CWWK/Topton/... Nxxx quad NIC router

[JonasRokas]

Load of questions coming in, sorry:

What temps are you getting exactly?
Have you tried the memtest without the NVMEs in the units?
What were/are you idle temps while in an OS (without a fan)?

If you see in my previous comment, my unit can't fully pass memtest86 because of the reboots, but the stress tests ran fine. Even stressapptest for an hour passed. If your idle temps are fine, maybe try the same test I did above while monitoring the temps and share your results.

I am getting over 85*c and up to 100*c, tests start failing above 90*c. The temps creep up slowly now after bios update and eventually fail. The 3 units are behaving differently, some fail quicker but eventually everything fails. I have tried all kinds of test from various versions of memtest, stability stress tests, hard drive, cpu, etc but everything fails.
40*c during idle even then I get BSOD in windows. Tried other operating systems such as proxmox and it’s more stable but VMs crash or freeze. The units are highly unreliable and flawed in design. For production environments these do not work, I wouldn’t even want them for home use.

Stability stress from AIDA64 says overheating right when starting the test. Temps are over 88*c. I will try with a fan and see if it helps but I am at my last straw with these 3 units. Topton has been very disappointing in customer service, the first 10 days just automated responses and after dispute with AliExpress they start responding but I have already tried everything short of installing an external fan…thermal paste redone, various bios updates, different stress tests, internal fans from chinese and Noctura, 20 sticks of ram from crucial, team, and gskill.

 

[idle_user]

I am getting over 85*c and up to 100*c, tests start failing above 90*c. The temps creep up slowly now after bios update and eventually fail.
...

Oh, wow. My condolences.
All of this reads like there's weak contact between the chip and the heatsink/copper plate if temps are high and rising during idle.

Either way, it's a definite flaw like you said.

 

[JonasRokas]

Even during medium temp the tests fail. Before the bios update, the tests and stability would fail at low temps.

 

[fta]

Unfortunately, looks like I misplaced my watt meter to test the power supply. I'll be purchasing one later and giving that a test.

To test the power supply, run prime95 with small fft on all cores and threads. If that doesn't cause the reboot, it's unlikely to be the power supply.

 

[idle_user]

To test the power supply, run prime95 with small fft on all cores and threads. If that doesn't cause the reboot, it's unlikely to be the power supply.

Well running small fft went fine and temps looked fine too.
But I tried large fft (for RAM and memory controller) with default settings and it immediately rebooted.

I tried again with:
2 cores only - it ran fine.
4 cores only - it ran fine.
6 cores only - rebooted
5 cores only - it ran fine.

Not sure what this means.

 

[hot22shot]

While I'm waiting for my n305 I'm considering buying some shims just to be sure, what thickness would you advise ?

 

[ohm]

6 cores only - rebooted
5 cores only - it ran fine.

Did u mean to say 6 cores ran ok and 5 cores didn't? Otherwise, based on all your tests so far, it seems instability occurs at load beyond a certain threshold, and may not be solely due to memory access issues. Maybe try to lower the package PL1 and PL2 to lower values between 10000(10W) and 15000W(15W) to reduce the power consumption and heat dissipation. Another thing to try is to progressively disable some cpu cores to lower the overall power consumption while maintaining higher clock speeds, and possibly to rule out defects(not very likely) with certain cores.

 

[ohm]

yeah, I did the paste again on one of them to test it and the same result but few degrees lower temp now. It looks like Topton did a decent job on the paste unlike their prior units.
I don’t know how else to get this to run cooler, I have already ran a fan inside and it’s helped to pass one memtest but temps are still high. There’s 2 nvme drives in there.

Simply repasting(removing and reapplying the thermal paste) is probably not enough. You've got to first remove the paste and then check for the presence of any gap between the cpu and heatsink. Closing any gap would involve filling it with a thermal pad or shim of suitable thickness, or shaving the motherboard standoffs(not advisable if you plan to pursue a refund/exchange with the seller). You could adjust the package PL1 and PL2 values to lower the power consumption and thus the cpu/platform temperature on load.

 

[fta]

Well running small fft went fine and temps looked fine too.
But I tried large fft (for RAM and memory controller) with default settings and it immediately rebooted.

I tried again with:
2 cores only - it ran fine.
4 cores only - it ran fine.
6 cores only - rebooted
5 cores only - it ran fine.

Not sure what this means.

This confirms the power supply is able to deliver the maximum power required. The small fft test will use considerably more power than the large fft test. Perhaps the PS is having issues at lower power levels, but to me this is further indication there is something wrong with the memory hardware on your box. The only way to truly rule out the PS, though, is to replace it with a quality one.

 

[lucker]

But I tried large fft (for RAM and memory controller) with default settings and it immediately rebooted.

