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.

Kahooli

Member
Dec 3, 2022
53
35
18
How did you get the copper heat pad off? I took off the screws and it still wouldn't budge from the case. I considered using something to pry it off but I didn't want to damage it.
In my case, there was almost as much thermal goop under the copper as above it. I twisted it a couple times and then pulled it off with my fingers. no issue there.... ya know except the mountain of goop.

I was literally asking my machinist buddies if they'd mill some cuts in the aluminum and copper block to add an 8mm heatpipe a week ago. lol.
 

Becks0815

Well-Known Member
Oct 15, 2022
241
336
63
How did you get the copper heat pad off? I took off the screws and it still wouldn't budge from the case. I considered using something to pry it off but I didn't want to damage it.
Remove the two screws, then just pull it off. It was only connected to the case with (a lot of) thermal paste.

Out of interest, based on the research I've done so far I thought I could run a WireGuard server/plugin directly in OPNsense. What is your reasoning for running it on a separate VM?
I had issues with disconnects, maybe also caused by my underpowered box running opnsense, I moved wireguard to my NAS, and after running wg-easy now for a while I don't plan to use the plugin anymore. Much easier to add clients. Just scan the QR code on WG easy, and your mobile phone is ready to be used. The only upgrade/change I will implement is to move the service to a VM on Proxmox, instead of running a docker service on the NAS. Two pieces of hardware less that can fail (Switch & NAS) and render VPN impossible.
 

Becks0815

Well-Known Member
Oct 15, 2022
241
336
63
Man I thought my topton thermal design just sucked. Nope. assembly error. mountains of shitty thermal goop and a crooked gap pad on the NICs.
I replaced both the gap pad and all the thermal goop (and under the copper) with generic paste and MX5. From 95C under load to 48C. Unreal.
What kind of box did you order, and from which company?
 

ocny153

New Member
Apr 28, 2023
12
5
3
I had issues with disconnects, maybe also caused by my underpowered box running opnsense, I moved wireguard to my NAS, and after running wg-easy now for a while I don't plan to use the plugin anymore. Much easier to add clients. Just scan the QR code on WG easy, and your mobile phone is ready to be used. The only upgrade/change I will implement is to move the service to a VM on Proxmox, instead of running a docker service on the NAS. Two pieces of hardware less that can fail (Switch & NAS) and render VPN impossible.
Thanks, I'll keep that in mind when I get round to setting up the VPN. The manual setup for clients without a QR code does sound annoying. I assume you wont use docker to run Wireguard on Proxmox, so you'll set up a VM/container running a linux distro and the wireguard service itself (eg not using wg-easy)?
 

Becks0815

Well-Known Member
Oct 15, 2022
241
336
63
Thanks, I'll keep that in mind when I get round to setting up the VPN. The manual setup for clients without a QR code does sound annoying. I assume you wont use docker to run Wireguard on Proxmox, so you'll set up a VM/container running a linux distro and the wireguard service itself (eg not using wg-easy)?
I have created a CT on Proxmox running Debian 11, installed docker on it, and now I am running wg-easy and portainer on this virtual machine. The CT does only share a virtual bridge/network connection with the VM running opnsense, but is separated from the network Proxmox itself runs on.

While I know software within Docker doesn't cause problems on a Linux host like it is the case compared to a direct installation, I want to keep the Proxmox host clean and out of the loop of any service I access from the WLAN. So: on Proxmox I only install and run VMs, and within these VMs I do what I want, but only within its own network.
 

Becks0815

Well-Known Member
Oct 15, 2022
241
336
63
Topton Computer Store
2023 New 2.5G Soft Router Pentium N6005 N5105 V5 4x Intel i226 Nics 2*NVMe 2*DDR4 Fanless Mini PC Firewall Appliance VPN Server
(I got the no ram no storage N5105)
Ah yeah, the N5105/N6005 from Topton were well known for the problems. A good amount of pages in the N5105 thread is dedicated to this. I wonder if the quality has improved on the N100/N200/N305 boxes, or if we will see similar issues here. At the moment there is not enough feedback on this. To be more precise, I think no one beside me had a closer look at the new series.
 

TraXter

New Member
Apr 14, 2023
7
13
3
So, I got the N200 Topton model yesterday after 14 days shipping from .cn to .de with "sellers shipping method". No extra taxes or fees.

