Topton Jasper Lake Quad i225V Mini PC Report

[andrea87]

Mine is currently running with PL1 15w and PL2 22.5W. I've set the minimum cpu frequency (there's an override on the multiplier in the bios) to 400MHz. With the kind of loads I get (10 ish % idle) the usual base frequency hovers around 800-1200MHz.

Read pages 80-85 of this thread, there's a lot of useful data and experimentation by several users.

Ram wise, I'm running 24 Gb as a 8+16. I don't feel the need (yet) to go 32gb, most of my daily loads are pretty well covered with this kind of memory.

Main OS is Proxmox, these are the vms/containers and their settings:
VM OPNsense: 4Gb allowed, actual use 2.5-2.7gb. Simple-ish setup, a few vlans, inbound VPN for easier access from outside.
VM Home Assistant + OS 8Gb allowed, actual use 5-6gb (hosts also InfluxDB and Grafana)
VM Ubuntu Server 8Gb allowed, 7gb in use. Hosts my wifi controller (TP-Link Omada), Docker + Portainer and a few containers with lightweight stuff)
CT Pihole - 256Mb assigned ~45mb actual use

With all of this running, I still have about 5gb free ram on the machine (18.5gb used)

 

[burtal]

http://pan.changwang.com:8888/s/nd8o05

new bios for v4 v5

Initial impresion ..... way better sustained cpu freq under load....for me at least, other did not had this issue

2.6+ Ghz after 15 min of prime95 on previous versions it woud drop to around 2.4 Ghs after couple of minutes

 

[Mike9474593]

Release Notes in this file (translation by deepl)


Version: FMI04 N5105
Date: 2023-04-18
Supported motherboards: N5105 V3 V4 V5


Updated the microcode issued by Intel on Feb 14, 2023.
Fixed Gen 11 virtualization crash issue

 

[Foxtrot]

Is that the one reported for OPNsense/pfsense VMs?
Patched by microcode 24?

 

[user-23]

Is that the one reported for OPNsense/pfsense VMs?
Patched by microcode 24?

yes, I think so. According to this Intel Github repo, the microcode version from 2023-02-14 changed the microcode of jasper lake processors from 24000023 to 24000024:

Intel-Linux-Processor-Microcode-Data-Files/releasenote.md at 2be47edc99eeeb5b6d76bf9450b74747cdaf3263 · intel/Intel-Linux-Processor-Microcode-Data-Files

Contribute to intel/Intel-Linux-Processor-Microcode-Data-Files development by creating an account on GitHub.

github.com

 

[Foxtrot]

Would this be a V5 version?

 

[Arcao]

I configured pfSense to update microcode to 24 during startup. I installed latest devcpu-data-intel package from FreeBSD:

fetch https://pkg.freebsd.org/FreeBSD:14:amd64/latest/All/devcpu-data-intel-20230214.pkg
pkg install devcpu-data-intel-20230214.pkg

And updated my /boot/loader.conf.local with this config:

cpu_microcode_load="YES"
cpu_microcode_name="/boot/firmware/intel-ucode.bin"

After reboot, microcode 24 is installed:

# dmesg | grep microcode
CPU microcode: updated from 0x1d to 0x24000024

Credit: Update Intel Microcode on FreeBSD - Thomas-Krenn-Wiki

 

[Arcao]

Hi guys,

Yesterday I received N5105 i226 box with fan (https://www.aliexpress.com/item/1005005101103120.html).

I use Gigabyte 256GB NVME SSD, and two Crucial SO-DIMM 4GB DDR4 2666MHz CL19.

I will use it as pfSense baremetal router. I am disappointed about the fan. Even though BIOS allows fan speed configuration with off state, it seems it doesn't work. So finally I disconnected fan from board. After tuning, here is my power consumption (with one ethernet cable connected and disconnected fan):

Idle: 4-5W, 24 deg C
Stress(30 minutes): 17W, 54 deg C

 

[Foxtrot]

Nice, I have the version without fan.
4-5W sounds great, could you share your settings?
Have to check if I received a similar power supply.


