Posts by wojciszpl

    Unfortunately, the Intel N100 doesn’t have the best idle power consumption.

    N100 itself think is pretty good for that purpose, minipc from Aliexpress with this processor are using often about 5W only. The problem is rather with low quality of the available on market mobo's with this processor


    Quote

    I had the same board from ASUS, and I can give you a few tips.

    Unfortunately, the Realtek LAN controller performs poorly in terms of power consumption. For some reason, it doesn’t want to lower its power consumption — it needs to be properly configured or have the correct drivers installed. There’s more information about this online.

    Or just replace it with Intel adapter :) I226 for example


    I've tried using NIC connected via USB, because of other reasons - my mobo is unstable with internal NIC plugged off (without cable) or when it's switched off in BIOS - then my Pcie x1 lane is not working and all disks disapper. I spent a lot of time on this issue and now I'd for fure disrecommend this mobo to everyone. So probably I'd be not able to switch off the NIC in BIOS so we will no get on using another NIC.


    But thought about 2.5G just for the performance. Which intel NIC You have and where You put it in this mobo? M.2 wifi socket?

    Can You share a link to it?


    Quote

    Also you need to enable ASPM for PCIe in the BIOS.

    Sure, set everywhere what was possible (the highest possible states) L0s/L1 + substates allowed


    Quote

    Check power mode for SATA devices, run cat /sys/class/scsi_host/host*/link_power_management_policy

    If you see 'max performace', you can try reduce power consumption using echo med_power_with_dipm | tee /sys/class/scsi_host/host*/link_power_management_policy powertop do the same when you run powertop --auto-tune


    Have med_power_with_dipm everywhere - I'm using powertop


    Quote

    In my tests, I was able to reduce the board’s power consumption to ~8 watts by installing an Intel I226 and an ASM1166 SATA adapter. However, this was with only one SATA SSD connected.


    I'm twice higher and even more, around 18-20W


    Quote

    However, I didn’t keep that board because I found a more energy-efficient setup for myself:

    i3-8100T + B360 motherboard + Intel I226.

    I also have one 16 GB NVMe Optane drive for the system, three SATA SSDs, and two SATA HDDs. When the HDDs are spun down, the total power consumption is around 8 watts.


    I thought about mobo for i3 12100 and now I regret my decission. I bought this mobo (asus) after recommendation found on this forum.

    With the 8th gen proc You're probably not able to use QSV properly for new video codecs?

    Do you also use a PicoPSU? Conventional power supplies are usually very inefficient at low loads.

    Did you connect the system hard drive directly to the motherboard?

    Yes, sure. I'm uring original PicoPSU. Tried even with another one, from Aliexpress but the results was worse.


    Quote

    I have the system hard drive connected to the motherboard and the data hard drives connected to the expansion card.

    I have exactly the same. SSD connected to the native SATA port, rest of disk connected to ASM1064 (plugged in into the Pcie x1 slot) - tried also ASPM1166 connected into M.2 socket - results was same or very similar

    OMV is installed on USB thumb drive


    Quote

    In the BIOS, I've actually only disabled the parallel port, serial port, and sound.

    Me to. Plus set all ASPMs to the maximum was available, switchthed off unused USB ports, etc


    Quote

    Have you ever measured the power consumption without the expansion card? So, just with the system hard drive?

    I'm not sure, but think I've tried this.


    Quote

    I also read somewhere that adapters with an ASM chip consume 4-5 W more... but that could just be a coincidence.

    Maybe, I'm not an expert about this kind of stuff. I just read topic here and on unraid forum with recommended sata cards and those was recommeded, because doesn't have port multiplier ( I don't know what is the implication of having it)

    I'm using a 2-port JMB58x card in the E key (Wifi) slot. I have a 4-port ASMedia card to go in the x1 slot but haven't installed it yet as I don't need the extra ports right now. So maybe my test wasn't helpful for you anyway.



