Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
Nicolai Mortensen

ATX12VO vs ATX12V efficiency

Recommended Posts

Posted · Original PosterOP

Hi everyone.

 

In the video "This is NOT a normal power supply... - ATX 12VO" Linus and his team managed to get the idle power consumption halfed with the new ATX standard, but a few things in the video raised some questions for me. I'm hoping theres a few extremely bright men and women here, that'll be able to dumb it down to a level I can understand fully.

 

In the video Linus said "Here's the thing. Converting from AC to DC power is extremely inefficient and conventional powersupplies do it multiple times for the multiple voltages that they need to supply. Now it's actually better to go from AC to 12 volt only, like we're doing here, and the derive the other required DC voltages from that [...] But Linus you might say, some highend powersupplies do that already anyway." Does this mean the old standard ATX12V powersupplies that function in this way, would be able to provide a similar efficiency as the new ATX12VO while Idling?

 

The we have the 80 Plus Titanium certification, which have a 10% load standard, that must be at least 90%. Would this be a competitor to the new standard or would you see similar results as in the video aswell here?

 

I hope someone will be able to explain it in debth and thank you in advance for your time

Link to post
Share on other sites
14 minutes ago, Nicolai Mortensen said:

Hi everyone.

 

In the video "This is NOT a normal power supply... - ATX 12VO" Linus and his team managed to get the idle power consumption halfed with the new ATX standard, but a few things in the video raised some questions for me. I'm hoping theres a few extremely bright men and women here, that'll be able to dumb it down to a level I can understand fully.

 

In the video Linus said "Here's the thing. Converting from AC to DC power is extremely inefficient and conventional powersupplies do it multiple times for the multiple voltages that they need to supply. Now it's actually better to go from AC to 12 volt only, like we're doing here, and the derive the other required DC voltages from that [...] But Linus you might say, some highend powersupplies do that already anyway." Does this mean the old standard ATX12V powersupplies that function in this way, would be able to provide a similar efficiency as the new ATX12VO while Idling?

 

The we have the 80 Plus Titanium certification, which have a 10% load standard, that must be at least 90%. Would this be a competitor to the new standard or would you see similar results as in the video aswell here?

 

I hope someone will be able to explain it in debth and thank you in advance for your time

i hope this fail to be honest, i didn't like the idea of taking 5v from the motherboard because motherboard makers will charge more for extra 5v ports for it and i dont see the benefits of it mostly my device is on load and never idle which wont benefits most people,, i guess OEMs will adopt this feature since they sell governments and companies that have normal work load that usually on 40% load

Link to post
Share on other sites

Putting the DC/DCs in the PSU is essentially equivalent minus the difference in cable loss, yes.


F@H
Desktop: i7-5960X 4.4GHz, Noctua NH-D14, ASUS Rampage V, 32GB, RTX2080S, 2TB NVMe SSD, 2x16TB HDD RAID0, Corsair HX1200, Thermaltake Overseer RX1, Samsung 4K curved 49" TV, 23" secondary

Mobile SFF rig: i9-9900K, Noctua NH-L9i, Asrock Z390 Phantom ITX-AC, 32GB, GTX1070, 2x1TB NVMe SSD RAID0, 2x5TB 2.5" HDD RAID0, Athena 500W Flex (Noctua fan), Custom 4.7l 3D printed case

 

Dell XPS 2 in 1 2019, 32GB, 1TB, 4K

 

GPD Win 2

Link to post
Share on other sites
Posted · Original PosterOP
22 minutes ago, Kilrah said:

Putting the DC/DCs in the PSU is essentially equivalent minus the difference in cable loss, yes.

The cable loss will happen either way, doesn't matter if it runs from the motherboard or psu, right? 

 

How can I recognise a psu that does dc to dc in this way? 

Link to post
Share on other sites
2 hours ago, Nicolai Mortensen said:

In the video Linus said

Ignore what Linus said. He got so many things wrong. The video should be pulled imo.

 

2 hours ago, Nicolai Mortensen said:

Does this mean the old standard ATX12V powersupplies that function in this way, would be able to provide a similar efficiency as the new ATX12VO while Idling?

Yes. New efficiency targets will require system integrators to use power supplies that are at least 70% efficient at 2% load for PSUs over 500w, and 70% at 10w for PSUs 500W and below. This is for both ATX12VO and ATX12V power supplies. A 12VO power supply makes it slightly easier to hit this target, but it's not necessary and there are already plenty of existing ATX12V power supplies that do this. For example the Corsair RM is a modified version of the RMx that was designed to meet this efficiency target for system integrators. It's functionally the same as the RMx, based on the same platform and design, however uses a different LLC controller which has a "kick" mode that boosts efficiency at low loads. It's also using a different fan, cables, and capacitors to keep costs slightly lower for SI's (it's about $10 cheaper than the RMx).

 

I'm also sceptical of if 12VO actually provides any better efficiency. PSUs are tested for efficiency. Motherboards are not. It's just moving the efficiency losses from generating the minor rails from the PSU to the motherboard where it won't show up in the PSU efficiency testing. The only difference is the very minimal loss you would see from the cables plugging in to the motherboard, but if you're plugging in peripheral devices they still need a cable whether it's from the PSU or from the motherboard and will see similar losses due to cable length. 

 

1 hour ago, Nicolai Mortensen said:

How can I recognise a psu that does dc to dc in this way? 

Pretty much any good power supply in the last 5-10 years will use DC-DC for its generation of minor rails, for example the seasonic focus Linus compared to in the video is DC-DC. If you look at PSU reviews (proper reviews not customer reviews) they will tell you if the PSU uses DC-DC for its minor rails. You can also look at photos of the internals to see if there are DC-DC converters, often a small panel on the secondary side (close to where the cables come out) with two small coils on it.

