Topton Jasper Lake Quad i225V Mini PC Report
- Thread starter ReturnedSword
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
Have you by any change experimented with doing AMD iGPU passthrough to drive dual monitors?
Looks interesting. Is the fan size 60 or 80mm now?Happy Halloween Ladies & Gents!Looks like Loksing is promoting a newer version of the N5105/N6005 with four new upgrades! Including booting from the TF Card slot now!! These products change constantly - so roll the roulette wheel and take your chances . . . technology marches on . . .
View attachment 25107
That's awesome! I was picturing a longer wire run, but 5ft is obviously nothing to worry about. I'd probably do the same if my equipment was close to a breaker panel. Would make for a very tidy setup.
Hi guys
A few days ago I received my barebone N5100 router and I installed a 128 nvme ssd, and a 3200mhz stick (running at 2933mhz)
Model:
- Techvision TVI7309X (reported by OS)
- GF-1338NP-12 (written on the pcb and reported by BIOS)
Notes:
- my unit had no gap between CPU and heatsink, but a sloppy thermal paste job nevertheless
- it supports secure boot
- Windows 10 activates automatically, but 11 doesn't
- power brick (12V / 4A) does not seem to be high quality, but stays cool (0.1W by itself)
- does not come with any documentation no website to support the product (eg BIOS updates)
- the system does not have serious thermal problems taking into account it is fanless. It idles at around 30 C.
- see my geekbench and userbenchmark results.
TDP:
Now here comes the weird part. Comparing it to the J4125 and N5105, the N5100 should be the clear winner when it comes to power consumption. In my tests with nothing connected to the unit I found this:
- ~ 9W running idle, ~ 23 W under load, ~ 1.7W powered off.
I never got it to go under 7W, and the curious thing is that I found this video with the same router using a N5105 CPU idle-ing at ~ 4W.
There are TONS of potions in the BIOS, but I do not have the knowledge needed to mess with most of them. My guess is that modifying some of these settings will help the power consumption. Or maybe a future BIOS update.
I would much apreciate if anyone can tell me which BIOS settings I can change to reduce the TDP to a minimum.
Thanks
A few days ago I received my barebone N5100 router and I installed a 128 nvme ssd, and a 3200mhz stick (running at 2933mhz)
Model:
- Techvision TVI7309X (reported by OS)
- GF-1338NP-12 (written on the pcb and reported by BIOS)
Notes:
- my unit had no gap between CPU and heatsink, but a sloppy thermal paste job nevertheless
- it supports secure boot
- Windows 10 activates automatically, but 11 doesn't
- power brick (12V / 4A) does not seem to be high quality, but stays cool (0.1W by itself)
- does not come with any documentation no website to support the product (eg BIOS updates)
- the system does not have serious thermal problems taking into account it is fanless. It idles at around 30 C.
- see my geekbench and userbenchmark results.
TDP:
Now here comes the weird part. Comparing it to the J4125 and N5105, the N5100 should be the clear winner when it comes to power consumption. In my tests with nothing connected to the unit I found this:
- ~ 9W running idle, ~ 23 W under load, ~ 1.7W powered off.
I never got it to go under 7W, and the curious thing is that I found this video with the same router using a N5105 CPU idle-ing at ~ 4W.
There are TONS of potions in the BIOS, but I do not have the knowledge needed to mess with most of them. My guess is that modifying some of these settings will help the power consumption. Or maybe a future BIOS update.
I would much apreciate if anyone can tell me which BIOS settings I can change to reduce the TDP to a minimum.
Thanks
Attachments
btw... there is a jumper on th pcb called AUTO_PWRON. You can use it to change the behaviour of the unit when power is restored (combined with the corresponding BIOS setting).
Welcome aboard.
Go to Advanced, then CPU power management control. What you are going to want to change are both the PL1/PL2 values in this menu (values are in mW, so 15W = 15000) and inside the "View / configure turbo options".
Other things you might want to set both there and in the general advanced setup:
Enable C-states
Set PECI to off
Enable turbo
Enable Speedstep
Set ACPI as auto
Go back a few pages in the thread, I've done a fair bit of testing on a device very similar to yours with a Topton 4xi225 / pentium 7505.
I imagine you're measuring power at the wall, be aware that most psu's below 10W load are quite inefficient. Mine at 6W output draws 9.5W at the plug.
