Reduce power consumption with powertop

[crowdx42]

26 minutes ago, Interstellar said:

 

No idea - completely headless. None of my dockers need it as far as I'm aware, I don't pass it through to Plex for example. Don't even have the drivers installed - is it a problem its being used briefly? Up to 1.5% if I look now. Edit: Back to 0.3%

 

Is there a way to find out what is using the GPU?

 

I have about 15 dockers running and a OPNSense VM running on cores 4 and 5 (isolated), stopping all the dockers/VMs drops the average usage to low single digit CPU usage and drops 1-2W off the power.

 

 

Are all your drives connected by the boards SATA ports? I have a similar setup but have two SAS cards and my system is idling at 80watts, intel i5 12600k and 16gig of memory on a Z690 motherboard with more drives BUT at idle, I don't believe that should have much of an impact.

 

[zff]

Excellent post, though the graph for the Seasonic should be included especially since the Corsair is included--I was confused why Seasonic Titanium PSUs are recommended--Titanium rating only guarantees 90% at 10% but the smallest Titanium-rated offering is 600W and 60W is not considered a low power build. Then I've skimmed through all the posts and see the Seasonic chart which is comparable to the Corsair (Corsair is actually a little more efficient at 20W). So because these two posts are mentioned, I can only assume other PSUs have worse curves and also the 80 Plus rating is not relevant at all (including Platinum) for low-powered builds since 10% of the lowest Titanium offering of 600W in the consumer market is still 60W.

 

If I were to build a NAS and home server today, what to consider for a CPU: Pentium Gold G7400, i3-12100, or i3-10100? All 3 have good enough transcoding capabilities via Quick Sync and I expect it to last for over a decade, which is not unreasonable since my desktop I'm typing on is an i5-2500k and I will finally upgrade at 14th gen when the DDR5 platform and pricing matures. It would do the usual NAS stuff but in the future might run Proxmox stuff and do some basic home surveillance. The most important thing is idle power consumption--is it true that "modern Intel processors from the past 10 years" all idle under 1W? If so, I guess it doesn't really matter what processor to choose at all and it only comes down to budget unless the server is working hard for the majority of the time? I assume accompany that with consumer-grade mini-ITX (4 builtin SATA ports is fine), I've settled with high capacity for power efficiency). And of course one of those Corsair or a used Seasonic Titanium (*all models* have similar efficiency curves? I only see this model (https://seasonic.com/prime-titanium-fanless) tested by that German site and if I go on Seasonics and search for their Titanium PSUs, I see it's neither of these two PSUs listed: https://seasonic.com/consumer/power-supplies?attr_80plus=15 which are presumably newer models. Given that even when used they command a premium, I want to make sure all models perform similarly well at the 3-8% utilization (assuming 650W PSU) that I expect my low power build to turn out, since I will likely have 0-2 HDDs running max (the remaining or all will be cold storage, I might just get a large SSD if I need online storage). If that's the case I will just get the cheapest used 600-650W Seasonic Titanium PSU regardless of a specific model, though I think I prefer fanless as it's one less point of failure (I see people replacing the stock fan with Noctuas for whatever reason).

 

Actually, I don't need a NAS until end of next year, but I doubt even then ATX12VO will be available for a mini ITX build at reasonable prices so it's probably not worth waiting.

 

One more question--how frequent should disks be spun down before it's not worth the wear and tear anymore? AFAIK the concern is largely outdated for modern drives, yet people continue to insist on leaving the NAS 24/7 for no reason other than it's so terrible to turn off drives, as if most desktop PCs don't have have at least 1 HDD and aren't turned off daily for decades and that was never a concern in that regard. NAS-rated drives are supposed to be "better built" so even if they are intended for servers that run 24/7, surely they can handle being turned off reasonably frequently.

 

Much appreciated and keep up the good work.

 

[Interstellar]

10 hours ago, crowdx42 said:

Are all your drives connected by the boards SATA ports? I have a similar setup but have two SAS cards and my system is idling at 80watts, intel i5 12600k and 16gig of memory on a Z690 motherboard with more drives BUT at idle, I don't believe that should have much of an impact.

 

 

Yes only SATA. No SAS etc cards.

 

I used to have a SAS card, can't remember which Highpoint card it was. A few years ago I used to have 8 odd 4TB drives hanging off it. Since upgrading to four 14TBs thus being able to ditch the card to only use internal SATA ports, I remember dropping a good 5W if not more when I removed it.

 

 

10 hours ago, mgutt said:

Depends on the container and how much load it produces. As for example a minecraft container produces a "huge" load through its java processes. But usual containers like plex, nextcloud, etc do not. A VM is different from that. It comes with a huge overhead because of its emulation part. So using VMs usually prevents reaching the deepest C-States.

 

Only 1W is strange. If I start minecraft or a Windows VM, the power consumption raises by around 10W.

 

For the iGPU? You should, but I think they are, because as far as I know powertop does not show the iGPU as long no drivers are installed.

