User Guide

Jetson AGX™ Orin Developer Kit Reviewer's Guide
6
Best in Class Performance
Up to 8X Higher AI Performance
The power-efficient Jetson AGX Orin System-on-Module (SoM) delivers up to 275 TOPS
1
of AI
performance within a 60-Watt power budget, an 8X improvement over the 32 TOPS delivered by Jetson
AGX Xavier. For designs requiring lower power profiles, customers can tune their designs for power
profiles ranging from 15W to 60W. Jetson AGX Orin also delivers up to 9X the DLA performance of
Xavier providing even higher energy efficiency for inferencing applications that purely run on the DLA
cores. Jetson AGX Orin also delivers up to 1.5X higher CPU performance and up to 1.5X higher DRAM
bandwidth that helps to reduce bottlenecks and latencies when running multiple concurrent inferencing
applications.
Figure 2 Jetson AGX Orin delivers 8x the AI performance of Jetson AGX Xavier
Significant Performance Speedups for Real-world AI Workloads
NVIDIA Jetson AGX Orin delivers significant performance speedups for various real-word AI workloads
as measured by industry accepted benchmarks and performance on widely used neural networks.
The industry standard MLPerf benchmark defined by a consortium of industry leaders including
NVIDIA, Google, Meta and others measures performance on common AI tasks for image classification,
object detection and natural language processing using widely used neural networks such as ResNet-
50, Mobilenet-v2, and GNMT. NVIDIA Jetson AGX Xavier is the currently at the top of the leaderboard
for this benchmark, beating the competition by significant margins. Read the blog
here
NVIDIA Jetson AGX Orin will further increase this lead and even beat the current leader, Jetson AGX
Xavier by a significant margin. The results of this benchmark will be revealed on April 6
th
after it is
reviewed and confirmed by the MLPerf benchmark organization.
1
INT8 TOPS