CWWK/Topton/... Nxxx quad NIC router

[Becks0815]

Kingnovy (or is it just another Topton spinoff?) now also offers the N95, N100 and N200, but the only differences on the web page I have spotted was a different picture for the case and copper element. Case A this time, without a copper heat pipe X. They also use an older picture for the NVME drive locations and show 2 PCIe adapters

 

[BoomBangCrash]

Just thought I'd mention an alternative: last week I ordered a similar box with another Alder Lake processor, the pentium gold 8505, currently available more cheaply than the N95/N100/N200/N305 boxes, but with specs that seem higher in some respects (cache size, total threads and max turbe freq). It does have a much higher wattage though so I guess the catch will be to see whether it overheats. The box I ordered has an active fan. It's coming by slow boat so I guess I won't know for some time, but thought I'd mention it here in case anyone else is interested.

Most interesting for me would be a wattage capped 8505 vs N100.

 

[Becks0815]

Game on - the box is here. RAM was accepted, just like the NVME drive. I only had some initial issues with no graphics being displayed, but the problem was I tried to use HDMI while the only output I was able to set at the moment is Display port.

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

The case is impressive heavy, but the connection between CPU and case is a single copper block only, without heat pipes. There is plenty of this grey paste applied on all kind of places. I removed them and replaced it with Arctic MX4.

I haven't spotted a gap between the copper block and the CPU, which is fine.

I have installed Ubuntu 22.04 and upgraded the kernel to 6.2.11, and powertop shows C10 as lowest possible power state

The initial power consumption in BIOS was 11.3W-12W using default Bios settings, 16GB RAM, no NVME drive installed, DP monitor attached plus keyboard in an USB port plugged in, measured using the Daijing power supply and at the wall (Voltcraft 3000)

Running ubuntu kernel 6.2.11 idle with powertop savings applied (except PM on the Intel NICs, which freezes my SSH connection), SATA, sound disabled, nothing plugged in except 1 ethernet cable and PL1 and PL2 set to 10W I am down to 8.7W, measured at the wall with the daijing brick. Not too bad for the first try, but I am sure there is plenty to play around in the Bios. I have found a sh*tton of settings I have never seen in any Bios, and the first ones I looked at were the ones for 386dx machines.

idle temp Ubuntu: 33°C CPU temp after sitting idle on the desktop here for 20 mins.

 

[Becks0815]

Forgot to mention:

There is an adapter installed to attach a second NVME drive. I have used the first one (without adapter)

USB - lsusb shows two hubs, one USB 2, the other USB 3 - I haven't tested which one is which

Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
Bus 001 Device 003: ID 05e3:0751 Genesys Logic, Inc. microSD Card Reader
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

Passmark test (using PL1/PL2 10W):

Intel N100 (x86_64)
4 cores @ 3400 MHz  |  15.4 GiB RAM
Number of Processes: 4  |  Test Iterations: 1  |  Test Duration: Medium
--------------------------------------------------------------------------
CPU Mark:                          4610
  Integer Math                     17518 Million Operations/s
  Floating Point Math              10816 Million Operations/s
  Prime Numbers                    23.7 Million Primes/s
  Sorting                          6007 Thousand Strings/s
  Encryption                       3354 MB/s
  Compression                      43163 KB/s
  CPU Single Threaded              1981 Million Operations/s
  Physics                          393 Frames/s
  Extended Instructions (SSE)      1933 Million Matrices/s

Memory Mark:                       2342
  Database Operations              2163 Thousand Operations/s
  Memory Read Cached               17170 MB/s
  Memory Read Uncached             11567 MB/s
  Memory Write                     8711 MB/s
  Available RAM                    15187 Megabytes
  Memory Latency                   33 Nanoseconds
  Memory Threaded                  25633 MB/s

So what tests are interesting now?

 

[Becks0815]

I have found two places for power limits. Anyone here who can explain what they trigger?

 

[Becks0815]

Think I got it.