From looking over previous pics of the 1338NP-12 boards, it appears you have a newer version (1.4). I see a few changes to the board, like smaller (probably more efficient) MOSFETs for some parts of the VRM, smaller inductor for what appears to be the iGPU rail, some capacitors, and the addition of that auto-power-on jumper. Apologies if someone has already mentioned this, I'm losing track of all the pages.
huh ? .... they are both based on Tremont architecture, so they are basically the same .... Elk and jasper launched in the same time, and i think jasper was official after Elk
Elk is "tunned" for IOT while Jasper is more tunned for low power laptops. and desktopsChange those breakers C should not be used in households
you will burndown before C cuts off power ... change them to B curveidle temp indicates that you indeed dont have a gap .... but did you actually verified by removing the NIC heatsink.
Last edited:
Yes, knowing about it from reviews I did verify while changing the thermal paste.
Quote from CW support
" V5 is mainly an upgrade
1. In the case of single M2, disable the speed of the other adapter board and add it to the separate M2 to make the speed reach X2,
2. There is also a clock chip added. No solid-state master control is selected, and any brand can use it.
3. Also fixed the TF card slot, changed the channel to USB, supporting system guidance"
" V5 is mainly an upgrade
1. In the case of single M2, disable the speed of the other adapter board and add it to the separate M2 to make the speed reach X2,
2. There is also a clock chip added. No solid-state master control is selected, and any brand can use it.
3. Also fixed the TF card slot, changed the channel to USB, supporting system guidance"
Yesterday I had a surprise with the power supply. The cable (europlug to 320-c7) failed. It looks like the brass contacts are deformed and don't make contact anymore with the psu side. Also, I've noticed (after a few days...) that almost the entirety of one of the plug's prongs is covered in plastic from the injection molding. Replaced the cable with a proper one and everything works again.
Here in Italy most of the breakers I've seen installed in any house's panel are C curve, either C10 for light / low power circuits or C16 for "normal" outlets. The local code (CEI 23/3-1, which applies EU EN 60898-1) I believe specifies either B or C curve breakers for household use. I'd have to ask an electrician friend to be certainly sure.
/OTChange those breakers C should not be used in households
Here in Italy most of the breakers I've seen installed in any house's panel are C curve, either C10 for light / low power circuits or C16 for "normal" outlets. The local code (CEI 23/3-1, which applies EU EN 60898-1) I believe specifies either B or C curve breakers for household use. I'd have to ask an electrician friend to be certainly sure.
This is incorrect. Both are Tremont-based chips. J6* is Elkhart Lake, N5* is Jasper Lake. Jasper Lake was actually released later than Elkhart Lake. Elkhart is intended for IOT/embedded devices, while Jasper for low power desktops, laptops, and tablets.
Please do not immediately believe seller's descriptions (those saying 11th/12th gen crap). Take time to double check Intel's site or Wikipedia.
Same here. I ordered from Topton Computer Store on Oct 5, 2022. Marked as shipped but still has status "Awaiting seller dispatch". I contacted them and the response is "the shipping info will be update soon please dont worry".
As you probably know, Aliexpress is running the 11.11 promotion of $3 off every $30. You can add the coupon "1111US30" at check out to take $30 off $200 orders. I got an AMD Ryzen R5-5600U (no ram, no SSD) from CWWK for $296.36, including tax and shipping. I just want to share this information with you all.
I wanted to mention something I think is very easily overlooked regarding shipping of these routers.
Take a look at the router I bought:
It says NO TAX & 3-10 days delivery. That is true, but ONLY IF YOU CHANGE THE SHIPPING METHOD to "seller shipping method".
As I've been quite familiar with the AE shopping process, when I first ordered the router I clicked the "buy it now" button without taking a look at other shipping options. After a day, I saw in the details of my order that it was going to be shipped in 9 days. I thought something was not quite right, so I re-read the listing.
I ended up canceling that order and re-placing the order this time with the seller shipping method selected (still free). And to my surprise, the item rived in 10 days after placing the order (to Romania), with no extra tax or costs.
I think people might be doing the same as I first did, from what I can read around.
I took a bit of time to fiddle with the BIOS setting you mentioned. None of them resulted in a lower TDP when idle, so the summary is:
- ~ 9W running idle, ~ 23 W under load, ~ 1.7W powered off.
So I took a look at the CPU Wattage in HWinfo, where I saw that the CPU was drawing 3 - 4W on idle. So I guess the power transformer might be inefficient. I tried other two supplies I had around (also cheap ones) and there was no difference. I've tried switching the NVMe with a SATA SSD and the TDP stayed pretty much the same. There was nothing else connected.
All this led me to another think I don't get. Take a look at the TDP of J4125, N5105 and N5100:
How can the N5105 in this video get even below 4W? I ordered the N5100 instead of N5105 specifically for the lower TDP, and now I am quite buffled to see they behave pretty much inversely to my expectation
Last edited: