GN: Intel to push ATX12VO hard for Alder Lake

DanNeely

Supreme [H]ardness
Joined
Aug 26, 2005
Messages
4,473
Just a small segment in a larger video, but they're reporting that unlike with the last generation both PSU and mobo makers are onboard with supporting the new standard. Hopefully they'll still have a good selection of traditional 24-pin ATX PSU mobos for a few more years; but Steve had nothing to say about it either way.

 
BTX

Although Mobo makers will like the ability to raise prices significantly while PSU makers don't drop prices even though PSU's become simpler.
BTX was more or less DOA because it was focused around even hotter CPUs that never happened. 12VO is an attempt to standardize what the big OEMs have all started doing in proprietary fashion several years ago. Most of it probably should've been done 10-15 years ago after Intel and AMD both switched to run the CPU everything off of 12V instead of 3.3V.

With the legacy voltages being mostly unused since the mid 2000's, there's no reason we couldn't've had a 12/14 pin main connector that removed all but 1 or 2 of the 3.3/5v lines and the corresponding grounds (along with the -12V only needed for RS232). If that had happened 12VO would have been a relatively minor update. It looks dramatic now because we didn't do the intermediate step of removing most of the long unneeded legacy clutter in the 24 pin never happened when it should have.

If it wasn't for the OEMs having jumped the gun (and especially if an intermediate 24 pin diet had happened), I'd've probably suggested holding off on pushing it on a larger scale for another 2 or 3 years just to give SATA a bit more time to finish fading away from mainstream systems. Unless part of it is Intel scheming that making adding an HDD more expensive will encourage OEMs to make faster SSD only systems I'm not really sure why they're apparently pushing it hard now. This is also why I'm really curious if Intel is just pushing mobo makers to have a significant number of 12VO models; or to be 12VO almost exclusively.
 
I think we're in for quite a few years of higher end modular PSUs just including a 12VO harness with their existing designs.

Dropping ATX24 is a good thing. Reduces overall complexity and waste of components. Loss of SATA power from the PSU is annoying for some, but in a few years even SATA data ports are going to start to become rare (as in 2 ports, unlike today's sixish).
 
I think we're in for quite a few years of higher end modular PSUs just including a 12VO harness with their existing designs.

Dropping ATX24 is a good thing. Reduces overall complexity and waste of components. Loss of SATA power from the PSU is annoying for some, but in a few years even SATA data ports are going to start to become rare (as in 2 ports, unlike today's sixish).


It is good to move forward, but this seems to be a poorly thought out standard. The drawbacks are serious though, and will have issues down the road. The first thing is you will have all the voltage regulators on the board instead of in the psu. The big issue will be that this will cause the price of motherboards to go up, and if one of them fail you have to replace the motherboard instead of the PSU. That is the same reason we have socketed cpus and not soldered on ones, easier to replace one part then the whole thing. Plus boards will have to be bigger to fit all this new stuff onboard along with running hotter. Will it really be good to have all this stuff next to the cpu poaring more heat into the case. Not to mention if you are working in your case and accidentally drop a screw into your voltage regulators while it is running.....

Next issue is going to be with SATA drives, they will all have to get both power and data from the board now, so you will have to pick a board that has as many power connectors you need ahead of time. You wont be able to get a bigger PSU down the road when you need to add more storage. Even if you add a SATA or raid card, drive power still needs to come from the board, no more buying a big PSU to run lots of drives in a storage drive system. There arent enough lanes in most systems to run 4 or 5 NVME drives on top of them being nowhere as big as spinning rust for storage, not to mention the price difference between flash and hard drives.

Last issue is with the power efficiency of the new standard, yeah it is great at idle. They are mostly just moving the voltage reg to the board instead of the PSU so you gain on the PSU side and loose on the mobo side. They are just moving the issue from one place to another place. Plus how many people with gaming systems actually sit at idle all day, most are playing games or farming crypto...so no real power savings then. Plus the heavy users are going to have to worry about keeping those voltage regulators on the board cool, before they were cooled by the PSU fan, now they arent. Welcome back to crappy small fans on your board keeping this stuff cool. or just running at high temps dumping heat into your system.