 

 

Minecraft/Windows VMs are probably way more intensive. As powertop shows for cores 4/5, they are isolated and only used by my OpnSENSE VM and they are spending 60-70% of their time in C7, thus obviously not that intensive.

 

corefreq-cli shows CPU Package power at about 4-5W with the VMs/Dockers running and 2-3W without them.

 

As for iGPU, turns out something is installed. I think because maybe UnRAID includes them by default now. I don't recall installing them!

 

Kernel driver in use: i915

Kernel modules: i915

 

 

Edited December 5, 2022 by Interstellar

[mgutt]

12 minutes ago, Interstellar said:

As for iGPU, turns out something is installed

This is fine. Especially for the 12th Gen and upwards iGPU it lowers power consumption in idle.

[kilonde]

On 10/30/2022 at 11:40 AM, J05u said:

When I run autotune i am getting 

powertop: /lib64/libncursesw.so.6: no version information available (required by powertop)
powertop: /lib64/libtinfo.so.6: no version information available (required by powertop)
modprobe cpufreq_stats failed
Loaded 0 prior measurements
Cannot load from file /var/cache/powertop/saved_parameters.powertop
File will be loaded after taking minimum number of measurement(s) with battery only 
RAPL device for cpu 0
RAPL device for cpu 0
Devfreq not enabled
glob returned GLOB_ABORTED
Cannot load from file /var/cache/powertop/saved_parameters.powertop
File will be load

 

In tunables all showing as good

 

But in Pkg(HW) as on screenhot bad, all zeros

 

What i can do to fix it?

I get the same error, have you managed to get it working? I'm using 10th gen Intel.

[mgutt]

36 minutes ago, kilonde said:

I get the same error,

There aren't any errors. Only warnings or irrelevant informations.

[kilonde]

On 12/5/2022 at 11:24 AM, mgutt said:

There aren't any errors. Only warnings or irrelevant informations.

Thanks, based on that output, the auto tune does not run, or am I wrong?

 

Google does not help much unfortunately.

[mgutt]

5 hours ago, kilonde said:

the auto tune does not run

It does and it does not "run". It changes only Linux configs. Easy verification: execute powertop and use TAB to switch to the last tab. Every entry is "good".

[BarbaGrump]

On 11/4/2020 at 7:30 PM, CS01-HS said:

I saw a few of these hard resetting link errors during my mover run. Thankfully (?) no CRC errors reported. ata3 is a spinning disk attached to an integrated ASM1062 controller.

 

I wonder if it might be related to the power-saving tweaks because nothing else changed. For now I've disabled them and will see if they reappear. Maybe coincidence but I'm posting in case others have the same issue.

Nov  3 23:17:30 NAS move: move: file /mnt/cache/Download/movie_1.mp4
Nov  3 23:17:33 NAS kernel: ata3.00: exception Emask 0x10 SAct 0x80 SErr 0x4050002 action 0x6 frozen
Nov  3 23:17:33 NAS kernel: ata3.00: irq_stat 0x08000000, interface fatal error
Nov  3 23:17:33 NAS kernel: ata3: SError: { RecovComm PHYRdyChg CommWake DevExch }
Nov  3 23:17:33 NAS kernel: ata3.00: failed command: WRITE FPDMA QUEUED
Nov  3 23:17:33 NAS kernel: ata3.00: cmd 61/00:38:58:44:51/04:00:2c:02:00/40 tag 7 ncq dma 524288 out
Nov  3 23:17:33 NAS kernel:         res 40/00:30:58:40:51/00:00:2c:02:00/40 Emask 0x10 (ATA bus error)
Nov  3 23:17:33 NAS kernel: ata3.00: status: { DRDY }
Nov  3 23:17:33 NAS kernel: ata3: hard resetting link
Nov  3 23:17:33 NAS move: move: file /mnt/cache/Download/movie_1.mp4
Nov  3 23:17:33 NAS kernel: ata3: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Nov  3 23:17:33 NAS kernel: ata3.00: supports DRM functions and may not be fully accessible
Nov  3 23:17:33 NAS kernel: ata3.00: supports DRM functions and may not be fully accessible
Nov  3 23:17:33 NAS kernel: ata3.00: configured for UDMA/133
Nov  3 23:17:33 NAS kernel: ata3: EH complete
Nov  3 23:17:35 NAS move: move: file /mnt/cache/Download/movie_2.mp4

 

This is exactly what I've experienced...see this post :

In short...my SATA-controller dont like some of the tunables coming from "powertop --auto-tune", but running all commands(except the SATA-one) suggested in 1st post works like a charm. 

