STH measured them at 23W idle so yes, they should go well below that. I don't even think their power draw under load was that high.
It's pretty hard to answer your question though, seeing as you have told us nothing about your setup.
Qotom Denverton fanless system with 4 SFP+
- Thread starter is39
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
It's pretty hard to answer your question though, seeing as you have told us nothing about your setup.
Sorry, I'll be back near the system next week, it's about 300km away from me and I don't have remote power monitoring.STH measured them at 23W idle so yes, they should go well below that. I don't even think their power draw under load was that high.
It's pretty hard to answer your question though, seeing as you have told us nothing about your setup.
I will double check if I was also measuring the drives with that number. Confirming that it is actually clocking down around 1ghz while idle (via powertop).
System has 64gb/RAM, one Ethernet connection and a 4 disk drive bay plugged into the SFF port.
Replying to myself, but... I would suggest avoiding the Nemix RAM kits. The one I received (2x32GB ECC) does POST correctly. Per Qotom's warnings, there's a fairly long wait before you get a display and BIOS beep, but it does at least boot... but both sticks in the kit are throwing ECC errors in Passmark's free edition of Memtest. The Nemix kit I got does have Micron chips on it, or at least they are labeled as Micron, but who knows if they are legit or counterfeit...Thanks, I poked around and realized the search is able to search within a thread too, just have to click some options.
I was looking at one of those "Nemix" kits on Newegg, figured that if they sell it for Denverton-based ASUSTOR NAS boxes, it would probably work...
The FOSS
memtest86+ didn't throw any errors at all, but it did not seem to be monitoring ECC on this platform...I have another 64GB kit on order, this set directly from Crucial/Micron (and it only cost like $10 more than the Nemix kit). , hopefully the legit Micron kit isn't tossing ECC errors and is clean.
I got a motherboard PDF from Qotom, and have a non-fuzzy pinout now to clarify the headers a little better.
Attachments
I also found this in the mainboard manual: 
Yet the M.2 slots don't mention being shared with the WWAN socket:
I could confirm that when installing 2x M.2 NVMe + M.2 WWAN all 3 devices show up and work as expected. So it seems the information in the product description is wrong and the WWAN socket actually shares the PCIe lane(s) with the USB3.0 controller that drives the second port on the front panel.
The only problem however with those WWAN cards is, that they are insane space-heaters and the card will sit directly underneath both M.2 disks, leading to extremely high temperatures on those. I'm running mine without a WWAN module now - my primary uplink is very reliable and mobile connections are pretty much useless anyways thanks to CGNAT...
Yet the M.2 slots don't mention being shared with the WWAN socket:
I could confirm that when installing 2x M.2 NVMe + M.2 WWAN all 3 devices show up and work as expected. So it seems the information in the product description is wrong and the WWAN socket actually shares the PCIe lane(s) with the USB3.0 controller that drives the second port on the front panel.
The only problem however with those WWAN cards is, that they are insane space-heaters and the card will sit directly underneath both M.2 disks, leading to extremely high temperatures on those. I'm running mine without a WWAN module now - my primary uplink is very reliable and mobile connections are pretty much useless anyways thanks to CGNAT...
Got my official Crucial kit (Crucial/Micron SKU MTA18ASF4G72HZ-3G2R [link goes straight to official Crucial store; mods can remove if this isn't kosher]) installed; passed 4 rounds of Passmark Memtest with zero errors (not even any ECC errors) in a "softball" test (bottom off of the system with it standing on its side in front of a fan to keep RAM cool as possible). It's now running through a second round, this time with the case closed up and sitting like normal. RAM temps appear to be way higher this time around (75C, if Passmark is accurate), but still no errors.
I feel like these passive systems could potentially benefit from a little bit of breathing room on the underside for the RAM & M.2 drives... Thinking about cutting a hole or two down there and gluing some mesh in place (my system will live in a crawlspace that has a fair share of small spiders that I don't want getting inside).
