Latency When Using LSTM Model with Coral USB Accelerator and Raspberry Pi Zero 2 W

[jin-park]

Description

Hello, I saw this issue on slow LSTM inference speeds due to some problems with the USB drivers. I am considering using the USB Accelerator with my Raspberry Pi Zero 2 W because I believe it may be faster than the CPU even with the large latency. Are there any plans to remove the need for an additional command (setting USB_ENABLE_BULK_DESCRIPTORS_FROM_DEVICE to 1), and if not, do you think it is possible to run inference on a single layer LSTM in maybe ~20ms or so with the USB Accelerator? Thanks.

Click to expand!

Issue Type

Support

Operating System

No response

Coral Device

USB Accelerator

Other Devices

No response

Programming Language

Python 3.9

Relevant Log Output

No response