It seems like a poor trade for most people, for not much advantage Seems more like a way to sell more PSUSs and more expensive motherboards. If they cut corners on your board and you have power issues, no way to fix this except buying a new board. Seems like it will just add to our already big issue with ewaste, with more mobos and PSUs going into the trash more often.
 
BTX was more or less DOA because it was focused around even hotter CPUs that never happened. 12VO is an attempt to standardize what the big OEMs have all started doing in proprietary fashion several years ago. Most of it probably should've been done 10-15 years ago after Intel and AMD both switched to run the CPU everything off of 12V instead of 3.3V.

With the legacy voltages being mostly unused since the mid 2000's, there's no reason we couldn't've had a 12/14 pin main connector that removed all but 1 or 2 of the 3.3/5v lines and the corresponding grounds (along with the -12V only needed for RS232). If that had happened 12VO would have been a relatively minor update. It looks dramatic now because we didn't do the intermediate step of removing most of the long unneeded legacy clutter in the 24 pin never happened when it should have.

If it wasn't for the OEMs having jumped the gun (and especially if an intermediate 24 pin diet had happened), I'd've probably suggested holding off on pushing it on a larger scale for another 2 or 3 years just to give SATA a bit more time to finish fading away from mainstream systems. Unless part of it is Intel scheming that making adding an HDD more expensive will encourage OEMs to make faster SSD only systems I'm not really sure why they're apparently pushing it hard now. This is also why I'm really curious if Intel is just pushing mobo makers to have a significant number of 12VO models; or to be 12VO almost exclusively.
Mostly being pushed now to stay ahead of various EU and California regulations regarding idle power consumption which is why OEM’s have started doing their own versions of 12VO. But that also means that Intel has to in some way support all those proprietary 12VO specs from all the vendors. One of Intel’s big points with OEM’s is they offer a complete package top to bottom with clearly defined specifications (not clear SKU’s) so having a specified and defined 12VO keeps with that bundle and takes something off the OEM’s plate.
 
What annoys with me this is, it adds to the complexity of motherboards, will likely make them die sooner (not that most don't get good use of their mobos), and as someone else said, is not likely to reduce the cost of power supplies. So in other words, we're all going to pay more for motherboards that are likely to not last as long. All to save .5% more electricity by doing the 12v to 5v/3.3v closer to the components on the motherboard.
 
What annoys with me this is, it adds to the complexity of motherboards, will likely make them die sooner (not that most don't get good use of their mobos), and as someone else said, is not likely to reduce the cost of power supplies. So in other words, we're all going to pay more for motherboards that are likely to not last as long. All to save .5% more electricity by doing the 12v to 5v/3.3v closer to the components on the motherboard.
It may or may not add complexity, 99% of MoBo’s are sold by OEM’s and never really changed or upgraded. They simply may not have those components or if they do just a single mini header which is a pretty standard part in the mobile space which is the bulk of PC sales to begin with. What this does is simply the OEM supply chain and give insensitive to phase out legacy consumer parts and interfaces.
Most OEM’s at this stage for the “cheap” alternative to SATA based SSD’s or HDD’s are using drives on the mSATA interface instead of the M.2 because it costs less for them to implement and ship which brings costs down over all. The OEM space is more competitive now than ever and in the end this gives more options, it’s not like this just makes the standard 24v AXT standard disappear, but it does pave the way for it to be phased out.
 
I think we're in for quite a few years of higher end modular PSUs just including a 12VO harness with their existing designs.

Dropping ATX24 is a good thing. Reduces overall complexity and waste of components. Loss of SATA power from the PSU is annoying for some, but in a few years even SATA data ports are going to start to become rare (as in 2 ports, unlike today's sixish).

There probably will be adapter cables made; both for existing modular and non-modular PSUs. They won't quite be dumb 1:1 adapters though. 12VO has 12v standby power not 5v, so they'll need a small DC-DC converter to change that. 12VO also has 50% more 12V power than the old 24pin connector; so you'll probably need to plug a molex into the adapter too, not just connect directly to the legacy plug.
 
