exactly what I was thinking.. and this makes me want to get rid of it.. and start a new project by building my own OPNsense box with probably older, but stronger hardware from trusted brands!
CWWK/Topton/... Nxxx quad NIC router
- Thread starter Becks0815
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
i think it maybe something to do with the i226 chips in these.
with opnsense running and no network cables connected to it i saw 7.3w power pull, thats with the fan still plugged into its usb port. now i know windows has lower power draw at idle so could be even lower.
what i will have to do is retest this at some point with the wall plug power meter and see how power usage is.
i suspect in bsd these nics run flat out or there is some sort of limit on how low the box will downclock or its not using all its lower power states.
this could be because n100 is newer than the older stuff that has less power consumption probably because of cpu microcode or kernel limitations on n100.
will test some more, its the only way i will get to the bottom of this.
with opnsense running and no network cables connected to it i saw 7.3w power pull, thats with the fan still plugged into its usb port. now i know windows has lower power draw at idle so could be even lower.
what i will have to do is retest this at some point with the wall plug power meter and see how power usage is.
i suspect in bsd these nics run flat out or there is some sort of limit on how low the box will downclock or its not using all its lower power states.
this could be because n100 is newer than the older stuff that has less power consumption probably because of cpu microcode or kernel limitations on n100.
will test some more, its the only way i will get to the bottom of this.
Hello,
I'm happy CWWK N100 2 port version user but I needed another device to have replacement and decided to go with mentioned here ODROID H4 Ultra (N305)
And I must say that its way better than CWWK (power consumption, support(bios updates), IO ports, pcie 3.0x4 adapters, temperatures)
I have created overview video, someone might find it useful:
If someone missed my cwwk videos (which tbh are must have to watch before buy), here they are:
I'm happy CWWK N100 2 port version user but I needed another device to have replacement and decided to go with mentioned here ODROID H4 Ultra (N305)
And I must say that its way better than CWWK (power consumption, support(bios updates), IO ports, pcie 3.0x4 adapters, temperatures)
I have created overview video, someone might find it useful:
If someone missed my cwwk videos (which tbh are must have to watch before buy), here they are:
That two port unit looks to be well designed with low power usage and good idle consumption, unfortunately I need a few more ports.
I tried with Windows as well, made sure all the drivers were installed and up to date. Using some monitoring software I could see the cores throttling all the way down and lower voltages reported, with the total package consumption estimated at 1.5 to 2 watts. At the plug the power draw was still around 10 watts. From experimenting (with it and without it), the SSD at idle is only adding about 0.1 watt. The network chips according to Intel's datasheet at 1G speeds draw a maximum of 0.9 W and 1.4 W when 2.5G, but this will be a lot less on typical short cable runs of a couple of metres. In my tests, unplugging a cable running a couple of metres negotiated at 2.5G shows it is using around 0.3 W.
pfSense shows that around 80% of the time the cores are at C3, and 15% of the time at C2.
So to recap (both headless so no keyboard or monitor):
Older 6 port i7 7500U (Kettop Mi7500L6) with 3 x 1Gbps network cables attached, 8Gig stick of memory and SSD, active PPPoE connection and routing background traffic for the entire home network, barely exceeds 5 watt idle, it jumps to about 10 watt maxing out the connection. Top of the box feels cool to the touch at the moment (room is 20c) as cooler than my hand temperature. Cores are running at 27c as reported in pfSense. The package is rated at 15W TDP.
6 port N100 (CWWK motherboard) with 2 network connections (1G and 2.5G), 8Gig memory and SSD, routing the background traffic from one PC as I'm just testing it, is 10-11 watts idle, peaking to around 15 to 17 watts when maxing out the connection. The heatsink feels warm, and temperatures reported are 36 degrees at idle. The best idle consumption with no network connections is about 9.4 watts. Package is 6W TDP.
I've tested using the same 12 volt supply on both to rule out differences there (identical wattage reported).
Given the N100 is a much newer process than my comparison to an existing i7 7500U type box, with the N100 advertised to give the same performance at much less power, or more performance for the same power as previous Intel chips, this is certainly not the case the way they implemented in some of these boxes, the opposite in fact. So something just isn't right with these. Somewhere something is burning 6 to 7 watts or so on just creating heat, which doesn't inspire much confidence.
Be nice to get to the bottom of it, but is it fixable by BIOS/driver or software settings, or is this just a poor hardware design?
Last edited:
Hello
I have this same board but unfortunatelly i am unable to flash the USB stick with Balena flashing tool under windows. It says the the image is not bootable and thats it. Any suggestions?
in case anyone is interested. tested a hp prodesk 400 g4 desktop mini with 256gb wd sn810 nvme, 2.5 mechanical drive, 16gb ram dual channel and i5 8500t cpu. running win11 it was idling around 5.7w from the wall. uses original hp 90w power adapter.
thats really good for these units. without the mechanical drive it would probably hit 5w on the dot.
thats really good for these units. without the mechanical drive it would probably hit 5w on the dot.
