TECHNOLOGY PROFILE · TECH-0264
TensorRT
NVIDIA optimizer and inference runtime for neural networks.
Reference link only — not reviewed this edition. This profile preserves the prepared catalog’s original descriptions. Advantages depend on the workload; comparisons are editorial, not benchmark results.
Where it has an edge
Builds optimized engines for supported NVIDIA workloads rather than executing an unoptimized training graph.
Best-fit problems
Latency-sensitive vision and other supported neural inference.
Tradeoffs & cautions
Engine portability and dynamic-shape support are constrained; calibration may be needed.
What to test before choosing
Replay representative concurrency and context lengths; measure p50/p95 latency, throughput, memory and error rate.
Runtime & hardware
Usually server CPU/GPU
Access & licensing
Open-source core; deployment costs separate
Execution
Local / self-hosted
Lifecycle
Not independently audited
Comparison basis
Editorial problem-fit judgement; no head-to-head benchmark performed
Connected concepts
Alternatives to investigate
ONNX Runtime · OpenVINO