CWWK/Topton/... Nxxx quad NIC router

[ohm]

Ali has some sale running, the unit is about 22$ cheaper now.
I'm thinking about getting a C variant of the N100 unit. Do your think its worth upgrading from a N5105 v4 (with the A case) to a N100? I'm running OPNsense bare metal on it.

Pros I see:

• Probably less heat
• Less power draw (I'm at 10W with the N5105)
• Nicer case (I like the C case)
• Can use same m.2

Cons

• Need to get some DDR5 (I do not have any spares atm)

I would sell the v4, replace my J5005 Gigabyte BRIX or replace the J4125 Qotom at my parents. I'm no way beyond the limits of my N5105, the unit is running strong. I'ts just the "want have" aspect. Should I get one or wait for v2, v3 or whatever? Do you think it's worth it? I sold a lot of old hardware (apu2, old NUC6C etc), now I have some pocket money to spend.

FreeBSD(13.1) on the latest OPNsense doesn't currently work too well on bare-metal Alder Lake CPUs: 261169 – Intel Alder Lake: data corruption with Read&Write files to FAT32 or UFS
Apparently fixed in 13.2, but no idea when this will be picked up by OPNSense. Almost returned my SSD before stumbling upon the root cause. But OPNsense on Proxmox 8 works well, rock stable and surprisingly lower idle temps.

 

[ohm]

I would wait then
Another con for this C variant: it only have USB2.0 ports

The 4 exposed ports are USB2.0, but it appears the internal jumper port is 3.0(based on output from lsusb).

 

[PANiCnz]

What sort of power usage is everyone seeing with a N100 device? The recent STH review suggests its higher than the older N5105 units. I dont think i need the extra performance from the N100 so am leaning towards an older unit.

 

[claymen]

FreeBSD(13.1) on the latest OPNsense doesn't currently work too well on bare-metal Alder Lake CPUs: 261169 – Intel Alder Lake: data corruption with Read&Write files to FAT32 or UFS
Apparently fixed in 13.2, but no idea when this will be picked up by OPNSense. Almost returned my SSD before stumbling upon the root cause. But OPNsense on Proxmox 8 works well, rock stable and surprisingly lower idle temps.

Wasn't that more to do with units that have both P and E cores, devices with a single core type weren't affected?

For Intel Adler Lake P core + E core processor (i7-12700T), copying files to FAT32 partition, the file corrutped (50%), but ZFS is fine. After disabling E core in the code by restrict the max cpu number, this issue is gone. And No E core processor has no such issue, like i7-12400.

 

[ohm]

Fix was apparently applied only to the small (E) cores.

 

[claymen]

View attachment 29710
Fix was apparently applied only to the small (E) cores.

Yeah I had read through that, all the discussion seemed to be in relation to CPU's that had both, I get that it affected the E cores, but wasn't clear if this was specific to the hybrid models which have both (as it does say for cpu's that have small cores, are they a small core if that's the only core it has..). Or is this more to do with assuming both core types are the same? Has anyone tested this on an E core only device? Though at this point it looks to have a fix anyway so shouldn't be an issue. For home use you could also use pfSense+ with a free license, as this is based on FreeBSD14 which has better support for Alderlake anyway.

 

[ohm]

Yeah I had read through that, all the discussion seemed to be in relation to CPU's that had both, I get that it affected the E cores, but wasn't clear if this was specific to the hybrid models which have both (as it does say for cpu's that have small cores, are they a small core if that's the only core it has..). Or is this more to do with assuming both core types are the same? Has anyone tested this on an E core only device? Though at this point it looks to have a fix anyway so shouldn't be an issue. For home use you could also use pfSense+ with a free license, as this is based on FreeBSD14 which has better support for Alderlake anyway.

I spent way too much time figuring out why the filesystem was always getting corrupted with OPNsense on a bare metal N100, even during installation. So yes, pretty sure this bug occurs on an E core only device.

 

[Mike9474593]

I would wait then
Another con for this C variant: it only have USB2.0 ports

That's a valid point. Guess I will wait then. Maybe changwang is releasing a v2, maybe the 5 port version has faster USB (USB-C).

FreeBSD(13.1) on the latest OPNsense doesn't currently work too well on bare-metal Alder Lake CPUs: 261169 – Intel Alder Lake: data corruption with Read&Write files to FAT32 or UFS
Apparently fixed in 13.2, but no idea when this will be picked up by OPNSense. Almost returned my SSD before stumbling upon the root cause. But OPNsense on Proxmox 8 works well, rock stable and surprisingly lower idle temps.

