Jetson AGX Orin 15TB SSD read/write speed issue (2 case)

[winterping]

Hello, this is Ethan.

I built Orin custom carrier board.

I have a few question.

1. The SSD installed is recognized, but the read/write speeds are slow.

The Orin has a read speed of 3.8GB/s and a write speed of 6GB/s. The Xavier also has the same SSD, and both are 6GB/s.

Why is the Orin slower than the Xavier?

-Specifications- (Please let me know if you need more information.)

SSD - PM1743 Samsung SSD 15TB.

lspci check - 16GT/s (Gen4), width x4

2. I’m getting an error.

AER: Corrected error received: 0004:01:00.0
[ 420.297719] nvme 0004:01:00.0: PCIe Bus Error: severity=Corrected, type=Data Link Layer, (Receiver ID)
[ 420.297719] nvme 0004:01:00.0: device [144d:a826] error status/mask=00000080/0000e000
[ 420.297721] nvme 0004:01:00.0: [ 7] BadDLLP
[ 420.314830] pcieport 0004:00:00.0: AER: Corrected error received: 0004:01:00.0
[ 420.314890] nvme 0004:01:00.0: PCIe Bus Error: severity=Corrected, type=Physical Layer, (Receiver ID)
[ 420.314891] nvme 0004:01:00.0: device [144d:a826] error status/mask=00000001/0000e000
[ 420.314892] nvme 0004:01:00.0: [ 0] RxErr (First)
[ 420.315502] pcieport 0004:00:00.0: AER: Corrected error received: 0004:00:00.0
[ 420.315509] pcieport 0004:00:00.0: PCIe Bus Error: severity=Corrected, type=Data Link Layer, (Transmitter ID)
[ 420.315510] pcieport 0004:00:00.0: device [10de:229c] error status/mask=00001000/0000e000
[ 420.315512] pcieport 0004:00:00.0: [12] Timeout

This error seems to be related to the SSD, but I don’t know why.

If this is too difficult to resolve, please tell me how to prevent the text from being printed.

Thank you!

[KevinFFF]

Hi Ethan,

What’s the Jetpack version in use?

@winterping

The Orin has a read speed of 3.8GB/s and a write speed of 6GB/s. The Xavier also has the same SSD, and both are 6GB/s.

Are you using the exact same SSD in both cases?
Have you also verified the read speed on AGX Orin devkit?

@winterping

I’m getting an error.

Are those errors happening when you perform SSD read speed test?

Please share the full dmesg for further check.

[winterping]

Hi kevin
thank you for reply.

Please understand if my answer is strange because I don’t know much about it.

jetpack version → nvidia-l4t-core 36.3.0-20240719161631

R36 (release), REVISION: 3.0, GCID: 36923193, BOARD: generic, EABI: aarch64, DATE: Fri Jul 19 23:24:25 UTC 2024

KERNEL_VARIANT: oot

TARGET_USERSPACE_LIB_DIR=nvidia
TARGET_USERSPACE_LIB_DIR_PATH=usr/lib/aarch64-linux-gnu/nvidia

1. Are you using the exact same SSD in both cases?
   yes, same connector, same SSD

2. Have you also verified the read speed on AGX Orin devkit?
   I, did. please check this capture image.

   
   

3. Are those errors happening when you perform SSD read speed test?
   Yes, it happened when benchmarking.

4. Please share the full dmesg for further check.
   sudo journalctl -k -b > loglog.log
   please check log file.

   loglog.log (1.7 MB)

If you need more specific information, please explain.

Thanks

[KevinFFF]

Jetpack 6.0(r36.3.0) is an initial release for Jetpack 6.
Could you also verify with the latest Jetpack 6.2.1(r36.4.4)?

Please try adding pcie_aspm=off in kernel command line to check if it could help for your case.
and run sudo jetson_clocks before you start the test.

We would suggest also verifying it on the devkit to clarify if the issue is specific to the custom carrier board.