Last edited:
I'm having such a strange issue with my rackmount version with the C3758R variant. I installed Pfsense and i have an issue with the speeds it is giving me. I have a 1Gbps/40Mbps asymmetric internet service and with my old Supermicro c2000 pfsense machine gives me 960Mbps/40Mbps consistent.
However when I try it with the QOTOM it gives me 450Mbps/10Mbps on any 2.5Gb NIC (haven't tested the 10gb ports just yet)
Also I had to disable hardware offloading for it to be stable, otherwise the interface is buggy and drops connections intermittently.
However when I try it with the QOTOM it gives me 450Mbps/10Mbps on any 2.5Gb NIC (haven't tested the 10gb ports just yet)
Also I had to disable hardware offloading for it to be stable, otherwise the interface is buggy and drops connections intermittently.
If you run top, what sorta interrupt rate are you seeing? If you see ksoftirqd pegging a core, you're CPU-bound, which could happen if you're unable to use hardware offloading. EDIT jut realized you're on pfsense, not sure what the equivalent would be on bsdHowever when I try it with the QOTOM it gives me 450Mbps/10Mbps on any 2.5Gb NIC (haven't tested the 10gb ports just yet)
Also I had to disable hardware offloading for it to be stable, otherwise the interface is buggy and drops connections intermittently.
Hardware offloading of what?I'm having such a strange issue with my rackmount version with the C3758R variant. I installed Pfsense and i have an issue with the speeds it is giving me. I have a 1Gbps/40Mbps asymmetric internet service and with my old Supermicro c2000 pfsense machine gives me 960Mbps/40Mbps consistent.
However when I try it with the QOTOM it gives me 450Mbps/10Mbps on any 2.5Gb NIC (haven't tested the 10gb ports just yet)
Also I had to disable hardware offloading for it to be stable, otherwise the interface is buggy and drops connections intermittently.
Is it a bare metal install or a virtualized one?
pppoE connection ?I'm having such a strange issue with my rackmount version with the C3758R variant. I installed Pfsense and i have an issue with the speeds it is giving me. I have a 1Gbps/40Mbps asymmetric internet service and with my old Supermicro c2000 pfsense machine gives me 960Mbps/40Mbps consistent.
However when I try it with the QOTOM it gives me 450Mbps/10Mbps on any 2.5Gb NIC (haven't tested the 10gb ports just yet)
Also I had to disable hardware offloading for it to be stable, otherwise the interface is buggy and drops connections intermittently.
They mentioned pfSense... the default setup disables NIC hardware offload for most (if not all) NICs by default. I'm really not sure why they do this, as Intel NIC drivers in FreeBSD should be fine.Hardware offloading of what?
Is it a bare metal install or a virtualized one?
That said, pfSense has been a bit "behind" with the Intel drivers and I know that for a period of time it couldn't handle the I226V that is in these systems. @Rolfe Shepsky have you considered trying OPNsense instead?
There are many different offload options, hence the question.
Yeah, BSD has worse driver support, sometimes significantly so. OPNsense used to be based on more recent FreeBSD versions but that no longer appears to be the case. In fact, it's the opposite now. Still doesn't hurt to try it though.
OPNsense 24.7 which comes out this month is supposed to be based on FreeBSD 14.1-RELEASE...OPNsense used to be based on more recent FreeBSD versions but that no longer appears to be the case. In fact, it's the opposite now. Still doesn't hurt to try it though.
(I'm actually waiting for it to come out to set up my Qotom box that just arrived this week. Very excited... unless you guys tell me I shouldn't be...)
pfSense CE has been based on FreeBSD 14.0-CURRENT since 2023-06-29 and pfSense Plus since 2023-02-15. pfSense Plus 24.03 and later are based on FreeBSD 15.0-CURRENT.
Versions of pfSense software and FreeBSD | pfSense Documentation
In other words, OPNsense has fallen behind quite a bit.
So, people still stuck with Kernel pinning with Proxmox VE v8? Proxmox VE v7. x hits EOL in 19 days....
Building Intel's version of the driver against the matching kernel headers and loading that as a module would presumably also work.
I run VyOS bare metal on mine, and they already switched to Intel's driver.