It is good to move forward, but this seems to be a poorly thought out standard. The drawbacks are serious though, and will have issues down the road. The first thing is you will have all the voltage regulators on the board instead of in the psu. The big issue will be that this will cause the price of motherboards to go up, and if one of them fail you have to replace the motherboard instead of the PSU. That is the same reason we have socketed cpus and not soldered on ones, easier to replace one part then the whole thing. Plus boards will have to be bigger to fit all this new stuff onboard along with running hotter. Will it really be good to have all this stuff next to the cpu poaring more heat into the case. Not to mention if you are working in your case and accidentally drop a screw into your voltage regulators while it is running.....
You're really overstating the issue with moving the voltage regulators increasing the failure risk on the mobo. Mobo's are full of voltage regulators already, virtually everything on them runs at voltages lower than the 12/5/3.3v the PSU delivers; and is run off of 12 to 1.Xv regulators already. If mobo makers using substandard parts there was a plausible kill pattern we'd've been seeing it on everything from the last 20 years.
Next issue is going to be with SATA drives, they will all have to get both power and data from the board now, so you will have to pick a board that has as many power connectors you need ahead of time. You wont be able to get a bigger PSU down the road when you need to add more storage. Even if you add a SATA or raid card, drive power still needs to come from the board, no more buying a big PSU to run lots of drives in a storage drive system. There arent enough lanes in most systems to run 4 or 5 NVME drives on top of them being nowhere as big as spinning rust for storage, not to mention the price difference between flash and hard drives.
Regarding sata drives, the mobo end power connector for them is a 4pin similar to the 12V CPU one but more compact that supports a string of several sata/molex connectors. Even if they delete a sata plugs to make room for the power ones, you'd end up with 4 or 6 instead of 6 or 8; not half as few.

As for big storage boxes with PCIe raid cards, future generations of them will almost certainly just move the 3.3/5v power generation onto the card itself and provide power as well as data connections; at which point it will be back to what your PSU can provide again. Assuming of course that future sata drives aren't just redesigned to run off 12V only to bypass the need for 3.3/5v from the mobo all together. 12VO does include cables with the old 4pin molex connector and only the 12V and 1 ground pin connected; so those can run directly from the PSU.

Alternately, cases that have a crapton of HDD slots will just include a power conversion module to supply the lower voltages the way they already do on most servers; which have been proprietary 12V for years. Aftermarket modules will probably become available for people who want to keep using their old cases.

Last issue is with the power efficiency of the new standard, yeah it is great at idle. They are mostly just moving the voltage reg to the board instead of the PSU so you gain on the PSU side and loose on the mobo side. They are just moving the issue from one place to another place. Plus how many people with gaming systems actually sit at idle all day, most are playing games or farming crypto...so no real power savings then. Plus the heavy users are going to have to worry about keeping those voltage regulators on the board cool, before they were cooled by the PSU fan, now they arent. Welcome back to crappy small fans on your board keeping this stuff cool. or just running at high temps dumping heat into your system.
The efficiency gains from moving 3.3/5v come from reducing cable loss. 12V from the PSU to the mobo will lose less over the same distance than 3.3/5v because resistive losses are I^2 * R, and for a given power amount the current will be less at higher voltages because of the lower current. Then the 3.3/5v only needs to go over short wires from the front of the mobo to the drive cages.

And again, mobos are already full of voltage regulators; the makers know how to keep them properly cooled and healthy. That's a total non-issue.
It seems like a poor trade for most people, for not much advantage Seems more like a way to sell more PSUSs and more expensive motherboards. If they cut corners on your board and you have power issues, no way to fix this except buying a new board. Seems like it will just add to our already big issue with ewaste, with more mobos and PSUs going into the trash more often.

Most people today use laptops. Most people who use desktops use whatever Dell/etc sells them with at most 1 or 2 sata devices. Being able to cut costs by not implementing any 3.3/5v out will probably encourage OEMs to move away from offering SATA at all on more systems.

Those of us who build our own systems are a very small minority of weirdos these days. Common parts haven't been optimized for our use cases for many years.
 
