What do you mean with it exactly?found it. i had the newer bios on. installed the one from february and the package c-states are back on the menu
Where did you find actual BIOS version for this device?
CWWK i5-1235U 6 port i226 report
- Thread starter fta
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
Where did you find actual BIOS version for this device?
Thx for the fast reply
Meant what is this with the Package C-States - what should be there or not misssing in BIOS?
I have the DDR5 version, so I think the BIOS won't be the right one
Last edited:
i am at 14-15w from the wall as measured on my smart plug when in c1 package, and most of the cores parked. I use ubuntu, have 2x16gb ram and one nvme drive. When i find the time to do it, i will try edit the bios file to enable lower c-states and test it if it helps with power consumption.
Edit: my guess is that if the cpu package would be allowed to go to lets say c6 or c7, i would think less then 10w should be possible. perhaps as low as 7w
No reason these devices can't idle at a few watt, the same chips and specs do in branded laptops and other devices (see Odroid H4), even without ASPM enabled the Odroid H4 device is around 5 watts. No matter what C-States or package C-States might be reported on these direct from China boxes, it is a struggle to get them at less than 10 watt because they are designed with corners cut, or using SoC's not for resale (engineering samples are not unheard of in these boxes) and so are not stable going in and out of C-States and its disabled in hardware.
yeah i just checked and the 1235u should be able to go to c10... so yeah, under 5w could be probably done. I have used the tools that @fta linked in his posts, but i am not sure how to proceed. There actually are no hidden lower c-states in the bios. There is only the C0/C1 and thats it, so i am not sure how to add the others
When I say no reason it can't do it, I mean no reason they could not have designed these to run at much lower power states, laptops running the same SoC's even with the screen on can draw much less power at idle than some of these boards. The problem is they haven't designed them to support low power states, they are mains powered, China doesn't pay our electricity bills, and if the components lifetime is reduced due to baking in extra heat 24/7, even better for them, they will happily sell us another one, and by not designing in power efficiency, its saved them money or allowed them to use lower quality components.
I've tried looking for Intel design papers for these SoCs, but its all NDA, you have to be a manufacture to get hold of them. I'm thinking there may be two types of recommended designs, one of them being a cheaper, quick and dirty design using fewer power rails at the cost of poor idle power consumption, and its this design China goes for every time. If I look at an Ordoid H4 board, despite fewer peripherals on board, it has more power rails (you can see these by counting the power inductors) than a typical DfC (direct from China) board based on the same SoC. This would allow more power gating or more sections of the SoC able to request lower voltages to save power.
If my suspicions are correct, this is why despite getting a hacked BIOS to allow turning on otherwise hidden C-States, if we are lucky and it remains stable, it doesn't actually drop the power consumption down anywhere near what should be achievable.
Several of us have started pulling up Protectli about their designs, which of course are just made in China by Yanling and have the same poor power consumption issues, and as Protectli haven't ever questioned power consumption, all their devices are just as bad, even their BIOS has had ASPM stuff hidden by Yanling like a lot of other boards we see DfC. Because many are raising these issues, they are starting to look seriously into the matter. They are one of the few companies that can place some pressure on their ODM to get improved designs, so it might be helpful if you and others pop over to https://www.reddit.com/r/protectli/ add your interest in better designed kit. There is currently a thread about the VP2430 that would be a good one to add your voice to as Protectli have been responding in that thread. That all feeds back to the China manufacturers and maybe where one goes, they all follow.
Last edited:
I just ordered this with the Pentium 8505 CPU since it was less than $200 for Prime Day. I'll be interested to see which revision I get. Probably the first since it's on sale, they are probably trying to clear stock.
EDIT: Arrived today, and nope, the 8505 model at least is still sporting the DDR 4 RAM v1 motherboard.
EDIT: Arrived today, and nope, the 8505 model at least is still sporting the DDR 4 RAM v1 motherboard.
Last edited:
It has been a bit since I've used my 1215u box. I just recently pulled it back out. Is there a go to bios to use for these now? I've seen several thrown around.
I know this thing is old news at this point, but three questions:
1 - What is the 4pin ATX power plug on the board for? Is that intended to be used if you want to use an ATX power supply instead of the 12V adapter?
2 -The "CPU Fan" 4pin PWM fan header ion the board doesn't seem to supply 12 volts, it can't even get my Noctua fan I bought to start spinning. Other fans I have spin super slow, like 300RPM or less. Is it mis labeled and is really a 5V fan header?
3 - What's the fan extension from standard 4-pin to some sort of micro connector? I don't see a place on the board to attach it.
EDIT: I put a multi-meter on the "12V CPU Header" and it's 5V. So unless that can be altered by the unlocked BIOS, I need to get a 5V fan
1 - What is the 4pin ATX power plug on the board for? Is that intended to be used if you want to use an ATX power supply instead of the 12V adapter?
2 -The "CPU Fan" 4pin PWM fan header ion the board doesn't seem to supply 12 volts, it can't even get my Noctua fan I bought to start spinning. Other fans I have spin super slow, like 300RPM or less. Is it mis labeled and is really a 5V fan header?
3 - What's the fan extension from standard 4-pin to some sort of micro connector? I don't see a place on the board to attach it.
EDIT: I put a multi-meter on the "12V CPU Header" and it's 5V. So unless that can be altered by the unlocked BIOS, I need to get a 5V fan
Last edited:
I installed the one here on mine last night. There are a lot of controls, more than I know what to do with but it boots up with my 8505 based model.
you can get a 12v fan. it just needs to be one with a low starting voltage. my 120mm runs at extremely low rpm, but the airflow is sufficient to provide nice extra cooling.
i have no clue about that extension. as far as i remember mine didnt have such a thing.
the 4 pin atx is alternate power source if you dont want to use the barrel jack. there are a few other mainboards that have this but its weird, because a normal atx powersupply wont work (because the 24 pin doesnt go anywhere and the powersupply doesnt get feedback from the mainboard). i would recommend sticking to the external brick unless you want to experiment.
Good news! With the new proxmox v9 kernel, there is no longer the bug that freezes the network when using the powertop --auto-tune command.
There is still an i226V driver performance bug when power management is enabled but it is corrected in kernel 6.15.6+.
With ASPM enabled + power management, my P8505 consumes 6W from the wall (shelly plug) with 2x 4GB sata drives, 64GB of RAM (Crucial) and 1x NIC UP.
There is still an i226V driver performance bug when power management is enabled but it is corrected in kernel 6.15.6+.
With ASPM enabled + power management, my P8505 consumes 6W from the wall (shelly plug) with 2x 4GB sata drives, 64GB of RAM (Crucial) and 1x NIC UP.
Your Pentium 8505 is upa dn running with 64GB of DDR4? I thought it was limited to 32GB.
Oops, you're right. I was thinking of the N305 version.
I only put 16GB in mine anyway since I'm using it as a baremetal OPNsense box anyway.
Wow. It's doing better than a lot of the N100 configurations. Whether that's the improvements in proxmox & kernal or hardware differences is an interesting question...