The platform PL settings steer the CPU power, the other the "total package" (no idea so far what this is). By setting PL1 to 12500 and PL2 to 25000 I was getting some better results in passmark:

Intel N100 (x86_64)
4 cores @ 3400 MHz  |  15.4 GiB RAM
Number of Processes: 4  |  Test Iterations: 1  |  Test Duration: Medium
--------------------------------------------------------------------------
CPU Mark:                          6199
  Integer Math                     17516 Million Operations/s
  Floating Point Math              10818 Million Operations/s
  Prime Numbers                    24.2 Million Primes/s
  Sorting                          9835 Thousand Strings/s
  Encryption                       4741 MB/s
  Compression                      68323 KB/s
  CPU Single Threaded              2105 Million Operations/s
  Physics                          547 Frames/s
  Extended Instructions (SSE)      3183 Million Matrices/s

Memory Mark:                       2549
  Database Operations              3387 Thousand Operations/s
  Memory Read Cached               19151 MB/s
  Memory Read Uncached             12618 MB/s
  Memory Write                     9612 MB/s
  Available RAM                    15363 Megabytes
  Memory Latency                   33 Nanoseconds
  Memory Threaded                  27060 MB/s
--------------------------------------------------------------------------

Power consumption then goes up to 26W instead, but stays at 9W when idle.

I am now running stress -c 4 now. Power consumption is stable at 22W, CPU temp. after 2 minutes is at 45°C on all cores. Not too bad, the case seems to be efficient.

 

[roarking]

I have found a sh*tton of settings I have never seen in any Bios, and the first ones I looked at were the ones for 386dx machines.

Thank you for testing the new N100 device and for the BIOS screenshots. Unfortunately the BIOS of this N100 mainboard seems to lack support for "Serial Port Console Redirection" despite the fact that it has a COM header (serial port).

I had extended discussion with CWWK support and asked if "Serial Port Console Redirection" could be added to the Jasper Lake N5105/N6005 V4+V5 mainboards, which are the predecessors of this N100 board. CWWK were open to this suggestion but in the end I was told that the manufacturer of the mainboards refused too enable this feature. All devices that have a COM port in the front panel seem to have it but the devices that only have a COM header do not

"Serial Port Console Redirection" would allow to monitor the entire boot process and BIOS access via serial connection (cable or bluetooth) as shown below:

 

[Becks0815]

The case C is good. I have set high power limits (PL2 25W, PL1 12.5W) and ran stress on Linux (100% CPU usage on all cores), with a starting CPU temp. of 28°C (room: 22°C), and after 20 minutes CPU core temp. is at 48°C

n100:~$ sensors
coretemp-isa-0000
Adapter: ISA adapter
Package id 0:  +48.0°C  (high = +105.0°C, crit = +105.0°C)
Core 0:        +48.0°C  (high = +105.0°C, crit = +105.0°C)
Core 1:        +48.0°C  (high = +105.0°C, crit = +105.0°C)
Core 2:        +48.0°C  (high = +105.0°C, crit = +105.0°C)
Core 3:        +48.0°C  (high = +105.0°C, crit = +105.0°C)

With these settings I get a CPU passmark of 6100. The N5105 is at 4100.

 

[TraXter]

Hey, happy to find this thread here. Have a N200 rolling in after I initially ordered a N6005 Topton A Model, because I wasn't aware there's a successor already. They luckily(?) offered a free upgrade to N200 model, as the N6005 is not produced any more.

I'm pretty excited to try a single 32GB DDR5 module to see if it works.

One thing I noticed, the PassMark specs given on CWWK/Topton offers do not make sense?

If I'm not missing something completely here, the N200 should outperform a N100 from specification. Although N100/N200 have only E-Cores, they should also outperform the N5105 and N95 easily. Maybe they are not accurate in the currently early phase, because only 2 samples are given. Cache specs are wrong as well.

With your tuned 6000 score for a N100 it gets even more strange. Once the box is here I'll give N200 results for comparison.

 

[Becks0815]

Enough Bios games for now.

My personal best settings are:

- All unneeded hardware off: sound, emmc, sata, tpm
- PECI off (found in the N5105 thread)
- EPB override over PECI enabled
- Platform PL1&2 disabled (no time yet to play with)
- power limit 4 override disabled
- Package PL1 = 10000, PL2 = 25000, Energy eff- turbo enabled
- C-States enabled, Package limit C10

Chipset -> PCH-IO Config -> PCI Express Config: every PCI root except #7 : ASPM auto, L1 Substates L1.1&L1.2, L1 low = enabled. On port 7 ASPM auto, L1 substaes disabled, L1 low enabled - otherwise network under Ubuntu wasn't working.