Most people today use laptops. Most people who use desktops use whatever Dell/etc sells them with at most 1 or 2 sata devices. Being able to cut costs by not implementing any 3.3/5v out will probably encourage OEMs to move away from offering SATA at all on more systems.
Dell already has started phasing out the traditional SASA drives, the director of HR needed the drive in her new desktop expanded as the 256GB SSD it shipped with us now too small. So I brought in a nice Samsung to replace it with. Only to crack the case open and find this little bugger instead. Fortunately a quick call to Dell confirmed it’s M.2 compatible so just drop one of those in and it’s G2G.
 

Attachments

  • 5993D0A2-28B8-4ADD-877A-0E18D121D926.jpeg
    5993D0A2-28B8-4ADD-877A-0E18D121D926.jpeg
    588.1 KB · Views: 0
Dell already has started phasing out the traditional SASA drives, the director of HR needed the drive in her new desktop expanded as the 256GB SSD it shipped with us now too small. So I brought in a nice Samsung to replace it with. Only to crack the case open and find this little bugger instead. Fortunately a quick call to Dell confirmed it’s M.2 compatible so just drop one of those in and it’s G2G.
2232 formfactor? First time I've seen an SSD in that size (vs a wifi adapter). The dual keying I think means it was m.2 sata, which has largely gone away.

And yeah, in the smallest/cheapest systems they're all SSD only; at and that point there's no reason not to use an m.2 instead. I was thinking about the gaming systems/etc they like to stick a 256/512GB SSD and 1/2TB HDD in. Nudging those to pure flash would be a very good thing. The price margin between a small SSD and 1TB hdd vs a cheap 1tb SSD is getting steadily smaller; and the bigger SSD is a much better user experience.
 
2232 formfactor? First time I've seen an SSD in that size (vs a wifi adapter). The dual keying I think means it was m.2 sata, which has largely gone away.

And yeah, in the smallest/cheapest systems they're all SSD only; at and that point there's no reason not to use an m.2 instead. I was thinking about the gaming systems/etc they like to stick a 256/512GB SSD and 1/2TB HDD in. Nudging those to pure flash would be a very good thing. The price margin between a small SSD and 1TB hdd vs a cheap 1tb SSD is getting steadily smaller; and the bigger SSD is a much better user experience.
Yeah but cracking that case open, it is already using a 12VO PSU which is tiny by the way, and reviving the SATA M.2 Key is probably cheaper and smaller than using the traditional SATA ports, fewer cables, fewer screws, faster to install, don't need to add the various drive cages to the case, and the board is specific and proprietary anyways, I should have taken pictures while I was in there, but it was a surprisingly sparse board very little going on. I never did any speed tests or anything but I would have to imagine that the m.2 SATA SSD in there can't be any slower than a cheap SSD.
 
2232 formfactor? First time I've seen an SSD in that size (vs a wifi adapter). The dual keying I think means it was m.2 sata, which has largely gone away.
yup and yup.
2232 is common in chromebooks, havent seen them in normal laptops though.
 
yup and yup.
2232 is common in chromebooks, havent seen them in normal laptops though.
And I pulled that out of a new (late 2020) Optiplex. My 2019 G5 shipped with one though a 512, quickly swapped that over into the secondary port, and replaced it with a Samsung 980 MZ. So I suspect that DELL at least is taking it more mainstream.
 
Yeah but cracking that case open, it is already using a 12VO PSU which is tiny by the way, and reviving the SATA M.2 Key is probably cheaper and smaller than using the traditional SATA ports, fewer cables, fewer screws, faster to install, don't need to add the various drive cages to the case, and the board is specific and proprietary anyways, I should have taken pictures while I was in there, but it was a surprisingly sparse board very little going on. I never did any speed tests or anything but I would have to imagine that the m.2 SATA SSD in there can't be any slower than a cheap SSD.
Yeah it's cheaper than an 2.5" sata. At this point though the price gap between m.2 sata and bottom end dramless NVME m.2 drives is more or less zero (and even the crappiest of these are still generally at least a bit faster than sata). That's why m.2 sata is fading away; not because 2.5" sata is having a revival.
 
You're really overstating the issue with moving the voltage regulators increasing the failure risk on the mobo. Mobo's are full of voltage regulators already, virtually everything on them runs at voltages lower than the 12/5/3.3v the PSU delivers; and is run off of 12 to 1.Xv regulators already. If mobo makers using substandard parts there was a plausible kill pattern we'd've been seeing it on everything from the last 20 years.

And most of those VRMs that provide low voltage rails are only providing milliamps or microamps of current, not high powered devices like the CPU, RAM and PCH or NB/SB. Often times linear regulators are used because the power draw is so low, and using a buck regulator for a rail would take up too much room and cost too much.

If you wanted to replicate the 5v rail from an ATX power supply on a 12VO motherboard and keep the same capacity, just take the entire CPU power regulation area and duplicate it somewhere else on the board. I'd probably add $20-30 to the BOM for just the 5v rail and more for the 3.3v rail.

If mobo makers using substandard parts there was a plausible kill pattern we'd've been seeing it on everything from the last 20 years.

There is a kill pattern, you're just not looking at the right space. You need to be looking at laptops, which have been dying en masse for decades due to VRM failures. Apple is probably the worst about this with their shit designs that run all parts of the machine at temperatures you'd expect from an easybake oven.

Motherboard manufacturers need to be motherboard manufacturers, they don't need to be making power supplies. 12VO is a terrible idea that will end badly, just like it would be for swapping any other two professionals into completely unrelated fields and expecting a good outcome.
 
Sucks for the 12 people who are still buying Intel.
I mean you can rip on them but OEM’s have been doing this since 2019. Intel is just standardizing it to make life easier, 24v ATX just doesn’t meat the more stringent standby mode regulations required to meet the various carbon neutral programs out there. And if you are building a system devoid of legacy connectors then 12VO is actually pretty decent.
 
I mean you can rip on them but OEM’s have been doing this since 2019. Intel is just standardizing it to make life easier, 24v ATX just doesn’t meat the more stringent standby mode regulations required to meet the various carbon neutral programs out there. And if you are building a system devoid of legacy connectors then 12VO is actually pretty decent.
OEMs did it to ensure you'd buy the extended warranty because you'd be forced to buy their proprietary bullshit if you didn't.
 
Sucks for the 12 people who are still buying Intel.
Just remember that Intel also created the current ATX12V spec that everybody seems to love. Regulations on idle power use in PCs are coming down the pike and everybody else refused to address them but Intel. ATX12VO will eventually replace that 26 year-old standard.
 
And most of those VRMs that provide low voltage rails are only providing milliamps or microamps of current, not high powered devices like the CPU, RAM and PCH or NB/SB. Often times linear regulators are used because the power draw is so low, and using a buck regulator for a rail would take up too much room and cost too much.

If you wanted to replicate the 5v rail from an ATX power supply on a 12VO motherboard and keep the same capacity, just take the entire CPU power regulation area and duplicate it somewhere else on the board. I'd probably add $20-30 to the BOM for just the 5v rail and more for the 3.3v rail.

I dont understand this quote. In direct opposition to your sentence: CPU, RAM, and PCH are already regulated on the motherboard from 12v down to their respective voltages on any board in the last 10 years. In the very next paragraph you use the CPU power regulation as an example of what they would need to do for 5V. So it seems you already know this. Can you clarify what you meant by "And most of those VRMs that provide low voltage rails are only providing milliamps or microamps of current, not high powered devices like the CPU, RAM and PCH or NB/SB. "
 
OEMs did it to ensure you'd buy the extended warranty because you'd be forced to buy their proprietary bullshit if you didn't.
No, not meeting my CGRT mandated Carbon Neutral usage points costs me fines. 12VO system keeps me in the clear with out having to constantly remind staff to turn the computers off when they are not in use. 24V didn’t, and we quite literally had somebody whose job it was at the end of day to go around and ensure all computers were powered down. Idle power consumption is a big deal for some of us, and proposed changes have those fines getting bigger, which puts me in a tough spot the new Intel’s are too power hungry and the AMD’s aren’t available from OEM. I’m seriously looking at M1’s for the electrical savings.
 
