Module load_balancer 

GPU load balancing for distributing index-building work across multiple devices.

This module provides:

• GpuLoadBalancer: runtime tracking of per-device workloads and selection of the least-loaded device for a new task.
• WorkloadDistributor: static splitting of a large index job into per-device contiguous chunks.

Pure Rust Policy

No CUDA runtime calls are made here. All load-balancing logic is Pure Rust and operates on abstract device descriptors (SimpleGpuDevice).

Structs

GpuLoadBalancer

Distributes GPU work across multiple devices using a least-loaded strategy.

SimpleGpuDevice

Lightweight descriptor of a GPU device used for load balancing decisions.

WorkloadChunk

A contiguous slice of a vector dataset assigned to a specific GPU.

WorkloadDistributor

Splits a large vector index job across multiple GPU devices proportionally to their memory capacity.