TechPrufer
Get started
Qdrant
Qdrant logoQdrant · Feature trail

GPU Accelerated Indexing

4 posts · Jan 2025 – May 2026

What this trail covers

Qdrant continues to optimize performance for vector search, demonstrating significant speed improvements for disk-based retrieval. This post details benchmarks showing Qdrant achieving 2x higher throughput and 50% lower latency than Elastic's DiskBBQ at the same recall target, using significantly less CPU and RAM per node. This is achieved through advanced quantization (TurboQuant 4-bit), async disk scoring, and a two-stage retrieval process. The results highlight Qdrant's efficiency and perform. GPU indexing is now available, delivering up to 4x faster HNSW index builds on dedicated GPUs, beneficial for high-write workloads. Multi-AZ replication is introduced for enhanced availability with a 99.95% SLA. Audit logging is now available, capturing API operations for compliance and security.

All Qdrant trails
  1. Qdrant 1.18 - TurboQuant - QdrantIntroduces TurboQuant, a new quantization method developed by Google Research, which offers twice the compression ratio of scalar quantization with similar recall and speed, and improved recall compared to binary quantization at equivalent storage budgets. Also adds memory monitoring per collection, the ability to add/remove named vectors without collection recreation, audit logging improvements with a query API and tracing IDs, per-collection API metrics, and new strict mode guardrails for memory and search batch size.
  2. Now Available on Qdrant Cloud: GPU Indexing, Multi-AZ, and Audit Logging - QdrantIntroduces GPU-accelerated indexing for up to 4x faster HNSW index builds, available on AWS clusters. Adds Multi-Availability-Zone (Multi-AZ) replication for enhanced availability (99.95% SLA) on Premium tier clusters. Implements audit logging for API operations (queries, upserts, deletes, collection management, snapshots) on all paid clusters, with logs stored persistently and retrievable via API for SIEM integration.
  3. Qdrant 1.13 - GPU Indexing, Strict Mode & New Storage Engine - QdrantIntroduces GPU-accelerated HNSW indexing on NVIDIA, AMD, and Intel GPUs, offering up to 10x faster indexing times compared to CPU-based methods for equivalent hardware cost. The in-house, Vulkan API-based implementation is vendor-agnostic and supports multi-GPU configurations. HNSW graph compression via Delta Encoding reduces memory footprint by up to 30% with no measurable performance degradation. Strict Mode is introduced to enforce operational controls and prevent inefficient usage in distributed deployments, with new collections defaulting to this mode. Filtering by named vectors is now supported with a `has_vector` condition. A new custom storage engine is also introduced for constant-time reads/writes of payloads and sparse vectors.