That I would take as a sign of hardware problem, meaning something is really not right with the memory management and, since you replaced the RAM module, more probable it's MB. If you re-flush BIOS and reset CMOS, then that would be it (what esle it would be?) The experienced engineers in the thread will correct me, of course, if I'm wrong.

Once I tried (unsuccessfully) to overclock ancient AMD on fairly cheap Asrock MB and whatever I did caused errors in prime95 or sometimes even the test failure, BSOD, freezes. Eventually I have dropped the idea, but the advised prime95 as a harware coherence test stick in my mind deep.

 

[idle_user]

Did u mean to say 6 cores ran ok and 5 cores didn't? Otherwise, based on all your tests so far, it seems instability occurs at load beyond a certain threshold, and may not be solely due to memory access issues. Maybe try to lower the package PL1 and PL2 to lower values between 10000(10W) and 15000W(15W) to reduce the power consumption and heat dissipation. Another thing to try is to progressively disable some cpu cores to lower the overall power consumption while maintaining higher clock speeds, and possibly to rule out defects(not very likely) with certain cores.

Nope, what I put was correct. I was going up by two cores each attempt. It failed at 6 core usage, so I bumped it down to 5 and it ran without reboot.

I have adjusted PL1 and PL2 to those values and still run into the same issues.
I don't suspect the cpu cores are the issue since it passed Prime95 Small FFT. Since it failed at Large FFT, I'm suspecting something is wrong with memory controller or maybe some incompatibility with the firmware.

These reboots have only happened on Prime95 Large FFT(with 6+ core usage) and memtest86 Test 8.
Everything else runs fine.

I even had it run stressapptest for 5 hours overnight and no issues. I'm hoping these reboots only happen during the intense memory tests which I don't think I'd reach to those levels during normal usage. (fingers crossed)

This confirms the power supply is able to deliver the maximum power required. The small fft test will use considerably more power than the large fft test. Perhaps the PS is having issues at lower power levels, but to me this is further indication there is something wrong with the memory hardware on your box. The only way to truly rule out the PS, though, is to replace it with a quality one.

Thanks for your support and insight. Are there recommended ones I should get? I'm not familiar at all with these power bricks. I posted an image of the one they supplied with it, so I'm not sure about its quality.

 

[ohm]

Nope, what I put was correct. I was going up by two cores each attempt. It failed at 6 core usage, so I bumped it down to 5 and it ran without reboot.

I have adjusted PL1 and PL2 to those values and still run into the same issues.
I don't suspect the cpu cores are the issue since it passed Prime95 Small FFT. Since it failed at Large FFT, I'm suspecting something is wrong with memory controller or maybe some incompatibility with the firmware.

These reboots have only happened on Prime95 Large FFT(with 6+ core usage) and memtest86 Test 8.
Everything else runs fine.

I even had it run stressapptest for 5 hours overnight and no issues. I'm hoping these reboots only happen during the intense memory tests which I don't think I'd reach to those levels during normal usage. (fingers crossed)



Thanks for your support and insight. Are there recommended ones I should get? I'm not familiar at all with these power bricks. I posted an image of the one they supplied with it, so I'm not sure about its quality.

If the issue can be reproduced even after capping the PL values which effectively runs the unit with lower power input, it should be safe to rule out power supply instability at least under high loads. I agree Prime95 pass on small fft and failure on large fft is a strong indicator of memory access issues. It could be worthwhile to see if loosening some memory access settings(speed, voltage etc) resolves the test failures - that would be strong evidence to strengthen your case towards returning the unit, or at the minimum have the option of retaining a stable albeit less optimized rig.

 

[idle_user]

If the issue can be reproduced even after capping the PL values which effectively runs the unit with lower power input, it should be safe to rule out power supply instability at least under high loads. I agree Prime95 pass on small fft and failure on large fft is a strong indicator of memory access issues. It could be worthwhile to see if loosening some memory access settings(speed, voltage etc) resolves the test failures - that would be strong evidence to strengthen your case towards returning the unit, or at the minimum have the option of retaining a stable albeit less optimized rig.

Am I able to adjust the memory timings in that BIOS? I'm not familiar with this one, but I couldn't find anywhere to up the voltage on the RAM a bit or turn down the timings.

At the moment, I'm still in the dispute with the seller on AliExpress. They mentioned again to return the unit to them. I said to them I will, but only if they have a pre-paid shipping label and pay any customs fees. But the conversation's language somehow turned into them expecting me to pay for the return. They also gave me some bizarre instructions to lie about what's in the box and the cost "for customs" purposes.
I tried to be reasonable with them at first by only requesting a ~50% partial refund when I began the dispute. Looks like AE is probably gonna step in, and lets see what happens. If all fails, I'm issuing a chargeback or go through paypal dispute (don't remember which method I used) - this will probably block my payment method or ban my AE account I'm willing to bet.

 

[ohm]

Am I able to adjust the memory timings in that BIOS? I'm not familiar with this one, but I couldn't find anywhere to up the voltage on the RAM a bit or turn down the timings.

At the moment, I'm still in the dispute with the seller on AliExpress. They mentioned again to return the unit to them. I said to them I will, but only if they have a pre-paid shipping label and pay any customs fees. But the conversation's language somehow turned into them expecting me to pay for the return. They also gave me some bizarre instructions to lie about what's in the box and the cost "for customs" purposes.
I tried to be reasonable with them at first by only requesting a ~50% partial refund when I began the dispute. Looks like AE is probably gonna step in, and lets see what happens. If all fails, I'm issuing a chargeback or go through paypal dispute (don't remember which method I used) - this will probably block my payment method or ban my AE account I'm willing to bet.

My unit is being used as the main router, so little chance of shutting it down to check out the bios settings - maybe someone else can help with this.

Don't settle for only 50% refund - argue that it is essentially deadweight if the system isn't even stable and you have no use for it other than the initial purpose you bought it for, eg as your main router. There is an off chance you may still be able to use it for less-critical work, but don't count on it.

 

[idle_user]

Don't settle for only 50% refund - argue that it is essentially deadweight if the system isn't even stable and you have no use for it other than the initial purpose you bought it for, eg as your main router. There is an off chance you may still be able to use it for less-critical work, but don't count on it.

That's the funny part. That's what I was trying to settle for in the dispute, but they are saying that's still too much.
I was really trying to be nice, but now the way they're handling it, I'm gonna do a full refund or die trying.

 

[andrea87]

Where are you from?

I would go through the dispute, claim the device as dead on arrival as it clearly doesn't boot. Offer a return, at the seller's expense. If it doesn't work with the dispute first, escalate through their chat system (write something like chat with tech support person) and provide all the info and screenshots you can. Keep the paypal dispute and card chargeback as last options, but inform AE as well if you go through their chat that if the issue doesn't get resolved by them (you essentially have a 2-300$ brick on your desk) you'll go through other ways.

I had to be very aggressive with my mini PC when it broke end of April, shipped the board back early May... the seller didn't give me any useful info for a month and I finally managed to get a full refund through AE after chatting with them twice.

Have a look also in this thread.

 

[vamega]

I'm new here, so sorry if this is the wrong place to post my build.

I picked up the Toptop n305 and added M.2 8 Ports SATA - 2xMini SAS card. Jumpered a sfx psu to power the drives, and 32GB DDR5 5600.

Plan to power the n305 with the psu later on, and get a second mini SAS card.

Were you able to figure out how to power the device via the PSU?

 

[abdullah]

Did you try that N100 bios on a N305 and/or did it work? CWWK hasn't posted anything new yet on their Chinese site

I haven't powered mine on yet, been busy with a few things.

Did anyone flush the BIOS from the link abdullah has mentioned? I would like to try the 4xNVMe board that came with ordered box, but that involves flushing special version of BIOS. In case the PCB doesn't work or doesn't meet the expectations I would very much prefer to roll back to the basics.

Now my box has BIOS 2.22.1287 (arrived with it). What is the version by the mentioned link?

I won't get to try it for another 2 weeks.. once I do, I'll let you know.

 

[sos_nz]

My N100 fanless minipc arrived from AliExpress last week. I put in a stick of 16GB DDR5 4800 (Crucial), and a Crucial P3 nvme drive. Booted up and installed bare metal OPNsense (23.7 RC1, based on BSD 13.2).

Other than needing to temporarily adjust back to a repo of the current stable release (23.1), it's been easy and fully functioning.

Idle temps ~44C. I'm v. happy.

 

[BenMc]

CWWK has a five port version of their new N100 CPU product line available on AliExpress. Has anyone looked at it yet?

It looks very promising to me because:

• Five I226-v ports instead of 4 (duh)
• Two M.2 slots w/o the need for adapter card. (Note, these are only x1 slots each)
• Two SATA 3.0 ports
• USB-C port
• Two HDMI 2.1 plus DP 1.something

This, to me, has a lot more flexibility than than 4 port models and its only sale for about $220 as of today.

Anyone know why I shouldn't purchase this?

EDIT: The HUNSN RJ39 appears to be identical to the CWWK product.

EDIT: Pulled the trigger... I bought the bare bones version and I'm adding 16 GB of Crucial DDR5 5600 RAM together with two 500 GB Western Digital SN570 SSDs.

I plan to make this my OpnSense router running as a VM on Proxmox with PCI passthru for the NICs. The two SSDs will be configured in a ZFS mirror (RAID-1).

But, for initial testing, I'll use OpnSense on Proxmox as a layer 2 switch (i.e. as a software bridge) to replace a dumb ethernet switch connecting my two PCs to my upstream router. I want to check stability before using this to replace my Asus router.

I'll try to remember to report my findings once this arrives and I've had time to futz with it.