re-ran formatting. * Change jni bindings to use the spark-rapids-jni implementation of murmur hash instead of the cudf version. Brought over cpp and java tests. * Documentation fix. * Fix cpp tests to actually call the spark_rapids_jni murmur hash. * First pass at xxhash64. cpp tests passing. * Improve cpp tests - null cases and more floating point edge cases. * Add Java tests. * Moved murmur32 hash implementaion from cudf to spark-rapids-jni Signed-off-by: db <[email protected]> * PR review changes. * Fix copyright data in Hash.java * Enable 32 bit decimal hash test. * Implement xxhash64 on the gpu Signed-off-by: db <[email protected]> * Add missing newlines. * PR review changes. * Remove default xxhash64 class constructor. Remove unused parameter (row index) from remaining constructor. * Fix issues with merge. * Rectify thirdparty/cudf issues. * Revert inadvertent change to pom.xml * PR review feedback changes. --------- Signed-off-by: db <[email protected]>
- Loading branch information