Screenshot_20200816-195621_Chrome.thumb.jpg.97cb5b2688fed936f43657d929505e3c.jpg

 

Also what Linus said in the video about converting AC to DC multiple times is wrong.

 

 

2 hours ago, Nicolai Mortensen said:

The we have the 80 Plus Titanium certification, which have a 10% load standard, that must be at least 90%. Would this be a competitor to the new standard or would you see similar results as in the video aswell here?

ATX12VO does not necessarily mean more efficient than a standard ATX12V power supply, at idle or at any load. You could have a 80+ Standard ATX12VO power supply. It doesn't even need to meet those low load efficiency targets I mentioned earlier (that's something separate) though it's likely that most will meet it since ATX12VO is aimed at system integrators who are expected to meet those targets.

The "50-60% efficiency at around 10% load" figure given by Linus in the video for standard ATX12V power supplies is very, very wrong. Any 80+ rated PSU is going to be much higher than that.

 

 

The results shown in the video are false. The test they performed was flawed and there's no way they could have got that big of a difference between power consumption between the two units if the two different systems they tested had same power draw. Very misleading.

I go in to more detail about that Here

 

Tldr; the video got a lot of things wrong.


CPU: Intel i7 6700k  | Motherboard: Gigabyte Z170x Gaming 5 | RAM: 2x8GB 3000MHz G.Skill Ripjaws 5 | GPU: Gigabyte Aorus GTX 1080ti | PSU: Corsair RM750x (2018) | Case: BeQuiet SilentBase 800 | Cooler: Corsair H100i AIO | SSD: Samsung 970 Evo 500GB + Samsung 840 500GB | HDD: Seagate Ironwolf 8TB + 2x Seagate Ironwolf 6TB | Monitor: Acer Predator XB271HU + Samsung BX2450

Link to post
Share on other sites
Posted · Original PosterOP

First of, fantasticly detailed description, thank you very much for that! The reason all of this interest me very much, is I'd like to build my own low power consumption server for NAS and Plex purposes. I'm in the process of learning about both software and hardware, that'll give me the best efficiency, security and longevity.

 

In your response from the linked post, you go in to details about the testing methods and mention Cybenetics as a good reference for tests. Do you know of a similar reference for the motherboards you're suspecting is the difference in the tests? I've never given them a second thought as a power consuming component, before you mentioned it. Now that you have, it is obvious that's it's an oversight on my part, of course that would consume power aswell.

Link to post
Share on other sites
16 minutes ago, Nicolai Mortensen said:

The reason all of this interest me very much, is I'd like to build my own low power consumption server for NAS and Plex purposes.

Just buy a regular PSU and call it a day. 80+ Gold might be worth it, but I doubt anything higher than that would recoup its upfront cost through efficiency savings over the life of the product, especially for something that is going to be idling majority of the time like a personal Plex Server. What I mean is if an 80+ Platinum PSU costs $20 more than an 80+ Gold unit upfront but it only saves you $2/year on your electricity bill it would take 10 years to make back that higher upfront cost.

 

ATX12VO PSUs or motherboards aren't available at the moment, and it's expected they'll be predominately OEM/SI focused anyway. If they do become available for the DIY market then I would guess selection would be rather limited.

 

49 minutes ago, Nicolai Mortensen said:

In your response from the linked post, you go in to details about the testing methods and mention Cybenetics as a good reference for tests. Do you know of a similar reference for the motherboards you're suspecting is the difference in the tests?

No, because there's never really been a need to test power efficiency of motherboards.

How much power a system (via motherboard) will draw will vary depending on the settings in the BIOS and what features it has, such as MCE (Multi-Core Enhancement), and of course whatever hardware (CPU, graphics card, peripherals, etc) plugged in to it. Different motherboards will have different power draws, especially boards with different chipsets (hence why X570 has a fan to cool the chipset and B450 does not), but the difference should be fairly minimal.

 

In regards to the LTT test the motherboard might have just been one change between the two systems that attributed to the difference in power draw shown between the two tests. The ATX12VO motherboard might have been more aggressively tuned from the maker for low power consumption while the other board they used might have aggressive default or overclocking profiles (eg. MCE) for "GAMING" that cause the CPU to draw more power. One board might default CPU core voltage to 1.15V while the other defaults to 1.18V. It could have also been power plan settings in Windows, the first board might have been set to run at 100% clock speed all the time while the other board was on the power saver plan that allows the CPU clock speed to drop to 20% during idle, drastically reducing power consumption. There could have been programs running in the background on the first test which resulted in higher power consumption...

There's any number of reasons that could have resulted in a different power load from the system that would be responsible for the different power draw from the wall.

Hooking up a PSU to a system and measuring how much power it's drawing from a wall is NOT a reliable way to test the efficiency of a power supply, especially if you're not monitoring the output (system power draw) and especially if you're changing hardware between the two tests. For reliable results you need to hook the PSU up to testing equipment that can generate a constant and known load that can be repeated.


CPU: Intel i7 6700k  | Motherboard: Gigabyte Z170x Gaming 5 | RAM: 2x8GB 3000MHz G.Skill Ripjaws 5 | GPU: Gigabyte Aorus GTX 1080ti | PSU: Corsair RM750x (2018) | Case: BeQuiet SilentBase 800 | Cooler: Corsair H100i AIO | SSD: Samsung 970 Evo 500GB + Samsung 840 500GB | HDD: Seagate Ironwolf 8TB + 2x Seagate Ironwolf 6TB | Monitor: Acer Predator XB271HU + Samsung BX2450

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×