    I bought it from Amazon.de last December. For the BIOS version `dmidecode` says

    Code
    Handle 0x0000, DMI type 0, 26 bytes
    BIOS Information
            Vendor: American Megatrends Inc.
            Version: 0405
            Release Date: 12/13/2023

    The other decode says

    I think I found a 'workaround' solution.

    Just in the grub options I put pcie_port_pm=off" to tell the system do not switch off the clock.

    It looks that pcie x1 on this mobo and NIC are sharing not only the bus, but also CLK is the same - I claimed it to ASUS, we will see what they answer (if anything)

    After that, problem doesn't appered so far.


    Maybe hmoff You can just Yours cat /etc/default/grub if You have that option also in Your grub setting - this would explain, why You didn't have such problem

    Yes, I have portainer running, but tried even stop everything (all docker containers) to check power usage, but the difference was really small (max 1-2W in peaks)

    You have the same BIOS version and same revision of mobo, so theoretically there is no reason for me to claiming a complaint for this mobo.

    Maybe You're just lucky and have a better made mobo ;)

    Depends. Depends on what kind of system. Does your system have enough ram for what it is doing? Most people don't have a system swapping a lot with swap being a significant contributor to flash wear. Would you rather have the system crash/OOM kill or swap? Just running swapoff -a is most likely going to be only part of the solution if you do want to disable swap.

    I have 16GB od ram, plenty of dockers, usually using only 50% of RAM, never saw more than 60-65%, so it seems to bo safe.

    on the other hand, I am using omv for years, from one year it's stored on the usb stick, with installed flash memery plugin, but I didn't know that despite of this plugin, some other things should be done.


    Today I just had problem with transferring big file via SMB, saw huge load average but small cpu usage and tried to found a issue of the problem. Problem was caused by intensive use of SWAP

    I had also problem with default hdparm used in OMV

    I used hd-idle instead and it helped!

    […]

    It's really interesting! I envy you such a result :)


    I can hardly go below 20W, after lots of optimisations in BIOS and Linux

    Below my power usage:

    L80EimE.png

    Do You know if Your SATA card has a port multiplier or not? I bought 2 different , recommended cards (ASM1166 and ASM1064, both are supporting ASPM but my power consumption is 2x higher than Yours)

    On the other hand, I'm not sure if this single part could do that difference.


    Can You please also share Your output of:

    dmidecode -t baseboard

    dmidecode -t bios


    and if it's not a big issue for You, photos of Your BIOS settings?


    My powertop is showing that my CPU has slightly more thins to do, but it's still mostly in the C10 state, so the power usage should be similar to Yours

    nvL3Is3.png

    This is also the case with my Asus. I had a 2.5 GBit Lan card installed in the PCIex slot. Just as I had deactivated the internal LAN, the 2.5 GBit card no longer worked. With a 2.5 GBit Lan card in the WLAN slot, there were no problems with the inner LAN port deactivated. I haven't had the Asus in operation for a long time.

    For the A+E Key (Wlan + BT) m.2 slot, there are also SATA cards with two SATA ports. Maybe it's better to use one in your case.

    it's exactly the issue I have.

    I had also sata card (jmb582) for the M.2 Wifi slot, but it's not enought SATA ports for me, that's why I need Pcie x1 -> 4SATA or M.2 -> 6 SATA


    After my investigation, it looks that Pcie x1 slot and internal NIC are sharing the same CLK. I heared about some beta BIOS which is splitting the clk for those ports, but not sure if it's true, until will get info from asus.

    I made a ticket to ASUS support, we will see what they answer.


    I lost a lot of my time for solving this problem, so I think no more asus in future :)


    Which mobo are You using now?

    Just tried left it for 90 minutes and it worked perfectly.

    thanks for cheking! looks then that I have probably some damaged board.

