Topton Jasper Lake Quad i225V Mini PC Report

[lukasig]

I will later try the BIOS settings you suggested and come back with a reply.

Since I have fiddled with the BIOS settings quite a bit, I first restored the the settings to default, and then proceeded to only change the settings you mentioned above. To my surprise, there was no change to the power consumption at the wall (still ~9W idle), even though I use the exact same hardware and power adapter as you...

Maybe resetting to defaults messed with other BIOS settings?

 

[T.Sharp]

Since I have fiddled with the BIOS settings quite a bit, I first restored the the settings to default, and then proceeded to only change the settings you mentioned above. To my surprise, there was no change to the power consumption at the wall (still ~9W idle), even though I use the exact same hardware and power adapter as you...

Maybe resetting to defaults messed with other BIOS settings?

I've reset the bios (same version you have) a few times, and when I change only those settings above, it drops from ~8W idle to ~5W. Balanced power plan in Win10. I did update everything thru Windows update, and disabled a couple things like News feed, cortana search, dropbox etc. Maybe there's just more junk running in the background, or it's a chipset driver thing?

I'm using a Belkin power meter just like the one in the video posted a while back, and it's possible that it's not super accurate. Not sure what other variables there could be.

 

[lukasig]

I've reset the bios (same version you have) a few times, and when I change only those settings above, it drops from ~8W idle to ~5W. Balanced power plan in Win10. I did update everything thru Windows update, and disabled a couple things like News feed, cortana search, dropbox etc. Maybe there's just more junk running in the background, or it's a chipset driver thing?

I'm using a Belkin power meter just like the one in the video posted a while back, and it's possible that it's not super accurate. Not sure what other variables there could be.

I've also tested with ubuntu and same result. Could also be my power meter, but i've tested power under 1W with it and seems to work fine. Besides, for me there is no change at all in power consumtion if I change the ASPM settings...

EDIT: did your win10 have drivers for the NICs? or do you use a wifi card?

 

[T.Sharp]

EDIT: did your win10 have drivers for the NICs? or do you use a wifi card?

It didn't have the NIC drivers before I ran Windows update, I have a USB Eth dongle I used for getting everything downloaded first.

EDIT : I found a bug. If I have ASPM set to auto on port 8, the system will not reboot when I hit restart inside of Windows. It shuts down completely and the power button won't turn it back on. I have to disconnect/reconnect power to get it to boot again. If I disable ASMP for only port 8, reboot works like it should, but idle power is ~0.5W higher.

I disabled port 8 in BIOS and NIC #3 disappeared from device manager. So 3 of the 4 NICs can be set to ASPM-auto without issue, but 1 cannot? Seems strange. No clue if these settings will break something in proxmox or pf/opnsense yet. Going to start messing with that today.

Also the CPU Mark score with the N5100 was 4365, much higher than the 3350 suggested on the Passmark site.

 

[lukasig]

It didn't have the NIC drivers before I ran Windows update, I have a USB Eth dongle I used for getting everything downloaded first.

ok...
thanks again for your BIOS settings info. I installed windows 10 today, and the curious thing is that... yes, the idle power consumption dropped to 4W.So I guess is more of a driver optimization thing. Anyway... I am quite happy with the result. Now I curious once I install pfsense on it, how low I can get it to stay when idle. Will come back with results.

 

[lukasig]

I've reset the bios (same version you have) a few times, and when I change only those settings above, it drops from ~8W idle to ~5W. Balanced power plan in Win10. I did update everything thru Windows update, and disabled a couple things like News feed, cortana search, dropbox etc. Maybe there's just more junk running in the background, or it's a chipset driver thing?

ok...
thanks again for your BIOS settings info. I installed windows 10 today, and the curious thing is that... yes, the idle power consumption dropped to 4W.So I guess is more of a driver optimization thing. Anyway... I am quite happy with the result. Now I curious once I install pfsense on it, how low I can get it to stay when idle. Will come back with results.

It didn't have the NIC drivers before I ran Windows update, I have a USB Eth dongle I used for getting everything downloaded first.

for offline drivers, I've been using Snappy Driver Installer for years now, and I think is one of the most useful things related to windows. Pretty much all drivers ever on your USB key.