I got only down to 7-8W.
Running Proxmox.

 

[Becks0815]

I am disappointed about the fan.

Had a similar issue with a fan not being steered. Solution was to bus an arduino nano 5V version, a 5V fan (Noctua) , a 10k temp. resistor and then just write a simple script which reads the temperature and adjusts the fan speed based on the findings.

It runs in my main PC since I installed it 2 years ago and steers a fan to cool down a hard drive stack

 

[Stovar]

Nice, I have the version without fan.
4-5W sounds great, could you share your settings?
Have to check if I received a similar power supply.


I got only down to 7-8W.
Running Proxmox.

I think he already mentioned and linked my old thread with the bios tweaks, either way 4-5 watts is very impressive I think that might be the lowest wattage number any N5105 owner has ever got around here but could be wrong its been a long 136+ pages!

But then he only has one network cable attached so that is a factor, if I recall right when I pulled out my 6x ethernet cables they were roughly taking up 1 watt if not a bit more each, so its possible to get it lower with just 1 Ethernet device, I think I hit around 6-7watts with nothing attached but with the normal 5 ethernet cables connected its 8.5-9watts idle.

 

[mcsv]

Stock thermal interface at 5105 v3 (or v4) are bad even if there's no visual problem (gap) and temps do not feel terrible at first at non-high load.

In my case I got okay-ish 50c at Proxmox under near idle loads with 5-6 VMs and case t +15-20 to env, but after 6 months run a cpuz stress bench, which resulted quickly 85+ and throttle (pl2=22w)

So, messing with cooling here worth it at least just to look at what a composition:
1 [case top] + thermal paste or glue + [copper brick with 2 screws]
2 middle thermal layer + [unexpected floating copper plate]
3 one more huge stockpile of thermal paste + [finally tiny cpu at pcb]

Huge amounts of paste everywhere is to fight with a gap in such crooked solution and I'm also re-applied way more than probably is necessary there of really old 10y+ mx2.
What, probably, can give a clue about "enough amount of thermal paste" - there's a clear strips around cpu - in my case those were covered in 1mm or so of thermals, but it looks like they should align and keep middle copper plate as close to cpu as possible.

So, I got now at 5th hour of same cpuz bench stable 75c, very hot case 60+, and, probably, almost no difference under usual low loads, still worth at least checking and fixing if u got tools already available.

 

[BobS]

Stock thermal interface at 5105 v3 (or v4) are bad even if there's no visual problem (gap) and temps do not feel terrible at first at non-high load.

In my case I got okay-ish 50c at Proxmox under near idle loads with 5-6 VMs and case t +15-20 to env, but after 6 months run a cpuz stress bench, which resulted quickly 85+ and throttle (pl2=22w)

So, messing with cooling here worth it at least just to look at what a composition:
1 [case top] + thermal paste or glue + [copper brick with 2 screws]
2 middle thermal layer + [unexpected floating copper plate]
3 one more huge stockpile of thermal paste + [finally tiny cpu at pcb]

Huge amounts of paste everywhere is to fight with a gap in such crooked solution and I'm also re-applied way more than probably is necessary there of really old 10y+ mx2.
What, probably, can give a clue about "enough amount of thermal paste" - there's a clear strips around cpu - in my case those were covered in 1mm or so of thermals, but it looks like they should align and keep middle copper plate as close to cpu as possible.

So, I got now at 5th hour of same cpuz bench stable 75c, very hot case 60+, and, probably, almost no difference under usual low loads, still worth at least checking and fixing if u got tools already available.

This is a post I made when I got the HUSN version with the N6005 to fix the thermals. Difference in what you have is that I removed the copper block. removed the paint and reassembled as per my post:

https://forums.servethehome.com/index.php?threads/hunsn-model-rj03l-w-n6005-v5-w-4-i226v-nics.38464/

BobS

 

[andrea87]

So, my unit suddently died tonight after six months of light use. I've tried resetting the cmos, running it with only one stick of ram, nothing. The pc doesn't boot and none of the two lights do light up. Any suggestion? I'm running a pentium 7505 topton box.

