Task Wise GPU Integration for Machine Learning Algorithms.

Task Wise GPU Integration for Machine Learning Algorithms.

Overview

GPU execution support has been introduced for supported Machine Learning algorithms to improve processing performance, execution efficiency, and scalability for machine learning workloads. Users can now enable GPU execution at task level for supported Classification and Regression algorithms using the IS_GPU option.

Feature Enhancements

  • Added task-level GPU execution support for supported Machine Learning algorithms.

  • Introduced IS_GPU toggle for eligible Classification and Regression tasks.

  • GPU configuration persists after saving and reopening workflows or pipelines.

  • Supported algorithms can execute using GPU processing where available.

  • Mixed execution flow with both GPU and CPU tasks is supported within the same workflow or pipeline.

  • Existing pipelines and executions continue to run on CPU by default without impact.

  • Separate GPU execution environment and dependency management added for GPU-enabled processing.

  • GPU execution results remain aligned with CPU execution outputs within acceptable tolerance.

Supported Classification Algorithms

  • Logistic Regression

  • Random Forest Classification

  • XGBoost Classification

  • Support Vector Machine (SVM)

Supported Regression Algorithms

  • Linear Regression

  • Random Forest Regression

  • XGBoost Regression

  • Ridge Regression

  • Support Vector Regression (SVR)

  • Lasso Regression

GPU Configuration

Users can enable or disable GPU execution for supported Machine Learning tasks using the IS_GPU option available within the task configuration.

Execution Behavior

  • GPU execution is applied only for supported Classification and Regression algorithms.

  • Unsupported tasks continue processing on CPU.

  • Users can enable or disable GPU execution at task level using the IS_GPU option.

  • Pipelines and workflows containing both GPU and CPU tasks execute successfully without interruption.

  • GPU execution settings are retained across save and reopen operations.

  • GPU execution results remain aligned with CPU execution outputs within acceptable tolerance.

Benefits

  • Improved model training and execution performance.

  • Better utilization of GPU-enabled infrastructure.

  • Flexible execution management with CPU and GPU support.

  • Enhanced scalability for machine learning workloads.

  • Backward compatibility with existing workflows and pipelines.