 

[T.Sharp]

ok...
thanks again for your BIOS settings info. I installed windows 10 today, and the curious thing is that... yes, the idle power consumption dropped to 4W.So I guess is more of a driver optimization thing. Anyway... I am quite happy with the result. Now I curious once I install pfsense on it, how low I can get it to stay when idle. Will come back with results.

No problem! Glad you got it working now.

for offline drivers, I've been using Snappy Driver Installer for years now, and I think is one of the most useful things related to windows. Pretty much all drivers ever on your USB key.

Yep, I use that sometimes as well, definitely handy. Usually I prefer to use Windows Update when possible, because it picks stable / tested drivers, even if they aren't the latest. Sometimes SD will recommend drivers that cause issues ime, or for things like intel chipset drivers, it will recommend the generic versions (usually dated 1968) over newer hardware specific drivers(?).

 

[sovking]

May you test these machines with freebsd/opnsense/pfsense to see it the power consumption is more similar to Linux or to Windows ?

 

[lukasig]

Yep, I use that sometimes as well, definitely handy. Usually I prefer to use Windows Update when possible, because it picks stable / tested drivers, even if they aren't the latest. Sometimes SD will recommend drivers that cause issues ime, or for things like intel chipset drivers, it will recommend the generic versions (usually dated 1968) over newer hardware specific drivers(?).

Yes, over time I also came to prefer the windows update drives for stability, but in ceses of missing network drivers on fresh install, SDI has its very good use.

 

[lukasig]

May you test these machines with freebsd/opnsense/pfsense to see it the power consumption is more similar to Linux or to Windows ?

not sure your question is very clear... at least for me...

 

[T.Sharp]

May you test these machines with freebsd/opnsense/pfsense to see it the power consumption is more similar to Linux or to Windows ?

With the base VGA install of opnsense, and PCI-E port ASPM enabled in bios, it's idling at 7W with 1 NIC connected. Doesn't seem to have the reboot issue with port 8 ASPM like Windows does.

I was confused for a bit because opnsense has a "beep" script that runs at startup / shutdown, but it makes a clicking / ticking sound, not the beep that bios plays on boot. I thought maybe the VRM or some other component was about to pop.

EDIT: I found a guide for enabling C3 states in opnsense, and now the power is down to 4.8W idle. The last comment in the thread states " Be aware that using lower C-states may introduce spurious instability - my Topton N5105-based unit showed that behaviour once I went lower than C1."

 

[sovking]

So with two-nic standard configuration of opnsense, it should use around 8-9W... So for limited bandwidth connection (100-300 Mbps), an APU2 or J4105 use a little bit less power for typical small/home firewall and routing.

 

[T.Sharp]

So with two-nic standard configuration of opnsense, it should use around 8-9W... So for limited bandwidth connection (100-300 Mbps), an APU2 or J4105 use a little bit less power for typical small/home firewall and routing.

I'm think with some tweaks, a single DIMM and a lower power SSD (mine has 4 NAND chips, if that matters), the power would be very similar. Each NIC seems to add ~0.5W at idle when connected, with ASPM enabled.

Not sure if the 5W idle using C3 states will be stable, going to take more experimenting.

Update : I have opnsense running as my primary router now, using 2 NICs. With ASPM, C3 states, and "powerD" set to minimum in opnsense, it's sitting at ~6W most of the time. It will hit 13W running speedtest with a ~470mb/s connection.

Surprisingly my Asus AC68U in AP-only mode, with 2.4GHz radio disabled and one eth LAN connection, draws 8W at idle...

Now to setup QoS and maybe VPN if I can figure out how to get NordLynx working with Wireguard.

 

[sqrwv]

Update : I have opnsense running as my primary router now, using 2 NICs. With ASPM, C3 states, and "powerD" set to minimum in opnsense, it's sitting at ~6W most of the time. It will hit 13W running speedtest with a ~470mb/s connection.