No, not meeting my CGRT mandated Carbon Neutral usage points costs me fines. 12VO system keeps me in the clear with out having to constantly remind staff to turn the computers off when they are not in use. 24V didn’t, and we quite literally had somebody whose job it was at the end of day to go around and ensure all computers were powered down. Idle power consumption is a big deal for some of us, and proposed changes have those fines getting bigger, which puts me in a tough spot the new Intel’s are too power hungry and the AMD’s aren’t available from OEM. I’m seriously looking at M1’s for the electrical savings.
You live in the EU don't you?

BC maybe?
 
I dont understand this quote. In direct opposition to your sentence: CPU, RAM, and PCH are already regulated on the motherboard from 12v down to their respective voltages on any board in the last 10 years.

Reading comprehension is important, you missed "not", where I'm saying that comparing low voltage, low current rails to high current rails like they're equal is wrong. DanNeely's post can basically be summed up as "because motherboards already have all of these low voltage and low powered rails, it's OK to add multiple high powered ones as well."

Can you clarify what you meant by "And most of those VRMs that provide low voltage rails are only providing milliamps or microamps of current, not high powered devices like the CPU, RAM and PCH or NB/SB. "

Not really sure what you need clarification on. The other parts of the board DanNeely was talking about, like the Ethernet PHY and audio chip don't use tons of power, so it's easy to create power rails for them. Many PHYs for example also need 2.5 and 1.8v, and those can be created with single chip linear regulators because they barely use any current. In fact, you WANT to use linear regulators in the case of audio, a buck regulator is an RF blaster and would make any audio sound like TV snow without tons of additional filtering.

Providing a 100W 5v rail on the other hand would require something on the lines of the complexity of the CPU power regulation circuitry. And running 12v through the motherboard to other devices like fans is just begging for additional problems. All of this additional circuitry and copper is going to significantly drive up the cost of motherboards and drastically increase their size. There are already cases of people blowing out fan headers on motherboards from fans that draw too much current.

I've done board level repair for decades, and I've seen the carnage that on-board voltage regulation and faulty power bus bars causes. It's not something that is good for the consumer and just promotes further eWaste. Many motherboard makers can't even get CPU power regulation correct, I definitely in no way would ever trust them with additional high power rails.
 
This standard will be annoying on SFF ITX builds. ATX doesn't bother me so much as I don't use SATA anymore, though I know it is still pretty common.

Not sure how long the transition will take on this, or if it will be successful.
 
This standard will be annoying on SFF ITX builds. ATX doesn't bother me so much as I don't use SATA anymore, though I know it is still pretty common.

Not sure how long the transition will take on this, or if it will be successful.
SFX will still have the other voltages, at least until they update the spec. Or you can get a separate 3.3/5V psu that'll take up a few cubic inches in the case.
 
Reading comprehension is important, you missed "not", where I'm saying that comparing low voltage, low current rails to high current rails like they're equal is wrong. DanNeely's post can basically be summed up as "because motherboards already have all of these low voltage and low powered rails, it's OK to add multiple high powered ones as well."



Not really sure what you need clarification on. The other parts of the board DanNeely was talking about, like the Ethernet PHY and audio chip don't use tons of power, so it's easy to create power rails for them. Many PHYs for example also need 2.5 and 1.8v, and those can be created with single chip linear regulators because they barely use any current. In fact, you WANT to use linear regulators in the case of audio, a buck regulator is an RF blaster and would make any audio sound like TV snow without tons of additional filtering.

Providing a 100W 5v rail on the other hand would require something on the lines of the complexity of the CPU power regulation circuitry. And running 12v through the motherboard to other devices like fans is just begging for additional problems. All of this additional circuitry and copper is going to significantly drive up the cost of motherboards and drastically increase their size. There are already cases of people blowing out fan headers on motherboards from fans that draw too much current.