I put in a Samsung 980 (non-Pro) NVMe and Crucial CT32G48C40S5 DDR5 32GB Module. Memory is detected fully :cool:. I reapplied the thermal paste with MX4 between cpu/gpu and also between copper block and black case. Adjusted the BIOS settings according to @Becks0815 thread earlier.

Tests are not bad but also not awesome:
Code:
Intel N200 (x86_64)
4 cores @ 3700 MHz  |  31.2 GiB RAM
Number of Processes: 4  |  Test Iterations: 1  |  Test Duration: Medium
--------------------------------------------------------------------------
CPU Mark:                          5265
  Integer Math                     19327 Million Operations/s
  Floating Point Math              11874 Million Operations/s
  Prime Numbers                    25.6 Million Primes/s
  Sorting                          7339 Thousand Strings/s
  Encryption                       4361 MB/s
  Compression                      54874 KB/s
  CPU Single Threaded              2164 Million Operations/s
  Physics                          201 Frames/s
  Extended Instructions (SSE)      2500 Million Matrices/s

Memory Mark:                       2141
  Database Operations              1796 Thousand Operations/s
  Memory Read Cached               18773 MB/s
  Memory Read Uncached             8507 MB/s
  Memory Write                     6788 MB/s
  Available RAM                    30255 Megabytes
  Memory Latency                   31 Nanoseconds
  Memory Threaded                  18188 MB/s
--------------------------------------------------------------------------
While the most CPU results match your's the overall score of 5265 is way below @Becks0815 's N100 model. There is quite some difference in especially the Physics, Sorting and overall memory values. They are mostly 2/3 of what to be expected. PerformanceTest FAQ - Poor PC Performance confirms the cpu results can be related to memory.

So, checking memory:
Code:
# sudo dmidecode --type 17
# dmidecode 3.3
Getting SMBIOS data from sysfs.
SMBIOS 3.5.0 present.
# SMBIOS implementations newer than version 3.3.0 are not
# fully supported by this version of dmidecode.

Handle 0x0028, DMI type 17, 92 bytes
Memory Device
        Array Handle: 0x0027
        Error Information Handle: Not Provided
        Total Width: 64 bits
        Data Width: 64 bits
        Size: 32 GB
        Form Factor: SODIMM
        Set: None
        Locator: Controller0-ChannelA-DIMM0
        Bank Locator: BANK 0
        Type: <OUT OF SPEC>
        Type Detail: Synchronous
        Speed: 4800 MT/s
        Manufacturer: Crucial Technology
        Serial Number: E7EE****
        Asset Tag: 9********
        Part Number: CT32G48C40S5.M16A1
        Rank: 2
        Configured Memory Speed: 4800 MT/s
        Minimum Voltage: 1.1 V
        Maximum Voltage: 1.1 V
        Configured Voltage: 1.1 V
        Memory Technology: DRAM
        Memory Operating Mode Capability: Volatile memory
        Firmware Version: Not Specified
        Module Manufacturer ID: Bank 6, Hex 0x9B
        Module Product ID: Unknown
        Memory Subsystem Controller Manufacturer ID: Unknown
        Memory Subsystem Controller Product ID: Unknown
        Non-Volatile Size: None
        Volatile Size: 32 GB
        Cache Size: None
        Logical Size: None

lshw -C memory
  *-firmware
       description: BIOS
       vendor: American Megatrends International, LLC.
       physical id: 0
       version: 5.27
       date: 03/18/2023
       size: 64KiB
       capabilities: pci upgrade shadowing cdboot bootselect socketedrom edd int13floppynec int13floppytoshiba int13floppy360 int13floppy1200 int13floppy720 int13floppy2880 int5printscreen int14serial int17printer int10video acpi usb biosbootspecification uefi
  *-memory
       description: System Memory
       physical id: 27
       slot: System board or motherboard
       size: 32GiB
     *-bank:0
          description: SODIMM Synchronous 4800 MHz (0.2 ns)
          product: CT32G48C40S5.M16A1
          vendor: Crucial Technology
          physical id: 0
          serial: E7EE****
          slot: Controller0-ChannelA-DIMM0
          size: 32GiB
          width: 64 bits
          clock: 505MHz (2.0ns)
     *-bank:1
          description: [empty]
          physical id: 1
          slot: Controller1-ChannelA-DIMM0
  *-cache:0
       description: L1 cache
       physical id: 32
       slot: L1 Cache
       size: 128KiB
       capacity: 128KiB
       capabilities: synchronous internal write-back data
       configuration: level=1
  *-cache:1
       description: L1 cache
       physical id: 33
       slot: L1 Cache
       size: 256KiB
       capacity: 256KiB
       capabilities: synchronous internal write-back instruction
       configuration: level=1
  *-cache:2
       description: L2 cache
       physical id: 34
       slot: L2 Cache
       size: 2MiB
       capacity: 2MiB
       capabilities: synchronous internal write-back unified
       configuration: level=2
  *-cache:3
       description: L3 cache
       physical id: 35
       slot: L3 Cache
       size: 6MiB
       capacity: 6MiB
       capabilities: synchronous internal write-back unified
       configuration: level=3
  *-memory UNCLAIMED
       description: RAM memory
       product: Intel Corporation
       vendor: Intel Corporation
       physical id: 14.2
       bus info: pci@0000:00:14.2
       version: 00
       width: 64 bits
       clock: 33MHz (30.3ns)
       capabilities: pm cap_list
       configuration: latency=0
       resources: iomemory:600-5ff iomemory:600-5ff memory:6002010000-6002013fff memory:6002016000-6002016fff
As the specs are similar to earlier modules mentioned here, there shouldn't be any significant difference. Was looking for XMP but it is missing completely in BIOS, so not sure if i can adjust anything else. Will let you know once I find something.
 