If possible try to see if/how ASPM and/or those C states affect latency and burst speed.
For example, compare pinging your LAN or do ipferf3 =1s burst(s) (with long idle time between bursts), to compare latency and burst speed, with those power savings from idle and without them.

 

[lukasig]

The only settings I changed in BIOS was under --- Chipset > PCH-IO Config > PCI-E Root Port (1-8) > ASPM set to Auto.
I did this for each PCI-E root port. EDIT : as well as DMI Link ASPM set to Auto.

i've done a bit more testing with the ASPM set to "Auto", and something strange came up. When using windows everything seems to be ok, but if using ubuntu server, once I unplug the HDMI cable, the PC immediately shuts off. If I start it with no connected display, it boots and then shuts off automatically....

 

[T.Sharp]

If possible try to see if/how ASPM and/or those C states affect latency and burst speed.
For example, compare pinging your LAN or do ipferf3 =1s burst(s) (with long idle time between bursts), to compare latency and burst speed, with those power savings from idle and without them.

Right now I only have a single desktop connected to the switch, not much of a home network. Pinging my laptop over the Asus AP shows latency bouncing around quite a bit, but I don't think that's a useful test. I'll see about getting another wired PC hooked up. I don't know anything about using iperf3, but I'll look into it.

I have "(FQ-)CoDel ECN" configured in Shaper for bandwidth limiting, and the bufferbloat test shows +0 added latency under load. Pinging their servers is ~24ms, which is a bit lower than typical when I was using the AC68U as the router. Speedtest from Spectrum ISP to Spectrum server (about 100 miles), shows ping of ~12ms, which is also lower than I was seeing previously. I know those won't tell you much about LAN performance, but that's all I have at the moment.

i've done a bit more testing with the ASPM set to "Auto", and something strange came up. When using windows everything seems to be ok, but if using ubuntu server, once I unplug the HDMI cable, the PC immediately shuts off. If I start it with no connected display, it boots and then shuts off automatically....

Very strange. Have you tried disabling ASPM on port 8? That was giving me issues with reboot / power on with Windows. I also have the auto power-on jumper and bios setting disabled currently, so maybe that plays into it? I haven't had any problems rebooting OPNsense yet, but will re-enable the jumper and see if anything changes.

 

[TripleZ]

How is the BIOS support on these - any source of updates? I've never used it, but is it possible to use something like coreboot?

 

[lukasig]

Very strange. Have you tried disabling ASPM on port 8? That was giving me issues with reboot / power on with Windows.

not yet... will try tomorrow
The problem is only using the ubuntu server. I installed pfsense and there is no issue not using a display. But even with powerd enabled, pfsense idles at 8-9W.

I also have the auto power-on jumper and bios setting disabled currently, so maybe that plays into it? I haven't had any problems rebooting OPNsense yet, but will re-enable the jumper and see if anything changes.

Yes, for now my auto power-on jumper is also disabled.

 

[T.Sharp]

But even with powerd enabled, pfsense idles at 8-9W.

Sounds like pfsense is disabling lower C-states too. I don't think PowerD did anything to reduce power draw until C3 states were manually enabled. I'm actually not sure if PowerD makes a difference even with C3 is enabled, I just had it set to minimum and left it there.
EDIT: PowerD is not used by modern chips

 

[sqrwv]

Right now I only have a single desktop connected to the switch, not much of a home network. Pinging my laptop over the Asus AP shows latency bouncing around quite a bit, but I don't think that's a useful test. I'll see about getting another wired PC hooked up. I don't know anything about using iperf3, but I'll look into it.

There's latency turning CPU and PCIe from low sleep states to active, so if they are idle/sleeping, the first time you access a site, it will have the added time of the hardware waking up.
I don't know if it's ms or us (micro-seconds) delay, if usec. we won't notice it.

Buffer bloat can't show it, because in that test it is already doing a full download.

Active State Power Management - Wikipedia
" While ASPM brings a reduction in power consumption, it can also result in increased latency as the serial bus needs to be 'woken up' from low-power mode, possibly reconfigured and the host-to-device link re-established. This is known as ASPM exit latency and takes up valuable time which can be annoying to the end user if it is too obvious when it occurs. "