Yeah, you are right. I did read about this but did not consider. OPNsense 23.7 should be based on 13.2 (source), but this is still a major drawback. Guess I will wait until 23.7 is released and then consider again. Maybe there is a V2 in two months or something different to spend my money

 

[ohm]

That's a valid point. Guess I will wait then. Maybe changwang is releasing a v2, maybe the 5 port version has faster USB (USB-C).

These are the V1 variants I suppose are/will be available limited by the number of PCIe lanes, choose your poison:

 

[flips33]

The 4 exposed ports are USB2.0

I'm seeing the same thing, a USB 3.0 bus that isn't used for front or rear ports:

lsusb -t
/: Bus 02.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/4p, 10000M
/: Bus 01.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/12p, 480M
|__ Port 1: Dev 2, If 0, Class=Human Interface Device, Driver=usbhid, 1.5M
|__ Port 2: Dev 6, If 0, Class=Mass Storage, Driver=usb-storage, 480M
|__ Port 3: Dev 3, If 0, Class=Mass Storage, Driver=usb-storage, 480M
|__ Port 4: Dev 4, If 0, Class=Mass Storage, Driver=usb-storage, 480M
|__ Port 7: Dev 5, If 0, Class=Mass Storage, Driver=usb-storage, 480M

but it appears the internal jumper port is 3.0(based on output from lsusb).

That would make sense, but aren't these 9-pin headers USB 2.0 only?

 

[ohm]

I'm seeing the same thing, a USB 3.0 bus that isn't used for front or rear ports:

lsusb -t
/: Bus 02.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/4p, 10000M
/: Bus 01.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/12p, 480M
|__ Port 1: Dev 2, If 0, Class=Human Interface Device, Driver=usbhid, 1.5M
|__ Port 2: Dev 6, If 0, Class=Mass Storage, Driver=usb-storage, 480M
|__ Port 3: Dev 3, If 0, Class=Mass Storage, Driver=usb-storage, 480M
|__ Port 4: Dev 4, If 0, Class=Mass Storage, Driver=usb-storage, 480M
|__ Port 7: Dev 5, If 0, Class=Mass Storage, Driver=usb-storage, 480M

That would make sense, but aren't these 9-pin headers USB 2.0 only?

Not an expert but Wikipedia says USB 3.0 has 9 wires:

 

[flips33]

Not an expert but Wikipedia says USB 3.0 has 9 wires:

It's a bit confusing but it's talking about USB 3.0 being 9 wires per USB port (header = two ports)

USB 2.0 is four wires per port (+ ID pin, so 2 x USB 2.0 ports = 9 pin header).
USB 3.0 is backward compatible + five additional wires per port (usually presented in a 10x2 header = 20 pin header)



USB 3.0 header cables look like this:

 

[abdullah]

My i3-N305 unit is due to arrive any day now...

Anyone have any idea what it will take to get coreboot working on these units?

Here's some research I've done for i3-N305:

I can see the Intel FSP is available for Alderlake N-Series processors for IoT platforms:

FSP/AlderLakeFspBinPkg/IoT/AlderLakeN at master · intel/FSP

Intel(R) Firmware Support Package (FSP). Contribute to intel/FSP development by creating an account on GitHub.

github.com

Coreboot intel-microcode repo, microcode-20230512-rev2 shows support for i3-N305:

intel-microcode/releasenote.md at main · coreboot/intel-microcode

Contribute to coreboot/intel-microcode development by creating an account on GitHub.

github.com

However Coreboot does not have anything for CWWK mainboards:

coreboot/src/mainboard at master · coreboot/coreboot

Mirror of https://review.coreboot.org/coreboot.git. We don't handle Pull Requests. - coreboot/coreboot

github.com

It may be possible to build using one of the other mainboards listed...

I did contact CWWK for the BIOS files, and they sent me a link - I asked if this will work for i3-N305 because the file names say N100, and chat confirmed it would. I don't know how long this link will work for so I have of course saved it elsewhere, I can't upload it here because it is 7.7mb and the forum throws an error saying it is too large for the site to process.

http://www.7immy.com:5244/d/%E4%B8%BB%E6%9D%BFBIOS/%E8%8B%B1%E7%89%B9%E5%B0%94%E7%B3%BB%E5%88%97BIOS/CW-12%E4%BB%A3%E4%B8%BB%E6%9C%BA-%E8%BD%AC%E6%8E%A5%E6%9D%BF%E6%AC%BE-N95-N100-N200-N305/CW-N100.rar?sign=xqCktRzniFvuJFAIVtnk9q19MSXOzeDDH9_MQ3iISyI=:0

I may try to extract files from the BIOS image using coreboot's bios extract tooling, GitHub - coreboot/bios_extract: collection of scripts to extract files from BIOS images

And then it may be a case of following the hands-on example of building coreboot from this brilliant paper by Jiming Sun et al,:

Building coreboot with Intel FSP

Since 1999, developers from around the world, some as individual contributors and others working on behalf of businesses and corporations, have formed a community around coreboot, an open source firmware project. coreboot is boot firmware primarily focused on x86...

link.springer.com

Although I may be being very naive here... I have zero experience with something this complex.

 

[fta]

I would only do this if I had a flash programmer on hand. You're going to brick it a some point and need to recover.

 

[RevAngel]

My bad shopping experience and a intentionally dragged out "no help, go fys" and answers straight out of a bullshit-lottery lead me to the following recommendation: do not buy these products, do not use aliexpress, avoid return costs to Chinese warehouses by not buying Chinese products. Do not expect help from Aliexpress or for bought Chinese products. In the end I just pay more and spend a lot of time for nothing at all.

 

[shanestorey]

I have one of the Topton N100, C variant running Proxmox quite nicely.

I'd like to get a second disk installed, but I can't get the second SSD running, it can be seen in the BIOS, but Proxmox doesn't see it.

Has anyone managed to get an SSD installed in the adapter M2 slot?

Seems that an SSD installed in the second slot also runs ~5 degrees hotter than in the primary slot.

 

[bhigh]

I'm playing catch up on this thread, so apologies in advance if I missed something.

Cases: Variation C seems like it would have the best heat dissipation, all else being equal. It looks like @Becks0815 has tested this.

N95 vs. N97 vs. N100 vs. N200:
- N95: 3.4 GHz boost, 16 EU at 1.2 GHz 15W TDP
- N97: 3.6 GHz boost, 24 EU at 1.2 GHz 12W TDP
- N100: 3.4 GHz, 24 EU at 750 MHz 6W TDP
- N200: 3.7 GHz boost, 32 EU at 750 MHz, 6W TDP

The higher graphics clock might account for some of the higher TDP on the N95. However the N97 has a higher boost clock and 24 EU at 1.2 GHz with a 12W TDP. I suspect the N95 comes from the worst yields, resulting in fewer functional EUs and more power usage. Probably best to avoid the N95 as it only saves $10 over the N100.

N100 seems to be a sweet spot out of the 4 core options.

For higher performance, the Pentium Gold 8505 and i3-305 seem to trade blows. The 8505 is about $20 cheaper from Topton, KingNovy, or CWWK.

The 8505 has 1P+4E cores/6 threads. The P-core should give it better single thread performance. Its E-cores are clocked at 3.3GHz so I'd expect it to be a little slower for CPU intensive multithreaded workloads. It also supports more memory (2 channels), has more cache (8MB), more PCIe lanes (20) and PCIe 4.0.

The i3-305 has 8 cores/8 threads clocked at 3.8 GHz, supports 1 memory channel, has 6MB cache, and 9 PCIe 3.0 lanes.

Max memory: Only 1 memory channel is supported. ARK says 16 GB max, but the KingNovy and Topton reps have both said it works with 32 GB. @TraXter tested with 32GB and it seems to have worked.

OPNsense and pfsense might be affected by a problem with FreeBSD 13.1 that that causes data corruption with FAT32 and UFS. ZFS doesn't appear to have a problem. 261169 – Intel Alder Lake: data corruption with Read&Write files to FAT32 or UFS

 

[bhigh]

The box came with a Daijing power supply, 12V/60W

I wasn't able to read the label. What size barrel plug does it have? Or what make and model did you replace it with?

Setup of 23.1 failed, I got constant errors about msdos_fs not accessible after the partition creation. According to google a common "feature" of FreeBSD13. So I started with 22.7 and went along the full upgrade road to 23.1_5

This sounds like the FreeBSD bug that was brought up later in the thread. Did you try installing with a ZFS root or only UFS?

Disabling PCID support may help too. This is probably why Proxmox is fine. Linux To Disable PCID For Intel Alder Lake & Raptor Lake Due To Issue With INVLPG - Phoronix

 

[ohm]

Has anyone managed to get an SSD installed in the adapter M2 slot?

It worked when I tried switching slots after getting the filesystem errors in OPNsense. Didn't try having both slots populated though.

 

[flips33]

Has anyone managed to get an SSD installed in the adapter M2 slot?

I have both m.2 slots populated with nvme. I'm running Linux now, but I did have Proxmox 7.4 installed using both nvme (zfs mirror)