Hi,
We are working on bringing up a design with the Hailo-8L connected via PCIe x1.
In our design, we are using the Hailo-8L in PCIe boot mode, with I2S_SDO being pulled up to 1V8.
We have verified
All the various voltage sources → as per specification
PCIe clock → 100 MHz, diff signal
CLKIN → 25 MHz
NRESET → 1V8
Unfortunately, we can’t enumerate the Hailo-8L device via PCIe.
We have further exposed GPIO_0 , GPIO_4 and GPIO_5 which in various reference designs is labelled as mandatory initial debug bus.
Unfortunately we can’t find any documentation on how the protocol can be used and/or how to interrogate the Hailo-8L over this debug bus.
Any reference would be greatly appreciated.
Thank you, Alex
omria
August 27, 2025, 11:00am
2
Hey @Alexander_Entinger ,
Welcome to the Hailo Community!
If your Hailo-8L isn’t showing up in lspci
, here’s what to check:
Power and Clock Setup
Release the reset signal (NRESET) only after power-up is complete
Ensure both clocks are stable first: 100 MHz PCIe clock (REFCLK) and 25 MHz clock (CLKIN)
Confirm I2S_SDO is pulled up to 1.8V for PCIe boot mode
Debug Pins The GPIO pins (GPIO_0, GPIO_4, GPIO_5) are for internal factory testing only - treat them as generic I/O.
If Problems Persist
Use SCU recovery mode to boot from alternate source and reload firmware
Verify reset and clock timing (PCIe devices are timing-sensitive)
Enable PCIe link training on your host
Bottom Line : This is typically a timing issue with power, clocks, or reset signals rather than hardware failure.
Hi @omria
We’ve verified power and clock setup, as well as power/reset/clock timing and could not determine any deviations. I2S_SDO is pulled up too.
Is there any other way to determine if the Hailo-8L is coming up and running, i.e. by interrogating its JTAG interface, even when running in PCIe boot mode?
Do you have additional information about the SCU recovery mode?