    I even tried today with external NIC via USB, and when nothing was plugged in into internal NIC or when NIC was disabled via BIOS, then Pcie x1 slot didn't work


    Can You please tell me what version of BIOS do You have? And when You bought the mobo, where and model of Your expansion pcie card?


    thank You!


    edit:// can You please share output of:
    dmidecode -t baseboard ?

    I am using kernel 6.8pve (but tested also 6.5pve - result was the same), because with standard kernel there was no support for the iGPU hardware acceleration.

    which one do You use? Do You use hardware accel (Plex, Jellyfin)?


    Would be greteful for You check and information, this will answer if this is global problem with that board or it's just mine is broken

    How did You achieve this?

    What kind of optimisation in BIOS?

    Which kernel are You using?

    I have exactly the same MOBO and my power consumption in idle (all disk in stanbdy) is ~20W

    I've tried everything, but nothing helped


    Can You put Your output from:

    lspci -vvv | awk 'BEGIN{RS=""} /ASPM/' | grep --color -P '(^[a-z0-9:.]+|ASPM )'

    and from powertop - Yout C-states achieved by CPU and "Device stats"

    ?

    In my case it looks like the Nvme port stay working fine, the same for Wifi card, but the Pcie x1 slot is causing issues.


    Below output of my lspci, for me looks like NIC is sharing bus with the pcie x1 slot in which I have ASM1064 controller:

    And exactly this controller is stopping working after link is down:

    After I bought this mobo I launched memtest for 24h (as always do for new hardware) to make sure that everything works smooth.

    And after a long period of time I see, that the problems occurs only in specific situation:

    Link down or unplugged cable RJ45 -> NIC is propagating this on the PCie bus and (I don't understand why) stuff connected to the Pcie x1 bus is going to D3Cold state from which it can't come back to live.

    Hi Colleagues, I am using either Asus N100I-D4 for some time, bought it after good opinion here in the omv forum.

    But I have big troubles with that motherboard regarding stability - I had my disks connected via ASM1064 into the Pcie x1

    But I found the issue that often after reboot the disk were not visibile (cold restart was needed - full power cycle)

    Also I found that after link down for NIC (for instance when router switched off due to missing power or after pluging off the cable (for test reasons)) controler also disappear from the system and also full power cycle was needed.

    Example logs below:

    I thought that the problem is the the controller, so I bought another, this time ASM1166 connected to M.2 slot and thus needed to connect my M.2 disk via adapter to Pcie x1 port.

    Result? This time disks from M2 port was fine, but instead, the Nvme disk was disconnected (which gave me 99% sure that the problem is not with the controler but with the MoBo behaviour and issue with PCIe x1 slot)


    I spent a lot of time searching for information and solutions, including chatgpt, etc but without succes


    Does anybody has idea what's wrong and how to fix it?

    Thank You for Your answer.

    Personally I checked comparision on the other bechmark, and the difference was quiet big:


    The question is - which bechmark is more valuable in reality.


    The reason why I changed it was:

    - Want to increase performance - I saw that some apps like Nextcloud was too heavy for J5040 (it worked but no quick as expected)

    - Having better iGPU with QVS supporting more video codecs (and this is fine - transcoding in jellyfin is much better)

    - Expected the power consumption will drop down


    If I would not read so many positive info about this board, I would not buy it - J5040 met 98% of my needs, had 4SATA by default + 2 thanks to M2 Wifi slot, was rock stable and as I see now - really good optimised in power consumption.


    That's why I;m so dissapointed and now I also thought the replace was not good idea.


    I tried switch off everything in bios but consumption is on the same level (17-18W in idle) - powertop -autotune helped a bit, because before it was a 21W


    Have no idea what to do more...

    Did You make any special settings in BIOS regarding power consumption?

    I ordered this mobo to replace my AsRock J5040, after reading this topic I expected to have similar or lower power consumption and much powerful CPU and iGPU

    But after setting everything and launch, I'm really disappointed - power consumption is ~19-20W in idle while it was 12-13W on J5040 (all disk in standby).

    I'm using PicoPSU 160W