Hi everyone
Got a N305 V2 with bios 09/27/2023 there is a new bios released 08/11/2023 Did someone already upgrade and could share feedback? I noticed some kernel crashes with Proxmox recently, and I already disable some power savings mentioned here Known Issues Proxmox VE 8.2 – Thomas-Krenn-Wiki
Got a N305 V2 with bios 09/27/2023 there is a new bios released 08/11/2023 Did someone already upgrade and could share feedback? I noticed some kernel crashes with Proxmox recently, and I already disable some power savings mentioned here Known Issues Proxmox VE 8.2 – Thomas-Krenn-Wiki
These sorts of tests of other equipment really confirm that these range of N100 network appliances from China, idling at 10-12 watt really do have something wrong with them, probably why they are all over Aliexpress being sold for very little money, and given they are burning through at least 6-7 watts possibly more than they should when idle, how long will they last running 24/7?
A network appliance should be setting the example for idle power consumption as they don't have sound cards or other peripherals you find on desktop type machines, no fans to power, less memory and they run headless a lot of the time (GPU throttled down), and are powered from low wattage more efficient power supplies. Given they are on 24/7 and mostly idle for a lot of time should be designed to waste no power whatsoever.
I think sites like this should be calling them up on this high idle power consumption, rather than glossing over it is "okay", and getting some answers.
yeah the idle power is way off. This person did a detailed analysis of a n100 device and its idle power is 4.2 watts which seems more accurate for a cpu that has a 6watt TDP. Their test was done on Ubuntu 23.04 (Lunar Lobster) on Linux kernel 6.2.0-26-generic. The device was purchased from aliexpress and seems to be a non branded mini pc. So I am not sure if saying just because its a cheap device that its faulty. As this persons n100 device is cheap as well.
Intel N100: A Challenge to ARM - bret.dk
The Intel N100 (and N200) seem to be Intel's attempt at trying to muscle into the market that high-end ARM SoCs look to have cornered. With the Rockchip
bret.dk
I will see if I can do some power testing with wall plug meter on the n100 box over the weekend. It's currently running home internet so can't just take it down.
Will get windows on it and test because I saw very low idle temps in windows which shows c states are working properly
Will get windows on it and test because I saw very low idle temps in windows which shows c states are working properly
As I've partly said before:
1) I think some of this is the difference between say an HP tiny with an R&D budget that probably dwarfs the Chinese manufacturers sales.
2) Another factor could be the inversion of the normal reminder that a 35w or 65w tdp does not mean that a cpu won't idle at low power. Perhaps the N100 "is so efficient anyway" that Intel didn't do a great job of idle management. Threads have shown us that it hates C states if even one peripheral isn't 100% compatible. Maybe that was true of many older processors, but as we expected 35 or 65 watts no-one really noticed.
3) in most systems there's a bridge/io chipset handling many of the peripherals and power management. N100 is all off the cpu, so maybe it's not as flexible. Or maybe the Chinese design is behind the differences and so is weak.
4) a common factor in the Chinese systems is the 2.5g networking. They're the first to give us that in cheap systems. Or many systems at all.
4a) It could be they've powered those badly, and they'd certainly need access to more watts that a ten year old 1g design. Even a watt of wasted power per port would be 4/6 watts across the system. That would be on the manufacturer, but hey $150.
4b) we know intel's 2.5g chips had lots of problems until recently. To give the benefit of the doubt maybe they've been made more stable with more aggressive power supplies. So waste a few watts at idle but keep the links up.
5) These systems also give us multiple usb3 and multiple displays. Even without design deficiencies that's a few more watts. Again idle usage might be CWWK's fault but we're getting options few others give us.
This is all conjecture, but once we're in the 0-15w range it's easy for small problems to make a big difference. Not too many years ago random peripheral chips could have eaten that unnoticed.
The point about the older 6 port device and it's better idle is fair. Assuming you want: 1g networking, less usb ports & speed, No NVME. Oh and for me in the UK it's well over twice the price £383 barebones. The branded NUCs are even more money and likely to give us a single 1g network. So different beasts entirely.
So in some ways the likes of cwwk aren't charging us enough to help them do it better. And we might not buy from them if it was more expensive. Catch 22.
1) I think some of this is the difference between say an HP tiny with an R&D budget that probably dwarfs the Chinese manufacturers sales.
