CWWK/Topton/... Nxxx quad NIC router

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

thimplicity

Member
Jan 11, 2022
64
10
8
It seems like my initial Topton box (N5105) is dying on me - which one is currently the recommended model here? I use it as a pfSense router with pihole and some docker containers running on it (all on Proxmox). Is the N100 sufficient? Looks like there are boxes supporting two NVMes now so that I can run a mirrow
 

slybunda

Active Member
Jan 30, 2023
132
71
28
Do basics first. Memtest and run CPU benches too like prime.
See how it goes from there.
Eliminate hardware first then it be software issue
 
  • Like
Reactions: audit13

slybunda

Active Member
Jan 30, 2023
132
71
28
Exactly why I still have my edgerouter 4 screwed to the wall as a backup for when I'm tinkering with the mini pc box.
 

Dunadan

New Member
May 11, 2021
10
4
3
Hi all
My version of the N100 topton router:


On stock BIOS and no fan the temps were in idle around 50C but the NVME was going to even 75-80C.
It has an aluminium block on top of the CPU and the case was very hot. If you put a cup of coffee on it it would be worm all the time ;)

I have done a little modding of the case by drilling around 20, 3mm holes in the bottom cover of it and placed 5v USB FAN (https://www.amazon.pl/dp/B00G05A2MU?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1) below the router blowing up into the case. Additionaly i have installed 120x120mm mesh filter into the case.

The USB fan is working with the slowest speed and the temps are: CPU - 38C and NVME is almoust 30C when idling and 50C when in heavy use.

i have unlocked the BIOS the advance power settings but do you guys have any tutorial or any other data of what to change in BIOS to lower the temps or make the router more power efficient?
 
  • Like
Reactions: KevinR

slybunda

Active Member
Jan 30, 2023
132
71
28
Hi all
My version of the N100 topton router:


On stock BIOS and no fan the temps were in idle around 50C but the NVME was going to even 75-80C.
It has an aluminium block on top of the CPU and the case was very hot. If you put a cup of coffee on it it would be worm all the time ;)

I have done a little modding of the case by drilling around 20, 3mm holes in the bottom cover of it and placed 5v USB FAN (https://www.amazon.pl/dp/B00G05A2MU?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1) below the router blowing up into the case. Additionaly i have installed 120x120mm mesh filter into the case.

The USB fan is working with the slowest speed and the temps are: CPU - 38C and NVME is almoust 30C when idling and 50C when in heavy use.

i have unlocked the BIOS the advance power settings but do you guys have any tutorial or any other data of what to change in BIOS to lower the temps or make the router more power efficient?
That model you got needs a fan on top of it
 

audit13

Member
Jun 26, 2024
38
9
8
Hi all
My version of the N100 topton router:


On stock BIOS and no fan the temps were in idle around 50C but the NVME was going to even 75-80C.
It has an aluminium block on top of the CPU and the case was very hot. If you put a cup of coffee on it it would be worm all the time ;)

I have done a little modding of the case by drilling around 20, 3mm holes in the bottom cover of it and placed 5v USB FAN (https://www.amazon.pl/dp/B00G05A2MU?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1) below the router blowing up into the case. Additionaly i have installed 120x120mm mesh filter into the case.

The USB fan is working with the slowest speed and the temps are: CPU - 38C and NVME is almoust 30C when idling and 50C when in heavy use.

i have unlocked the BIOS the advance power settings but do you guys have any tutorial or any other data of what to change in BIOS to lower the temps or make the router more power efficient?
I have the same model running opnsense on bare metal.

How did you unlock the bios?
 

Dunadan

New Member
May 11, 2021
10
4
3
Problem is with a fan at the bottom it's not gonna pull in much air unless it's a loud fan. There is just not enough clearance from the table it sits on. Also there is no vents for the air to go out from other than. The bottom so that's just going to recirculate hot air.
Thats why the fan should push the air into the case instead of pullig.
 

audit13

Member
Jun 26, 2024
38
9
8
I followed this advice:

but remember to update the 1.nsh file with -bios.

it should look like this FPT -BIOS -F ....bin

The moded bios is from this file bios15_edited.bin bios15_edited.bin
Thanks for the information.

Unfortunately, I made a mistake and my Topton is not the same unit although the case looks similar. The motherboard inside my unit does not have a slot for the battery.
 
  • Wow
Reactions: Dunadan

phil-2024

New Member
Sep 7, 2024
27
21
3
Just a quick update as I've since received a mini PC with an Intel N97 (this clocks a bit faster than the N100). Connected to Wi-Fi but no keyboard and monitor, (has a fan running), in Windows at idle it is drawing just 4 watts. This is some 7 watts less than some of these boxes here, and that is without having to enter the BIOS and play about with any settings. I stand by my comments that some of these network appliances drawing 11-12 watts at idle are seriously flawed.
 
  • Like
Reactions: audit13

phil-2024

New Member
Sep 7, 2024
27
21
3
4 watts at the AC socket, so actual power usage. The one I have (which is going to be used for media playback) is an NiPoGi N97 with 8GB DDR4 and 256GB SSD. It does have room to add a 2.5" drive.

The N97 clocks a bit faster than the N100 and consequently uses a bit more maximum power (if configured to do so) than the N100, I suspect the N97 is just a better binned N100 as otherwise they are identical in specification.

Out of curiosity I am going to put pfSense on it to see what the idle power consumption is.

When powered off or in sleep, this box is drawing just half a watt, compared to 2.5 watts of these network appliances, not that network appliances spend anytime powered off typically, but its an indication of how poorly designed they are.
 