I might have to change some of the PCI power settings later. High chance I have switched off every NIC except ETH0. But now the system runs at 6.3W idle, measured at the wall, while still having a CPU passmark of 5700 (at 20W peak).

 

[tusk9541]

Game on - the box is here. RAM was accepted, just like the NVME drive. I only had some initial issues with no graphics being displayed, but the problem was I tried to use HDMI while the only output I was able to set at the moment is Display port.

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

The case is impressive heavy, but the connection between CPU and case is a single copper block only, without heat pipes. There is plenty of this grey paste applied on all kind of places. I removed them and replaced it with Arctic MX4.

I haven't spotted a gap between the copper block and the CPU, which is fine.

I have installed Ubuntu 22.04 and upgraded the kernel to 6.2.11, and powertop shows C10 as lowest possible power state

The initial power consumption in BIOS was 11.3W-12W using default Bios settings, 16GB RAM, no NVME drive installed, DP monitor attached plus keyboard in an USB port plugged in, measured using the Daijing power supply and at the wall (Voltcraft 3000)

Running ubuntu kernel 6.2.11 idle with powertop savings applied (except PM on the Intel NICs, which freezes my SSH connection), SATA, sound disabled, nothing plugged in except 1 ethernet cable and PL1 and PL2 set to 10W I am down to 8.7W, measured at the wall with the daijing brick. Not too bad for the first try, but I am sure there is plenty to play around in the Bios. I have found a sh*tton of settings I have never seen in any Bios, and the first ones I looked at were the ones for 386dx machines.

idle temp Ubuntu: 33°C CPU temp after sitting idle on the desktop here for 20 mins.

So the case doesn't have the heatpipes like shown in the pic of the listing, probably only reserved for the more powerful/expensive CPUs, but more concerning, the 4 mounting holes around the CPU have gone unused, so there may be still no proper mounting pressure.

 

[Becks0815]

So the case doesn't have the heatpipes like shown in the pic of the listing, probably only reserved for the more powerful/expensive CPUs, but more concerning, the 4 mounting holes around the CPU have gone unused, so there may be still no proper mounting pressure.

if it is the case, then it is not relevant. I have replaced the original thermal paste with MX4. then I overclocked/tuned the N100 a bit, using PL1 =10W and PL2 = 25W, and ran a stress test under Linux, with all 4 cores at 100% usage. After 20 minutes, all CPU core temp were at 48°C (with 38°C as starting temp., about 2-3 secs after I started the test). This is enough cooling for me.


More tests next week - the NVME drive died and I have no replacement here.

 

[weiyiaw85]

Hey guys, I'm also doing research on 8505 / N305 / I7-1265u

What my objective is to use this small appliance to build a virtual environment as home lab to test
- PFsense / Sophos XG Home version (FREE) / Opensense
- VM I think is must - ESXI / Proxmox
- TrueNas
- Windows 11
- Kali linux
- If resource enough maybe will try out Ntop , Zeek and others

Home not much device 8 CCTV, with about less than users connect via wifi only

when I did some research it
- Previous N5105 alot of bugs , 2.5G interface not stable, end up I think is the heat cause it , not sure will happen on 8505/N305/I7-1265u
- VM ESXI / Proxmox not recognize 2.5G interface (forgot is which model already) not sure resolved or not
- only N305 support DDR5, others is DDR4 (I don't know how much different on this part)
- 8505 only 5 Core ...but benchmark result better than N305
- but N305 benchmark lower than 7505
(if see benchmark result , look like never end)

in term of brand I found look like CWWK have a better reputation ? any others better manufacture ?

 

[Becks0815]

With your tuned 6000 score for a N100 it gets even more strange. Once the box is here I'll give N200 results for comparison.

There are only 2 samples of the N200. All it takes is to use bad settings in the Bios to get bad passmark results. Just look at the ones I got. I played around a bit with PL1 and PL2 and had values between 4600 and 6200, depending on how hard I overclocked the machine. Now take a case with bad thermal design, a CPU gap or something similar and you can't do what I did.