I've done board level repair for decades, and I've seen the carnage that on-board voltage regulation and faulty power bus bars causes. It's not something that is good for the consumer and just promotes further eWaste. Many motherboard makers can't even get CPU power regulation correct, I definitely in no way would ever trust them with additional high power rails.
To be clear, I'm not attacking you, I was sincerely asking you to clarify something.

This topic reminds me of https://en.wikipedia.org/wiki/Law_of_triviality

Its interesting how enthusiastic people get about it (me included!), but in the end, as far as I can tell, it doesn't really matter all that much compared to other hardware issues. The perfect subject for a community forum :p

Hardware manufacturers (dell, hp, etc) already do strange proprietary methods of 12VO. I think it would be nice if they standardized. ATX12V 24pin can continue for all i care, although I cringe every time I see those 24pins and know inside that I use about half of them to their actual capabilities (and 1 of them NOT AT ALL!).


DanNeely didn't mention anything low power in the post you quoted, so this was confusing to me.

The buck converter we use on our 12 drive SAS/SATA backplane is 12A (20A MAX). It comes in under $2 at high enough quantities. Add in all the inductors and capacitors around that, it's probably going to be under $10. Many motherboards don't need 12 drives. So I think your $20-$30 estimate is too high.
Regardless this cost would presumably (although maybe slowly?) be subtracted from the PSU which no longer needs to do this job.

Motherboards already do run +12V to fans, so I think there is no difference here.

If you can't trust a motherboard maker to make CPU regulators then I think you are simply buying the wrong motherboards. Cheap motherboards will have cheap parts no matter what you do. It's the same with cheap PSUs.
 
To be clear, I'm not attacking you, I was sincerely asking you to clarify something.

This topic reminds me of https://en.wikipedia.org/wiki/Law_of_triviality

Its interesting how enthusiastic people get about it (me included!), but in the end, as far as I can tell, it doesn't really matter all that much compared to other hardware issues. The perfect subject for a community forum :p

Hardware manufacturers (dell, hp, etc) already do strange proprietary methods of 12VO. I think it would be nice if they standardized. ATX12V 24pin can continue for all i care, although I cringe every time I see those 24pins and know inside that I use about half of them to their actual capabilities (and 1 of them NOT AT ALL!).


DanNeely didn't mention anything low power in the post you quoted, so this was confusing to me.

The buck converter we use on our 12 drive SAS/SATA backplane is 12A (20A MAX). It comes in under $2 at high enough quantities. Add in all the inductors and capacitors around that, it's probably going to be under $10. Many motherboards don't need 12 drives. So I think your $20-$30 estimate is too high.
Regardless this cost would presumably (although maybe slowly?) be subtracted from the PSU which no longer needs to do this job.

Motherboards already do run +12V to fans, so I think there is no difference here.

If you can't trust a motherboard maker to make CPU regulators then I think you are simply buying the wrong motherboards. Cheap motherboards will have cheap parts no matter what you do. It's the same with cheap PSUs.
With PCIE5 right around the corner, the 3.5" and 5.25" form factor drives going up in price, and NVME drives going down, I think we will soon see a point where outside specialized tasks the overwhelming majority of machines built will just need a small subset of M.2 ports installed on the backside of the MB. The biggest thing I see missing from the spec is around the CPU for increased power supply for cooling options as most bigger options require either a SATA or Molex connector to power them.
 
This standard will be annoying on SFF ITX builds. ATX doesn't bother me so much as I don't use SATA anymore, though I know it is still pretty common.

Not sure how long the transition will take on this, or if it will be successful.

Unless your SFF build needs to use SATA drives, I don't see how it will be more annoying.
 
Probably legacy ones, most ITX boards just put additional M.2 keys on the back. Plenty of space there.
I suspect mITX will be leading the charge to delete SATA support as the legacy feature to kill. Not 100%, there probably will be at least a few boards with 4x sata for mini DIY NASes; but cutting the pair of SATAs most common on the form factor today and the new mini power plug that would be needed to support them with 12VO will let them be able to size the 3.3/5v hardware to its minimum level. 3.3v sized to the PCIe slot and number of m.2 slots included. 5V to the number of USB ports. No need to worry about whatever other stuff might be stuffed into the box.
 
Back
Top