Benchmarker: Scope of support

Compatible model

Only single-input models are supported.

The four-dimensional array structure of images should be organized Batch, Number of Channels, Height, and Width.

Batch size: The number of combined input datasets that the model processes simultaneously.

Channel: 3 for RGB or BGR and 1 for Grayscale.

Input size: In computer vision tasks, input size refers to the size of the input images.

Supported Hardware

Arm MCU/NPU

Target Device	.tflite
Renesas RA8D1 (Arm Cortex-M85)	O (only INT8)
Renesas RA8D1 (Arm Cortex-M85) with helium	O (only INT8)
Alif Ensemble DevKit-E7 Gen2 (Arm Cortex-M55 + Ethos-U55)	O (only INT8)
Alif Ensemble DevKit-E7 Gen2 (Arm Cortex-M55 + Ethos-U55) with helium	O (only INT8)
Arm Virtual Hardware Corstone-300(Ethos-U55/U65)	O (only INT8)
Arduino Nicla Vision(Arm Cortex-M7/M4)	O (only INT8)
NXP i.MX 93(Arm Cortex-A55/M33+Ethos-U65)	O (only INT8)

NVIDIA

When benchmarking on Jetson, it is essential for the model file and target device to match the Jetpack version.

Target Device	.trt	.engine	.tflite	.onnx
Jetson Nano JetPack 4.4.1	O	O	X	O
Jetson Nano JetPack 4.6	O	O	X	O
Jetson Xavier NX JetPack 4.6	O	O	X	O
Jetson Xavier NX JetPack 5.0.2	O	O	X	O
Jetson TX2 JetPack 4.6	O	O	X	O
Jetson AGX Xavier JetPack 4.6	O	O	X	O
Jetson AGX Orin JetPack 5.0.1	O	O	X	O
Jetson Orin Nano JetPack 6.0	O	O	X	O
AWS-T4	O	O	X	O

Raspberry Pi

Target Device	.tflite	.onnx
Raspberry Pi ZeroW	O	X
Raspberry Pi Zero2W	O	X
Raspberry Pi 2B	O	O
Raspberry Pi 3B	O	O
Raspberry Pi 3B+	O	O
Raspberry Pi 4B	O	O
Raspberry Pi 5	O	O

Intel

Target Device	.zip(bin+xml)
Xeon W-2223	O