Blogs›NVIDIA›BEV Pooling Optimization for Physical AI

BEV Pooling Optimization for Physical AI

posts

2026

This feature thread tracks the evolution of optimizing Bird's-Eye-View (BEV) pooling for NVIDIA GPUs, crucial for physical AI applications like autonomous vehicles and robotics. Initial efforts focused on the fundamental BEV pooling operation, which gathers depth-weighted image features and scatters them into BEV grid cells. Subsequent developments, exemplified by BEVPoolV3, introduce algorithmic and implementation optimizations to significantly reduce latency. These include reducing duplicate depth loads, utilizing efficient scatter maps, precomputing indices to eliminate runtime integer division, and optimizing output writes. The optimization workflow involves classifying the memory regime (DRAM-bound vs. L2-resident), eliminating redundant scatter traffic, mapping kernel implementations to target GPU architectures, and validating bottlenecks with profiling tools like NVIDIA Nsight Compute. Performance benchmarks demonstrate substantial speedups across different GPU architectures and precision formats (FP16, FP8), highlighting the importance of architecture-specific optimization strategies.

2026

Accelerating BEV Pooling on NVIDIA GPUs for Physical AI Applications | NVIDIA Technical Blog

6/24/2026

Introduces BEVPoolV3 with four key algorithmic changes: reduced duplicate depth loads, a five-array INT32 scatter map, precomputed indices eliminating runtime integer division, and interval-owned output writes. Details a practical workflow for optimizing BEV pooling operators: classifying the working set by L2 fit, eliminating redundant scatter traffic, mapping kernel implementation to the target NVIDIA GPU architecture, and validating bottlenecks with NVIDIA Nsight Compute. Demonstrates up to 22x speedup (FP16) on DRAM-bound paths and up to 42x speedup (FP8) on L2-resident paths over BEVPoolV2 on NVIDIA RTX GPUs.