Last edited:
  • Like
Reactions: amitsax

Becks0815

Well-Known Member
Oct 15, 2022
241
336
63
There is quite some difference in especially the Physics, Sorting and overall memory values. They are mostly 2/3 of what to be expected. PerformanceTest FAQ - Poor PC Performance confirms the cpu results can be related to memory.
I bought the one with the lowest CL values I could find, and with standard values and without playing around with PL1/PL2 the N100 is at around 4600 passmark. So 5200 is good.
 

TraXter

New Member
Apr 14, 2023
7
13
3
I bought the one with the lowest CL values I could find, and with standard values and without playing around with PL1/PL2 the N100 is at around 4600 passmark. So 5200 is good.
5200 is with your BIOS settings applied, with "optimized BIOS defaults" I get around 4000. So something is off here. My CL40 vs your CL38 should not make that big of difference I would assume.
 

ocny153

New Member
Apr 28, 2023
12
5
3
Crucial CT32G48C40S5 DDR5 32GB Module
5200 is with your BIOS settings applied, with "optimized BIOS defaults" I get around 4000
Could it be that using more memory than supported by the CPU causes this? It would seem unlikely, but who knows! I've ordered the same RAM but a 16GB module (CT16G48C40S5) to put in my CWWK N100. When it gets here in 2 to 3 weeks I can run the same tests.
 

fta

Active Member
Feb 19, 2017
155
210
43
94
5200 is with your BIOS settings applied, with "optimized BIOS defaults" I get around 4000. So something is off here. My CL40 vs your CL38 should not make that big of difference I would assume.
What was your PL1/PL2 for both cases? The N200 has a higher boost clock, but otherwise is essentially the same as the N100. If the N100 can hit the PL2 of 25W under full load, there's not going to be a big difference between the N100 and N200 under short term load. Under long term load with PL1=6, it's the same due to the restrictive PL1.
 

TraXter

New Member
Apr 14, 2023
7
13
3
Could it be that using more memory than supported by the CPU causes this?
I don't think its the size, because here is a 32GB test with faster memory results:

After a looking through this thread and doing some BIOS tweaks this is what I'm getting on the test.
I have a Topton unit that looks like Variation C from the first posts images.

Code:
                          PassMark PerformanceTest Linux

Intel N200 (x86_64)

4 cores @ 3700 MHz  |  31.1 GiB RAM
Number of Processes: 4  |  Test Iterations: 1  |  Test Duration: Medium
--------------------------------------------------------------------------

CPU Mark:                          5529
  Integer Math                     19107 Million Operations/s
  Floating Point Math              11902 Million Operations/s
  Prime Numbers                    26.3 Million Primes/s
  Sorting                          7736 Thousand Strings/s
  Encryption                       4032 MB/s
  Compression                      52081 KB/s
  CPU Single Threaded              2151 Million Operations/s
  Physics                          494 Frames/s
  Extended Instructions (SSE)      2503 Million Matrices/s

Memory Mark:                       2658
  Database Operations              2759 Thousand Operations/s
  Memory Read Cached               19515 MB/s
  Memory Read Uncached             12539 MB/s
  Memory Write                     9733 MB/s
  Available RAM                    28961 Megabytes
  Memory Latency                   31 Nanoseconds
  Memory Threaded                  28111 MB/s

--------------------------------------------------------------------------


Results submitted: https://www.passmark.com/baselines/V10/display.php?id=503740048746
@Kenjutso would you share your CL timing on your DDR5 module?