[mgutt]

3 hours ago, BarbaGrump said:

In short...my SATA-controller dont like some of the tunables coming from "powertop --auto-tune

Depending on the used controller a firmware update can fix this:

https://forums.unraid.net/topic/102010-recommended-controllers-for-unraid/?do=findComment&comment=1185707

[BarbaGrump]

23 hours ago, mgutt said:

Depending on the used controller a firmware update can fix this:

https://forums.unraid.net/topic/102010-recommended-controllers-for-unraid/?do=findComment&comment=1185707

Well, in my case its a mobo-integrated ASmedia 1164 controller in the "Asustor Lockerstor 4 2nd gen" which I run Unraid on, updated with latest bios provided from Asustor(v1.17)

[mgutt]

2 minutes ago, BarbaGrump said:

updated with latest bios

I don't think this updates the controller firmware, but I'm not sure. Maybe ask asustor.

[CS01-HS]

On 12/14/2022 at 8:09 AM, BarbaGrump said:

In short...my SATA-controller dont like some of the tunables coming from "powertop --auto-tune", but running all commands(except the SATA-one) suggested in 1st post works like a charm. 

 

My fix was to move the spinning disks from the integrated ASM controller to the integrated intel. I've had no problems running autotune with only SSDs on the ASM.

 

Before that, with mgutt's explicit config/go commands as a basis, I excluded the ASM's host3 and host4 from power-saving - intel are 1 and 2. 

linkpm_count=1
for i in /sys/class/scsi_host/host*/link_power_management_policy; do
   echo 'med_power_with_dipm' > $i
   if [ $linkpm_count -eq 2 ]; then
     break
   fi
   ((linkpm_count++))
done

 

[kilonde]

On 12/8/2022 at 4:54 PM, mgutt said:

It does and it does not "run". It changes only Linux configs. Easy verification: execute powertop and use TAB to switch to the last tab. Every entry is "good".

Thanks again, that makes sense, the tunables tab is showing good for all parameters. Yet my Idle stats dont look promising:

 

 

And it looks like only two frequencies - 800MHz or 3.6 GHz

 

 

Can I make some further changes? All C states are enabled in bios, no VMs. 

[mgutt]

Added two interesting things to the initial post:

Quote

MTU

Surprisingly a user was not able to reach any C-State other than C1E after setting the MTU to 9000.

 

NVMe

A user found out that a specific NVMe SSD allowed only C2, while with a Samsung Evo Plus the system reached C10.

 

[J05u]

powertop autotune should be entered inside powertop, or I can just run in terminal?

can't figure out where I can run it inside powertop in terminal

[Marc Heyer]

17 minutes ago, J05u said:

powertop autotune should be entered inside powertop, or I can just run in terminal?

can't figure out where I can run it inside powertop in terminal

Yes, in terminal just type:

powertop --auto-tune

[JorgeB]

@mguttcould you please add to the first post that if users start getting ATA link errors, or even disks dropping after using this to first try to disable it? It seems multiple controllers have issues with this.

[Rick Sanchez]

My Xeon setup uses 120w in use and idle! I wish I could drop its usage massively!

 

Can this help?

[cuddles]

On 10/21/2020 at 1:05 PM, mgutt said:

 

PCIe cards

Some PCIe cards prevent states other than C0 or C3 (Intel). As an example my Qnap Aquantia 10G card allowed only C3 while all my Intel 10G cards allowed C7. Without any installed card I'm able to reach C10.

 

 

What exact Intel Cards you own and use to reach C7 ?

[ryry]

Well, I guess I did a stupid.  I'm away from home at the moment and decided to run the 3 install commands from OP then reboot remotely.  Not home for a few days.  After shutting down, Unraid isn't even getting an IP - does anyone see how running the commands at the beginning of the post then restarting could result in a no-boot scenario?

 

For some extra info, it's an i7-3770k on a z77 motherboard.

 

edit 2: maybe it was already installed via the tips and tweaks plugin? feeling pretty dumb right now...

Edited December 28, 2022 by ryry
more info

[mgutt]

26 minutes ago, ryry said:

maybe it was already installed via the tips and tweaks plugin

As far as I know the plugin does not contain it anymore.

 

26 minutes ago, ryry said:

the post then restarting could result in a no-boot scenario?

Never heard about something similar.

 

52 minutes ago, cuddles said:

What exact Intel Cards you own and use to reach C7 ?

X550-T with Single RJ45 and X710 with Dual SFP+

[ryry]

I had someone try and restart it. Still no boot. Invisible to router. Is there a common debug path to a no boot scenario? Would like to avoid a screen but I don’t think I have any options. 

[mgutt]

9 hours ago, ryry said:

Is there a common debug path to a no boot scenario?

- Delete the two network config files in /config

- move "menu default" in syslinux/syslinux.cfg after "label Unraid OS Safe Mode":

 

by that your server boots in safe mode and obtains a new IP address through your DHCP server.

 

[mgutt]

On 12/27/2022 at 4:31 PM, JorgeB said:

@mguttcould you please add to the first post that if users start getting ATA link errors, or even disks dropping after using this to first try to disable it? It seems multiple controllers have issues with this.

I added a warning:

 

And a further explanation to the initial post: