Convert Apache Spark UDFs into GPU-accelerated RapidsUDF implementations.
npx skills add https://github.com/NVIDIA/cudf-spark --skill udf-convert-to-cudf
Please help me install this Agent Skill. Skill: udf-convert-to-cudf Source: https://github.com/NVIDIA/cudf-spark/tree/main/skills/udf-convert-to-cudf Command: npx skills add https://github.com/NVIDIA/cudf-spark --skill udf-convert-to-cudf
This Skill helps you convert an existing Apache Spark UDF into a GPU-accelerated RapidsUDF so it can run efficiently on NVIDIA GPUs while preserving CPU compatibility.
Use this skill to convert my tested Spark UDF into a RapidsUDF and update the comparison test for GPU execution.
High-intent search queries and answers about installing and using this skill.
To convert an Apache Spark UDF to a GPU RapidsUDF, implement the RapidsUDF interface, rename the class, and translate row-based logic into columnar evaluateColumnar operations using cuDF ColumnVector APIs.
Implementing columnar execution in a Spark RapidsUDF requires using cuDF Java APIs and correct ColumnVector type mapping to translate row-based UDF behavior into evaluateColumnar operations.
Preventing GPU memory leaks in Spark RapidsUDFs requires explicit GPU resource management and memory-leak-safe evaluation during the columnar execution logic implementation.
You can maintain CPU compatibility during UDF migration by updating comparison tests to verify that CPU and GPU results match exactly within your stepwise migration workflow.
Updating Spark tests is required when migrating a UDF to GPU to validate and harden the implementation, ensuring comparison tests verify CPU and GPU results match and check for memory leaks.
Use a GPU RapidsUDF instead of a standard Spark UDF when you need production-ready GPU acceleration for Spark jobs and have a tested CPU UDF ready for stepwise columnar migration.
NVIDIAConvert Apache Spark UDFs into equivalent Spark SQL expressions.
NVIDIAOptimize cuDF RapidsUDF implementations for GPU performance with profiling and benchmarking.
nop2099Check disk space, RAM, and GPU status against predefined thresholds.
NVIDIAValidate and compare Spark UDF implementations across CPU and GPU paths.