I've ordered the same RAM but a 16GB module (CT16G48C40S5) to put in my CWWK N100. When it gets here in 2 to 3 weeks I can run the same tests.
Awesome, thanks! I mean, speed is not that important for me. It's rather the size, as I will be running Proxmox with kubernetes vms for a homelab on the box, but still be curious what's causing the difference and see your results.

What was your PL1/PL2 for both cases? The N200 has a higher boost clock, but otherwise is essentially the same as the N100. If the N100 can hit the PL2 of 25W under full load, there's not going to be a big difference between the N100 and N200 under short term load. Under long term load with PL1=6, it's the same due to the restrictive PL1.
Stock (~4000 Score):
Platform PL1 Disabled
Platform PL2 Enabled with Power 10000

Tuned (~5200 Score):
Platform PL1 Enabled with Power 10000
Platform PL2 Enabled with Power 25000
 
Last edited:
  • Like
Reactions: ChlorophyllJoe

TraXter

New Member
Apr 14, 2023
7
13
3
Just noticed I had set Platform PL1/2 and not Package PL1/2 in the Turbo sub menu. So I retested and still got 5161 - so essentially no difference.

Tuned (~5161 Score):
Platform PL1 Disabled
Platform PL2 Disabled
Turbo Menu/Package PL1 with Power 10000
Turbo Menu/Package PL2 with Power 25000
 
  • Like
Reactions: ChlorophyllJoe

fta

Active Member
Feb 19, 2017
155
210
43
94
Stock (~4000 Score):
Platform PL1 Disabled
Platform PL2 Enabled with Power 10000

Tuned (~5200 Score):
Platform PL1 Enabled with Power 10000
Platform PL2 Enabled with Power 25000
This makes sense. PL2=10W is really low. The CPU can't boost very much at that power level. In the second case, it will boost to 25W until you hit Tau, which is probably 28s, and then drop down to 10W. This will perform significantly better than your first case in pretty much any benchmark.
 

TraXter

New Member
Apr 14, 2023
7
13
3
OK, it's Proxmox!

I cleared CMOS, applied Package PL1 10000/PL2 25000, ran memtest86 for an hour without errors. Then installed Ubuntu Server 22.04 and ran passmark:

Bash:
Ubuntu 22.04 Server:

                          PassMark PerformanceTest Linux


Intel N200 (x86_64)
4 cores @ 3700 MHz  |  31.1 GiB RAM
Number of Processes: 4  |  Test Iterations: 1  |  Test Duration: Medium
--------------------------------------------------------------------------
CPU Mark:                          5578
  Integer Math                     19371 Million Operations/s
  Floating Point Math              11938 Million Operations/s
  Prime Numbers                    26.1 Million Primes/s
  Sorting                          7928 Thousand Strings/s
  Encryption                       4070 MB/s
  Compression                      54220 KB/s
  CPU Single Threaded              2111 Million Operations/s
  Physics                          485 Frames/s
  Extended Instructions (SSE)      2458 Million Matrices/s

Memory Mark:                       2593
  Database Operations              2754 Thousand Operations/s
  Memory Read Cached               18932 MB/s
  Memory Read Uncached             12171 MB/s
  Memory Write                     10210 MB/s
  Available RAM                    31391 Megabytes
  Memory Latency                   34 Nanoseconds
  Memory Threaded                  27861 MB/s
--------------------------------------------------------------------------
That are the results I was looking for! Then reinstalled Proxmox:
Bash:
Proxmox:


                          PassMark PerformanceTest Linux


Intel N200 (x86_64)
4 cores @ 3700 MHz  |  31.1 GiB RAM
Number of Processes: 4  |  Test Iterations: 1  |  Test Duration: Medium
--------------------------------------------------------------------------
CPU Mark:                          5219
  Integer Math                     19344 Million Operations/s
  Floating Point Math              11912 Million Operations/s
  Prime Numbers                    26.0 Million Primes/s
  Sorting                          6890 Thousand Strings/s
  Encryption                       4413 MB/s
  Compression                      53916 KB/s
  CPU Single Threaded              2177 Million Operations/s
  Physics                          200 Frames/s
  Extended Instructions (SSE)      2504 Million Matrices/s

