Poor Performance - P41 Plus

[Papaya0]

Hello,

I am experiencing poor transfer performance between two P41 Plus drives in Ubuntu. Have tested moderate sized files (7GB) and larger files (50GB); results are the same. After a quick burst, R/W significantly declines.

Please see the below iotop results as a proxy during a transfer of a folder of 7GB files. The burst lasts for about 5 seconds and then significantly declines.

Here is the drive information for the two:

Capacity : 2.05 TB (2,048,408,248,320 bytes)
DevicePath : /dev/nvme0n1
DeviceStatus : Healthy
Firmware : 004C
FirmwareUpdateAvailable : The selected drive contains current firmware as of this tool release.
Index : 0
MaximumLBA : 4000797359
ModelNumber : SOLIDIGM SSDPFKNU020TZ
ProductFamily : P41 Plus
SMARTEnabled : True
SectorDataSize : 512

Capacity : 2.05 TB (2,048,408,248,320 bytes)
DevicePath : /dev/nvme1n1
DeviceStatus : Healthy
Firmware : 004C
FirmwareUpdateAvailable : The selected drive contains current firmware as of this tool release.
Index : 1
MaximumLBA : 4000797359
ModelNumber : SOLIDIGM SSDPFKNU020TZ
ProductFamily : P41 Plus
SMARTEnabled : True
SectorDataSize : 512

 Here is my current computer information

OS: Ubuntu MATE 24.04.1 LTS x86_64  
Kernel: 6.8.0-51-generic 
Shell: bash 5.2.21 
WM: Metacity (Marco) 
Terminal: mate-terminal 
CPU: 11th Gen Intel i5-11600K (12) @ 
GPU: Intel RocketLake-S GT1 [UHD Gra 
GPU: NVIDIA GeForce RTX 3080 Ti 
Memory: 15490MiB / 63748MiB
Motherboard: Z590-AORUS-MASTER

Thank you for your time an I hope there is something I can do to remediate this issue. 

[SolidigmGleb]

Hello Papaya0,

Thank you for providing detailed information about the performance issues with your P41 Plus drives. Please, explore some other factors that might be influencing the performance:

1. Adjust Power Management Settings: Ubuntu may apply aggressive power-saving settings that can throttle SSD performance. Disable power management for NVMe devices with the following command:

   bash

   echo 'autonomous_power_state_transition=0' | sudo tee -a /etc/nvme/hostnqn

2. Monitor System Resources: During file transfers, check for any potential bottlenecks such as high CPU or RAM usage which could impact performance. Use the htop command to monitor these resources in real-time.

3. Check NVMe Driver Compatibility: Ensure that the NVMe driver is functioning optimally. You can check this by running lsmod | grep nvme in the terminal.

4. Test Disk Performance: Utilize dd to test raw disk performance and identify any abnormalities:

   bash

   sudo dd if=/dev/zero of=/tmp/output bs=8k count=10k; sync; rm /tmp/output

If none of the above steps work, there might be a more complex issue at play. In that case, you can create a ticket on the Solidigm Support hub and we will be happy to assist you.

Kind regards,
Gleb
Solidigm Customer Support

[Papaya0]

Thanks you the help.

1. I have performed this task

2. No bottlenecks, I routinely monitor this.

3. I have performed this command as written and have received the below, I cannot ascertain the results:

nvme                   61440  4
nvme_core             212992  5 nvme
nvme_auth              28672  1 nvme_core

 4. Should I be changing this command depending on where my drive is mounted? Mine is /dev/nvme0n1

[Papaya0]

To note, task 1 had not remediated the issue