Official Technical Resources: NVIDIA (page 1)
Official Technical Resources · NVIDIA. Developer documentation, models and hardware technical resources. This source has 264 entries admitted under this category's review policy, listed across 18 pages; each links directly to the original.
Authority tiers in this category: A1–A2.
Memory Management
NVIDIA cuVS uses RAPIDS Memory Manager (RMM) through RAFT so GPU algorithms can allocate temporary buffers, output arrays, and staging memory through one…
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: Retrieval-Augmented Generation (RAG) · Embedding · LLM Inference Optimization
Link last checked: 2026-09-25
Dynamic Batching
Dynamic batching wraps an existing ANN index and combines many concurrent small search requests into larger GPU batches.
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: Retrieval-Augmented Generation (RAG) · Embedding · LLM Inference Optimization
Link last checked: 2026-09-25
Product Quantization
Product Quantization, or PQ, is a GPU-accelerated preprocessing algorithm for compressing vectors.
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: Retrieval-Augmented Generation (RAG) · Embedding · LLM Inference Optimization
Link last checked: 2026-09-25
Spectral Embedding
Spectral Embedding is a GPU-accelerated preprocessing algorithm for graph-based dimensionality reduction.
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: Retrieval-Augmented Generation (RAG) · Embedding · LLM Inference Optimization
Link last checked: 2026-09-25
cuVS Bench Backends
This page explains how cuVS Bench separates benchmark orchestration from the system that actually builds and searches an index.
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: Retrieval-Augmented Generation (RAG) · Embedding · LLM Inference Optimization
Link last checked: 2026-09-25
cuVS Bench
cuVS Bench is a reproducible benchmarking tool for ANN search implementations.
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: Retrieval-Augmented Generation (RAG) · Embedding · LLM Inference Optimization
Link last checked: 2026-09-25
cuvs_bench.synthesize_dataset
Generates a synthetic dataset of arbitrary size from a small sample of a real dataset, for benchmarking vector indexes at scales (100M, 1B, 10B+) where the…
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: Retrieval-Augmented Generation (RAG) · Embedding · LLM Inference Optimization
Link last checked: 2026-09-25
Methodologies
Vector search indexes should be compared by both search quality and performance.
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: Retrieval-Augmented Generation (RAG) · Embedding · LLM Inference Optimization
Link last checked: 2026-09-25
FastTokens Tokenizer
Reduce frontend tokenization latency for long-context BPE models
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: Model Deployment Options · LLM Inference Optimization · LLM Observability and Tracing
Link last checked: 2026-09-25
Metrics Developer Guide
Reference for creating counters, gauges, and histograms with Dynamo's metrics API, surfaced on the local /metrics endpoint.
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: LLM Observability and Tracing · Model Deployment Options · Model Evaluation and Benchmarks
Link last checked: 2026-09-25
Agent Harnesses
Point coding-agent CLIs at a Dynamo deployment
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: AI Agent · Tool Calling · LLM Inference Optimization
Link last checked: 2026-09-25
Agent Hints
Per-request serving hints for agentic workloads
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: AI Agent · Tool Calling · LLM Inference Optimization
Link last checked: 2026-09-25
Agent Trace Replay
Capture an agent workload once, then replay it on real or simulated infrastructure
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: AI Agent · Tool Calling · LLM Inference Optimization
Link last checked: 2026-09-25
Agent Tracing
Export Dynamo request traces, tool-call metadata, and Perfetto timelines
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: AI Agent · Tool Calling · LLM Inference Optimization
Link last checked: 2026-09-25
Priority Scheduling
Request priority across the Dynamo router and backend engines
Evidence use: For understanding GPU computing, training, retrieval, inference serving, reliability or deployment decisions.
Limitations: Directly evidences only the implementation and claims of NVIDIA's technology stack. Recheck versioned interfaces and experimental capabilities before use.
Related concepts: AI Agent · Tool Calling · LLM Inference Optimization
Link last checked: 2026-09-25