Memory Mark:                       2133
  Database Operations              1792 Thousand Operations/s
  Memory Read Cached               18605 MB/s
  Memory Read Uncached             8451 MB/s
  Memory Write                     6725 MB/s
  Available RAM                    30098 Megabytes
  Memory Latency                   31 Nanoseconds
  Memory Threaded                  18173 MB/s
--------------------------------------------------------------------------
Well back down a third on the memory results. Then tried PVE Kernel 6.2:
Bash:
Proxmox 6.2 Kernel:

                          PassMark PerformanceTest Linux


Intel N200 (x86_64)
4 cores @ 3700 MHz  |  31.1 GiB RAM
Number of Processes: 4  |  Test Iterations: 1  |  Test Duration: Medium
--------------------------------------------------------------------------
CPU Mark:                          5208
  Integer Math                     19258 Million Operations/s
  Floating Point Math              11888 Million Operations/s
  Prime Numbers                    23.6 Million Primes/s
  Sorting                          6944 Thousand Strings/s
  Encryption                       4351 MB/s
  Compression                      54363 KB/s
  CPU Single Threaded              2161 Million Operations/s
  Physics                          209 Frames/s
  Extended Instructions (SSE)      2479 Million Matrices/s

Memory Mark:                       2336
  Database Operations              1823 Thousand Operations/s
  Memory Read Cached               19191 MB/s
  Memory Read Uncached             11038 MB/s
  Memory Write                     8511 MB/s
  Available RAM                    30281 Megabytes
  Memory Latency                   32 Nanoseconds
  Memory Threaded                  18342 MB/s
--------------------------------------------------------------------------
To notice is quite a significant bump in write/read results, and +200 overall score. But still not reaching the ubuntu results.

Well and this fixed it finally in Proxmox: echo "powersave" | tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

Code:
Proxmox with Kernel 6.2 and powersave profile:

                          PassMark PerformanceTest Linux


Intel N200 (x86_64)
4 cores @ 3700 MHz  |  31.1 GiB RAM
Number of Processes: 4  |  Test Iterations: 1  |  Test Duration: Medium
--------------------------------------------------------------------------
CPU Mark:                          5517
  Integer Math                     19288 Million Operations/s
  Floating Point Math              11881 Million Operations/s
  Prime Numbers                    25.7 Million Primes/s
  Sorting                          7633 Thousand Strings/s
  Encryption                       4022 MB/s
  Compression                      53773 KB/s
  CPU Single Threaded              2121 Million Operations/s
  Physics                          483 Frames/s
  Extended Instructions (SSE)      2438 Million Matrices/s

Memory Mark:                       2578
  Database Operations              2768 Thousand Operations/s
  Memory Read Cached               18786 MB/s
  Memory Read Uncached             12178 MB/s
  Memory Write                     10339 MB/s
  Available RAM                    30251 Megabytes
  Memory Latency                   34 Nanoseconds
  Memory Threaded                  27930 MB/s
--------------------------------------------------------------------------
Clearly the physics and mem results went up. I'm OK with that, knowing the hardware is fine. Still interested in your results!
 
Last edited:

Becks0815

Well-Known Member
Oct 15, 2022
241
336
63
Mine with PL1/PL2 10/25, run on Proxmox with the PVE 6.2 kernel, with Opnsense and another VM still running in the background (mostly idle). So they are a bit lower than it would be the case with only Proxmox.

Code:
Intel N100 (x86_64)
4 cores @ 3400 MHz  |  15.4 GiB RAM
Number of Processes: 4  |  Test Iterations: 1  |  Test Duration: Medium
--------------------------------------------------------------------------
CPU Mark:                          5536
  Integer Math                     17239 Million Operations/s
  Floating Point Math              10708 Million Operations/s
  Prime Numbers                    23.5 Million Primes/s
  Sorting                          9499 Thousand Strings/s
  Encryption                       4399 MB/s
  Compression                      55736 KB/s
  CPU Single Threaded              2091 Million Operations/s
  Physics                          373 Frames/s
  Extended Instructions (SSE)      2553 Million Matrices/s

Memory Mark:                       2043
  Database Operations              1726 Thousand Operations/s
  Memory Read Cached               18993 MB/s
  Memory Read Uncached             12494 MB/s
  Memory Write                     9538 MB/s
  Available RAM                    3089 Megabytes
  Memory Latency                   32 Nanoseconds
  Memory Threaded                  27474 MB/s
--------------------------------------------------------------------------
 
  • Like
Reactions: TraXter