Jetson Orin AGX benchmarks for image and video processing

Updated: August 16, 2026

These benchmarks measure the Fastvideo SDK imaging pipeline on NVIDIA Jetson AGX Xavier — demosaicing, JPEG and JPEG2000, denoising, and resize — for 2K images. The page shows the real-time imaging throughput the Xavier module delivers for embedded camera applications.

How these numbers were measured. Fastvideo benchmarks are measured on NVIDIA GPUs, timing pure GPU processing at the resolution, bit depth, format and quality stated on this page. See the full benchmark methodology and test environment and the Fastvideo SDK benchmark report (PDF). Our tools are open source on GitHub and built on NVIDIA CUDA.

Check a single module. Every module in the Fastvideo SDK comes with a command-line sample. Run it on your own images on your own GPU and compare the result with the figures published here.

Check the whole pipeline. A module in the middle of a pipeline cannot be judged on its own: its input is hard to prepare, and its output is an intermediate product you cannot look at. What matters is how it changes the final image and the total time of the schema. Two of our applications run the complete pipeline and report the time of every stage — FastVCR (live cameras or RAW files from disk, so no camera is required) and Fast CinemaDNG (DNG, offline). The benchmark module is included in the free builds of both.

Detailed benchmark data. Measured results are published on our benchmark pages: all benchmarks.

Jetson Orin AGX Tech Specs

GPU 512-core Volta GPU with Tensor Cores
CPU 8-core ARM v8.2 64-bit CPU, 8MB L2 + 4MB L3
Memory 16 GB 256-Bit LPDDR4x | 137 GB/s
Storage 32 GB eMMC 5.1 DL
Accelerator (2x) NVDLA Engines
Vision Accelerator 7-way VLIW Vision Processor
Encoder/Decoder (2x) 4Kp60 | HEVC/(2x) 4Kp60
Size 105 mm x 105 mm
Deployment Module (Jetson AGX Xavier)

We have done performance benchmarks at Jetson AGX Xavier for the key components of Fastvideo SDK. We've tested images with 2K and 4K resolutions and got the following averaged benchmarks. For each particular algorithm we've measured an average kernel time that we need to process an image which has already been uploaded to GPU memory. This is not latency, but average kernel time on GPU.

Jetson Orin AGX performance benchmarks for 2K images (1920×1080)

HQLI Demosaic (8-bit, RGGB) – 0.30 ms
HQLI Demosaic (16-bit, RGGB) – 0.36 ms
DFPD Demosaic (8-bit, RGGB) – 0.45 ms
DFPD Demosaic (16-bit, RGGB) – 0.96 ms
MG Demosaic (16-bit, RGGB) – 2.23 ms
JPEG encoder (8-bit, quality 90%) – 0.42 ms
JPEG encoder (24-bit, quality 90%, 4:2:0) – 0.62 ms
JPEG encoder (24-bit, quality 90%, 4:4:4) – 0.75 ms
JPEG decoder (8-bit, quality 90%) – 0.86 ms
JPEG decoder (24-bit, quality 90%, 4:2:0) – 1.35 ms
JPEG decoder (24-bit, quality 90%, 4:4:4) – 1.37 ms
24-bit image resize (algorithm Lanczos3) from 1920×1080 to 960×540 – 1.48 ms
24-bit image resize (algorithm Lanczos3) from 1920×1080 to 1919×1079 – 2.34 ms
Denoise (8-bit, wavelet 9/7, 7 dwt levels) – 1.6 ms
Denoise (24-bit, wavelet 9/7, 7 dwt levels) – 4.4 ms
J2K Encoder (24-bit, wavelet 9/7, 7 dwt levels, cb 32, cr 12, lossless, single) – 23.3 ms
J2K Encoder (24-bit, wavelet 9/7, 7 dwt levels, cb 32, cr 12, lossy, single) – 11.1 ms
J2K Encoder (24-bit, wavelet 9/7, 7 dwt levels, cb 32, cr 12, lossy, batch) – 10.5 ms

Jetson Orin AGX performance benchmarks for 4K images (3840×2160)

HQLI Demosaic (8-bit, RGGB) – 0.53 ms
HQLI Demosaic (16-bit, RGGB) – 0.93 ms
DFPD Demosaic (8-bit, RGGB) – 1.43 ms
DFPD Demosaic (16-bit, RGGB) – 2.18 ms
MG Demosaic (16-bit, RGGB) – 5.8 ms
JPEG encoder (8-bit, quality 90%) – 1.23 ms
JPEG encoder (24-bit, quality 90%, 4:2:0) – 1.83 ms
JPEG encoder (24-bit, quality 90%, 4:4:4) – 2.67 ms
JPEG decoder (8-bit, quality 90%) – 2.13 ms
JPEG decoder (24-bit, quality 90%, 4:2:0) – 4.0 ms
JPEG decoder (24-bit, quality 90%, 4:4:4) – 4.15 ms
24-bit image resize (algorithm Lanczos3) from 3840×2160 to 1920×1080 – 5.45 ms
24-bit image resize (algorithm Lanczos3) from 3840×2160 to 3839×2159 – 8.6 ms
Denoise (8-bit, wavelet 9/7, 7 dwt levels) – 4.6 ms
Denoise (24-bit, wavelet 9/7, 7 dwt levels) – 13 ms
J2K Encoder (24-bit, wavelet 9/7, 7 dwt levels, cb 32, cr 12, lossless, single) – 92 ms
J2K Encoder (24-bit, wavelet 9/7, 7 dwt levels, cb 32, cr 12, lossy, single) – 42 ms
J2K Encoder (24-bit, wavelet 9/7, 7 dwt levels, cb 32, cr 12, lossy, batch) – 39 ms

Jetson Orin AGX performance benchmarks for 12-bit images 4032×2192

JPEG encoder (gray, 12-bit, quality 90%) – 2.07 ms
JPEG encoder (color, 12-bit, quality 90%, 4:2:0) – 3.1 ms
JPEG encoder (color, 12-bit, quality 90%, 4:4:4) – 5.0 ms

About the author

Fyodor Serzhenko, PhD, is the founder and CEO of Fastvideo. He earned his PhD at the Moscow Institute of Physics and Technology (MIPT) in 1993. Since 2009 he has led the development of Fastvideo’s GPU-accelerated image codecs and ISP modules, including the Fastvideo SDK. Connect on LinkedIn.

Why you can trust these results

Fastvideo has built GPU-accelerated image processing software since 2009. Our codecs comply with the standards they implement — JPEG (ITU-T T.81 / ISO IEC 10918) and JPEG2000 (ITU-T T.800 / ISO IEC 15444) — and our tools are open source on GitHub. Every performance figure we publish is reproducible: see our benchmark methodology, download the Fastvideo SDK benchmark report (PDF), and measure it on your own GPU.