[v17] MB-68591: GPU-Accelerated Vector Search

[CascadingRadium]

• IVF Indexes can now optionally use GPUs for training purpose using a best-attempt strategy.
• Multi-GPU support is added, using a softmax based load balancing algorithm to evenly distribute load across all the GPUs.
• Requires:
  • MB-67052: Add GPU bindings go-faiss#46
  • [v17] MB-68591: Add GPU Field Indexing Option bleve_index_api#77
  • [v17] Store field-level indexing options in zap segments #345

[Copilot]

Pull Request Overview

This PR adds GPU acceleration support for FAISS vector index training in the zap package. The implementation provides a best-effort approach to use GPUs when available and configured, with graceful fallback to CPU training when GPU resources are unavailable or insufficient.

Key Changes:

• Added GPU training capability with automatic fallback to CPU training
• Introduced vector index options to persist GPU configuration preferences
• Implemented intelligent GPU device selection based on available memory

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 10 comments.

File	Description
faiss_vector_gpu.go	New file implementing GPU training logic, device selection algorithm, and resource management for FAISS index training
faiss_vector_gpu_test.go	New test file with basic tests for GPU enumeration and device selection
section_faiss_vector_index.go	Updated to support vector index options, integrate GPU training into merge and write operations, and persist GPU configuration

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull Request Overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.