  • Like
Reactions: Stovar and audit13

matt_garman

Active Member
Feb 7, 2011
228
63
28
Just a quick update as I've since received a mini PC with an Intel N97 (this clocks a bit faster than the N100). Connected to Wi-Fi but no keyboard and monitor, (has a fan running), in Windows at idle it is drawing just 4 watts. This is some 7 watts less than some of these boxes here, and that is without having to enter the BIOS and play about with any settings. I stand by my comments that some of these network appliances drawing 11-12 watts at idle are seriously flawed.
Can you go in this system's BIOS and see what can be tweaked and what the default settings are? Also, how many network ports does it have?

I don't recall anyone hitting four watts with the CWWK-style network appliance n100 boxes, but people are getting below 10 with BIOS tuning. So it would be a better "apples to apples" comparison if we check BIOS settings - it's possible your system ships with some of the power tuning tweaks discussed here already applied.

The one I have is an NiPoGi N97 with 8GB DDR4 and 256GB SSD.
Emphasis mine. Is that a single 8GB DIMM or two 4GB DIMMs? I believe all the CWWK-style network appliance boxes are DDR5, so that's a difference. Also another reason to take a look at your BIOS settings, to see what voltage and speed your RAM is running. I can't recall if RAM voltage/speed tweaking has been attempted on the CWWK style boxes to reduce power draw. But DDR5 is generally faster than DDR4, so I would expect the performance ceiling of your system to be lower than a similar DDR5-based system. At low-to-idle loads it likely doesn't matter though.
 

phil-2024

New Member
Sep 7, 2024
27
21
3
Can you go in this system's BIOS and see what can be tweaked and what the default settings are? Also, how many network ports does it have?
The BIOS settings are as good as identical as I've seen on these network appliances, still I see no ASPM options for power settings. The CPU Power settings default to identical settings as the CWWK style BIOS. You'd be hard pressed to tell them apart. Also the package power when shown in Windows on the CWWK devices will report a couple of watts at idle and CPU frequency will drop to about 400MHz, so I'm pretty sure the CPU is entering low power modes, but there is some hardware issue that means something is still burning up a lot of power.

I don't recall anyone hitting four watts with the CWWK-style network appliance n100 boxes, but people are getting below 10 with BIOS tuning. So it would be a better "apples to apples" comparison if we check BIOS settings - it's possible your system ships with some of the power tuning tweaks discussed here already applied.
I've seen about about 9.5 watt on my CWWK appliance but typically its 10.5 to 11 watt. BIOS power tuning as far as I can see really only relates to maximum power usage, i.e. making sure it can't overheat or crash due to trying to use more power than can be supplied, and by default, the BIOS is set to allow the CPU to use as little as possible with light or no loads (Max C states and Speedstep enabled etc).

For context, my existing network appliance from several years ago, using a much older less efficient CPU (i7 7500U), still with 6 network ports and 3 connected and live, idles at 5 watt, and I didn't need to tweak anything to get it that low.

Emphasis mine. Is that a single 8GB DIMM or two 4GB DIMMs? I believe all the CWWK-style network appliance boxes are DDR5, so that's a difference. Also another reason to take a look at your BIOS settings, to see what voltage and speed your RAM is running. I can't recall if RAM voltage/speed tweaking has been attempted on the CWWK style boxes to reduce power draw. But DDR5 is generally faster than DDR4, so I would expect the performance ceiling of your system to be lower than a similar DDR5-based system. At low-to-idle loads it likely doesn't matter though.
This mini-PC has 8GB DDR4 as a single DIMM. DDR5 should use less power, around 20% less power is quoted compared to DDR4, so making these CWWK appliances looking even worse in comparison to my older network appliance or other mini-PCs using DDR4 but the same Intel N100 chips.

So to sum up, something is seriously wrong with these CWWK appliances, they are drawing far too much power at idle, just what is burning that extra 6-7 watts? Have they upped the voltages somewhere due to instability from a design flaw or they are using a low binned batch of Intel CPUs that they've got for cheap that crash in lower power states, so this batch has ended up in cheap mains powered network appliances as they are no good in laptops? This is China we are talking about, you can't rule anything out and no one puts a brand name on these boxes :) Whilst the extra in electricity costs isn't going to break most peoples bank accounts, and many are happy sitting fans on these boxes, what about the long term reliability? I guess time will tell with that one.
 

matt_garman

Active Member
Feb 7, 2011
228
63
28
I've seen about about 9.5 watt on my CWWK appliance but typically its 10.5 to 11 watt. BIOS power tuning as far as I can see really only relates to maximum power usage, i.e. making sure it can't overheat or crash due to trying to use more power than can be supplied, and by default, the BIOS is set to allow the CPU to use as little as possible with light or no loads (Max C states and Speedstep enabled etc).
FWIW, I've got my CWWK n100 4x2.5gbps port appliance idling at 7.5-8.0 watts. That's with 1 network port attached, keyboard, and HDMI video. If I remove the keyboard and HDMI (i.e. only power and network connected), it saves another watt, i.e. 6.5-7.0 watts. This is with Proxmox v8.2.7, essentially a vanilla install (tweaked with powersave governor and powertop --auto-tune). I did do a number of BIOS tweaks discussed in this thread. I'm measuring AC power draw at the wall via Kill-a-Watt.

However, I don't have another similar device with which to compare (unless you count a CWWK n5105 device, with which I got essentially the same results). And I think at this low of a power draw, our numbers aren't necessarily comparable without some kind of common control device, or precision calibrated instrumentation.