2) Another factor could be the inversion of the normal reminder that a 35w or 65w tdp does not mean that a cpu won't idle at low power. Perhaps the N100 "is so efficient anyway" that Intel didn't do a great job of idle management. Threads have shown us that it hates C states if even one peripheral isn't 100% compatible. Maybe that was true of many older processors, but as we expected 35 or 65 watts no-one really noticed.
3) in most systems there's a bridge/io chipset handling many of the peripherals and power management. N100 is all off the cpu, so maybe it's not as flexible. Or maybe the Chinese design is behind the differences and so is weak.
4) a common factor in the Chinese systems is the 2.5g networking. They're the first to give us that in cheap systems. Or many systems at all.
4a) It could be they've powered those badly, and they'd certainly need access to more watts that a ten year old 1g design. Even a watt of wasted power per port would be 4/6 watts across the system. That would be on the manufacturer, but hey $150.
4b) we know intel's 2.5g chips had lots of problems until recently. To give the benefit of the doubt maybe they've been made more stable with more aggressive power supplies. So waste a few watts at idle but keep the links up.
5) These systems also give us multiple usb3 and multiple displays. Even without design deficiencies that's a few more watts. Again idle usage might be CWWK's fault but we're getting options few others give us.
This is all conjecture, but once we're in the 0-15w range it's easy for small problems to make a big difference. Not too many years ago random peripheral chips could have eaten that unnoticed.
The point about the older 6 port device and it's better idle is fair. Assuming you want: 1g networking, less usb ports & speed, No NVME. Oh and for me in the UK it's well over twice the price £383 barebones. The branded NUCs are even more money and likely to give us a single 1g network. So different beasts entirely.
So in some ways the likes of cwwk aren't charging us enough to help them do it better. And we might not buy from them if it was more expensive. Catch 22.
Not really. Just for you, i plugged in the original psu for my device (x86-p5), and did a photo. My idle is between 5.6 and 6.1 watts, while both ports of i226v are connected at 2.5gbps speed.
This device is cwwk x86-p5, with 8gb ram, no ssd, booting from usb sd card reader + sd card, running openwrt 23.05. If you can't go down this low with your device, then either your bios settings, os settings, or used hardware are the issues. I don't have a photo, but my acemageic t8 plus with N95 and two gigabit realtek lan can idle around ~4w using roughly the same config. So no, it's not the platform thats the issue.
However that isn't the same device we are discussing and it might well be that particular board is okay. I can assure you that the 6 port N100, as well as the 4 port that I've seen reviewed with idle figures posted about, will not idle at those levels, and even 6 watts is a bit on the high side. I've tried booting from USB without SSD and its no better, a bit worse if anything as the USB stick was drawing more power than the SSD at idle.
However , it shows they can do it sometimes and shows the issue isn't due to the N100 or having 2.5G ports, but due to a flaw with some of these types of boards. Its all well and good saying they are cheap and we should accept it, but these companies are making a profit still, they are not a charity doing us a favour.
Nothing to do with research, it doesn't take any research to build a motherboard, you just have the design but there is no research, its all done by reference designs and software already available from Intel.
It could be any number of things, its not for us to need to fathom it out, and if someone is happy with the units even though they are inefficient and clearly not following good design practices and possibly cheapened out on the DC to DC conversion or elsewhere that prevents proper power management, absolutely fine be me. They might burn out in 2 years time or may go on for ever, time will tell.
I'm just raising the issue and would think websites like 'servethehome' reviewing these devices should be a lot more critical on idle power consumption (they test it, but then ignore it), not only does it cost extra to run these devices, electronics running hotter than they need to 24/7 isn't good for reliability as something is burning that power when it shouldn't be.
Chinese companies like CWWK know exactly what they are doing, if they can make a few dollars more profit by skimping on the design they will and do. Quality control along with design is poor not because they aren't charging us enough, its because they want to make more money, and will ship these things abroad where we loose most of our legal protections for faulty goods and most of us aren't going to pay to ship it back when stops working in a few years time, we just bin it. A lot of these boxes were shipping without the CPU in contact with the heat sink, so that is how much they care.
so ran this bios update and now iv lost 2 network ports, also cpu doesnt clock down and stays at full clock making cpus idle power pull measured at 5.4w in hwinfo64.
cant be the wrong bios they given me?
Hi everyone, I have a question if somebody can help me. I would like to flash modded bios to this Topton N100, because I suspect that it's always running at the highest power mode (it's running kind of hot - around 50C under load and that's with replaced thermal paste and a fan on top. Sadly I don't have a outlet power meter to confirm my suspicions currently). Original BIOS has all the power pages removed.
Does anybody have the same model and had success flashing modded bios with missing pages unlocked?
Does anybody have the same model and had success flashing modded bios with missing pages unlocked?