I've tested, and there's +12V on the input, the fuse is fine and I don't see any skid marks. Trying to reset cmos does nothing.

 

[T.Sharp]

So, my unit suddently died tonight after six months of light use. I've tried resetting the cmos, running it with only one stick of ram, nothing. The pc doesn't boot and none of the two lights do light up. Any suggestion? I'm running a pentium 7505 topton box.

I've tested, and there's +12V on the input, the fuse is fine and I don't see any skid marks. Trying to reset cmos does nothing.

Do you have a power meter to see if it's drawing anything?
Any parts of the board getting warm when it's plugged in?

 

[Foxtrot]

Did you ckeck both sides of the PCB?
Also the one holding the CPU?

 

[andrea87]

Just tested. The machine pulls about 4-5W at idle for about a minute, there's a significant hotspot (up to 50ish °C) near the VRM area. Looks like an hardware fault somewhere, it might be the +5V SB regulator with a short / dead IC on the supply line somewhere. I'll try messaging TopTon to see if they help in any way.

 

[Stovar]

So, my unit suddently died tonight after six months of light use. I've tried resetting the cmos, running it with only one stick of ram, nothing. The pc doesn't boot and none of the two lights do light up. Any suggestion? I'm running a pentium 7505 topton box.

I've tested, and there's +12V on the input, the fuse is fine and I don't see any skid marks. Trying to reset cmos does nothing.

Just tested. The machine pulls about 4-5W at idle for about a minute, there's a significant hotspot (up to 50ish °C) near the VRM area. Looks like an hardware fault somewhere, it might be the +5V SB regulator with a short / dead IC on the supply line somewhere. I'll try messaging TopTon to see if they help in any way.

Darn just sounds unlucky, maybe the mainboard went bad. Any word on Toptons support?

I may go with Topton next but theres always a concern on how good the support really is with any of these Asian brands, if memory recalls right early on this mega thread some had dead units and got no reply back even.

 

[miro123]

Stock thermal interface at 5105 v3 (or v4) are bad even if there's no visual problem (gap) and temps do not feel terrible at first at non-high load.

Should I adjust for zero air gap or put a little bit PCB pressure /like all PC cpu coolers/?

There is fundamental mechanical problem with all boxes - outside case shell is actually is in direct contact with CPU - add to this tiny aluminium sheet used for casing, and I'm wondering that all goes ok for such long time.

 

[olavrb]

Thanks @Stovar for #2489.

I found "acustic noise reduction" as "acoustic noise mitigation" here:

• Advanced -> Power & performance -> CPU - Power Management Control -> CPU VR Settings -> Acoustic Noise Settings -> Acoustic Noise Mitigation

Edit: "Converted" those BIOS settings into tables with exact names taken from Changwang BIOS from 2023-04-18 (N5105 v3, 4L-V01)

Spoiler: Click to show

Advanced → ACPI Settings

Key	Value	Default
Enable ACPI Auto Configuration	Auto	Disabled

Advanced → CPU Configuration

Key	Value	Default
PECI	Disabled	Enabled

Advanced → Power & performance → CPU - Power Management Control

Key	Value	Default
Boot performance mode	Turbo Performance	Turbo Performance
Intel(R) SpeedStep(tm)	Enabled	Enabled
Platform PL1 Enable	Enabled	Enabled
Platform PL1 Power	10000	50000
Platform PL2 Enable	Enabled	Enabled
Platform PL2 Power	12000	65000
C states	Enabled	Enabled

Advanced → Power & performance → CPU - Power Management Control → CPU VR Settings → Acoustic Noise Settings → Acoustic Noise Mitigation

Key	Value	Default
Acoustic noise mitigation	Enabled	Disabled

Chipset → PCH-IO Configuration → PCI Express Configuration

Key	Value	Default
DMI Link ASPM Control	Auto	Auto

Chipset → PCH-IO Configuration → PCI Express Configuration → PCI express root port <1-8>

Key	Value	Default
ASPM	Auto	Disabled