CUDA Kernels API Reference

This module provides CUDA-accelerated functions for trajectory planning and dynamics computation with automatic CPU fallback.

Availability Check

Core CUDA Kernels

Trajectory Kernels

Dynamics Kernels

Potential Field Kernels

High-Level Wrappers

CPU Fallback Functions

Memory Management

Memory Pool Class

Grid Configuration

Performance Tools

Constant Memory

Module Constants

Environment Variables

MANIPULAPY_FASTMATH: Set to “1” to enable fast math optimizations (~2x speedup with relaxed IEEE 754 compliance)
MANIPULAPY_USE_FP16: Set to “1” to use 16-bit floating point precision for memory-bound kernels