My estimation is: The N200 will beat the N100, but it will mostly depend on how good the cooling is and how long the N200 can keep the high frequency without overheating. Beside that I wouldn't worry. Both the N100 and N200 are for sure nice and capable CPUs for oponsense or similar tasks.

 

[spitefulmonkey]

I'm trying to decide between the n200 or n305. I will be bare metal pfsense. My current isp is 2 gig but later this year they are upping to 5 gig. For home use would the 200 be enough or should I just opt for the 305?

 

[Becks0815]

I'm trying to decide between the n200 or n305. I will be bare metal pfsense. My current isp is 2 gig but later this year they are upping to 5 gig. For home use would the 200 be enough or should I just opt for the 305?

Can you get any of these boxes with a 5 gbit NIC?

 

[roarking]

Can you get any of these boxes with a 5 gbit NIC?

I do not know if a 10GbE NIC makes sense in these boxes from a networking perspective but it is possible to plug one in



I used an Intel X550-T2:

root@n5105:~# uname -ar
Linux n5105 5.10.0-20-amd64 #1 SMP Debian 5.10.158-2 (2022-12-13) x86_64 GNU/Linux
root@n5105:~# lspci
...
00:1f.5 Serial bus controller [0c80]: Intel Corporation Device 4da4 (rev 01)
01:00.0 Ethernet controller: Intel Corporation Ethernet Controller 10G X550T (rev 01)
01:00.1 Ethernet controller: Intel Corporation Ethernet Controller 10G X550T (rev 01)
02:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller SM951/PM951 (rev 01)
03:00.0 Ethernet controller: Intel Corporation Device 125c (rev 04)
04:00.0 Ethernet controller: Intel Corporation Device 125c (rev 04)
05:00.0 Ethernet controller: Intel Corporation Device 125c (rev 04)
06:00.0 Ethernet controller: Intel Corporation Device 125c (rev 04)
root@n5105:~#

This N5105 mainboard has only PCIe 3.0 x1 in the M.2 slot but the new mainboards with N100/N200/i3-N305 should allow PCI 3.0 x4 directly or x2 with an adapter board:

root@n5105:~# lspci -vv
...
01:00.0 Ethernet controller: Intel Corporation Ethernet Controller 10G X550T (rev 01)
        Subsystem: Intel Corporation Ethernet Converged Network Adapter X550-T2
...
                LnkCap: Port #0, Speed 8GT/s, Width x1, ASPM L1, Exit Latency L1 <16us
...
                LnkSta: Speed 8GT/s (ok), Width x1 (ok)
...
                LnkCap2: Supported Link Speeds: 2.5-8GT/s, Crosslink- Retimer- 2Retimers- DRS-
...
        Kernel driver in use: ixgbe
        Kernel modules: ixgbe

 

[Chura]

For OPNsense, Should I jump on N100 already or wait bit more ?

 

[alexw1982]

For OPNsense, Should I jump on N100 already or wait bit more ?

If you’re not in a rush I would wait for prices to come down a little. Otherwise, go for it!

 

[Becks0815]

For OPNsense, Should I jump on N100 already or wait bit more ?

I was not able to really test the box with opnsense so far because of a hard drive failure (old NVME drive laying around). I'll get the replacement tomorrow and then spend some time playing around with the tunings, but as far as I can see: at least the CWWK version is pretty good. The Bios has almost all the required features (except serial console, something even the V4/V50 N5105 didn't have), the one I got didn't have a cpu/case gap, and with the case version (I called it version C) I was able to run a stress test with PL1 = 10W and PL2 =25W, resulting in a passmark of 6200 (instead of the expected 4600) for 20 minutes, with 48°C core temp. at the end of the test.
I also was able to switch off 3 of the NICs in the Bios and switch on some more energy saving features and ended with 6W idle usage under Ubuntu. With two NICs switched on I was at around 7W idle under opnsense before the drive dried. Both times energy consumption measured at the wall, so based on my findings with the efficiency of the power supplies in this range (about 73-75%) we are talking about 5W such a machine draws while being idle. I am fine with that. My J3160 4 port box pulls around 7W during normal usage as opnsense router (around 20% CPU usage), I am sure the N100 will end in the same region while still offering much more CPU power if required (passmark of a J3160 is around 1100).