And seeing as CPU/GPU/RAM consume the lion's share of the power in a typical desktop computer (the exception is people with lots of 3.5" HDDs... and those run their motors off 12V), double-conversion makes the whole system less efficient.
48V is a win only when you are supplying many kW of 48 V to an entire rack of computers at once, from ganged AC-DC converters.
I dont think you could convert 48V directly to 0.8-1.5V with acceptable performance/efficiency. So using 48V would require all separate components in your PC use two-staged conversion, adding cost, complexity and heat dissipation on boards.
The conversion from 170 DC needs to use a transformer for safety isolation anyway, so you can freely choose PSU output voltage without much impact on efficiency of the PSU.
Where is 48V regulation stricter? It has been a de-facto standard worldwide for over a century because it's at the limit of what regulations typically consider low-voltage.
Yeah, as is most legacy telecom equipment, because 48V was readily available and easy to be compliant with in telecom facilities, which then spread to other applications that either interoperated with telecom equipment or that used multi-purpose telecom equipment and components for other uses, like general audio equipment. It's not an accident that my audio interface has a 48V power-supply built-in.
its really annyone to try to run some thick damn cables in small cases. so we can just use like 2-4-6 pin connectors instead of a million pin connector x 3
please do some very basic research.
you just described a connector and cable choice problem and not a 12v vs 48v problem.
your graphics and cpu power could be a single xt90 or xt120 power connector with just 2 cables per connector, which would be a bit thicker, but not much and thus be much easier to manage. much much easier actually.
any proper high power connector uses 2 connections (one source + 1 ground)
so if you want simple cable management there is your solution.
it has NOTHING AT ALL to do with 12v or 48v.
and cpu + graphics card power cables are already purely 12v connections anyways.
so again you are asking for the completely wrong thing, because you don't understand the actual solution, that would make your life easier, while also NOT breaking all compatibility with all hardware ever made for pc as it all depends on 12v pretty much.
That would be better than using something like the 12V High Failure Rate connector.
the nvidia 12 pin fire hazard has nothing to do with what voltage gets used.
it has 0 safety margins, is extremely fragile, is designed to be a melting and fire risk at low and high power.
and it crucially is NOT smaller than other power connectors. xt90 or xt120 connectors are about the same size and actually safely carry 720watts (xt120 at 12v 60 amps).
this is nothing new. nvidia decided to push a known fire hazard, instead of going with safe connectors like xt90 or xt120.
safety margins for the nvidia 12 pin fire hazard were considered a big reason for the melting, but it is AMONG the reasons.
as got shown by 9070 xt cards and 5070 cards and others melting at vastly lower max power.
that is why increasing the voltage at the same power and thus dropping amps a lot can NOT fix an inherent fire hazard.
it could fix a missing safety margin, but that is NOT the only issue, but again one in a long list of reasons the nvidia 12 pin fire hazard melts and catches on fire.
Also have you ever tried to unplug an xt90 or xt120? Lol
are you saying, that safe reliable tightly fitting connectors are a problem now?
a connector, that the person in the video has no problem pulling apart in one go?
like wtf.
and as another reminder, IF the insane move would get made to jump to 48v, then the connectors, that you'd wanna use would still be xt90/xt120 style connectors with 2 big contacts and massive surfaces areas REGARDLESS, because that is just how you design proper connectors, although maybe you go down to an xt60 connector with guess what the same design just scaled down..
oh so you want to break ALL COMPATIBILITY for all hardware ever made in the past? interesting...
your graphics card? nah that won't work anymore. your storage? nope.
but hey at least you can get 0 benefit from the change, because we're talking about a few cents of copper, so there is that right?
and there are no "huge currents" in desktop systems.
even if you have a 720 watt graphics card, that could all be powered by a single small xt120 power connector with proper safety margins and LESS cables than today, because it would be 2 thicker cables, rather than lots of smaller cables.
there is no problem with 12v and i am sick of people being so willing to set EVERY SINGLE DEVICE, that uses 12v on fire, because they feel like, that 48v is a nicer number.
Sort of an aside: this makes me wonder how USB-PD works when going through a motherboard. Does the USB controller for PD-capable ports get fed 12V and then buck/boost it to the requested voltage? Or does it only support 5 and 12v modes and just switches which rail gets connected to the port?
For the most part I just pretend all motherboard ports are still only capable of 5v 500mA and live in blissful ignorance.
Yeah there's been a big push for 48V in servers i think. Weird how desktops are still stuck at only 12v, laptops use 20v. But the inertia with 12v is so strong i doubt it will change.
I feel from the very introduction of ATX12VO in like 2019(?) most consumers saw it for what it was - an effort to shift costs from PSU makers to motherboard makers, since that's where all the conversion to lower voltages will now have to take place. Might make sense for OEMs though, but that depends on everyone aside from California tightening efficiency standards enough that OEMs consider it worth the cost of transition.
IMO it makes sense. Basically nothing uses 5v or 3.3v anyway so for most systems the cost is just removed entirely.
The problem is that Intel half-assed the standard. To do it properly, they needed to get a whole new standard that got rid of the legacy 3.3v and 5v connections on motherboards (e.g. SATA and PCIE both have 3.3 and 5v pins that no one really uses (and ideally would have gotten rid of on board pciex16 since GPUs hanging off motherboards is just a really bad idea).
5 V and 3.3 V are still very much used. Notably, USB needs 5 V and M.2 SSDs run on 3.3 V. A lot of the small chips on the board doing various things (e.g. on board audio) also usually don't run on 12 V.
I still see a few advantages, though. Having the voltage conversion on board allows you to avoid way overbuilt rails for those voltages and cuts down on transmission distance, as transmitting power with such low voltage is really inefficient over longish cables.
One potential problem I foresee is cheaper boards having bugger all current capacity, at least without risking your motherboard's life. As it stands, they can just pass the PSU's power through to a handful of USB ports without worrying about you trying to run a webcam, audio interface, stream deck, capture card, and fancy USB lighting rig while also charging a few devices through a bus-powdered hub.
My worry is that they'll slap on a bargain basement voltage regulator that's effectively spec'd for only mouse/keyboard type power draw whereas I'd expect even a cheap PSU wouldn't have much issues supporting that mess reliably. I'm mostly thinking of the -DS3H and -HDV etc. class of boards being cost cut in this way, and it's not as if you can trust consumers to know (or care) what the technical limits of each port is.
I more-or-less am rooting for 12vo (even just to finally nix that ancient 24pin) but I've no doubt these companies would gladly neuter their products to save a buck in components and tell consumers, "Well you should have bought the $250+ model if you expected to actually be able to really use it!"
Any component that's directly connected to the motherboard with no power cables from the PSU can receive 3.3 and 5V from the motherboard in the 12VO standards, so it's irrelevant that m.2 uses 3.3V.
What? Do you think Intel would conspire to set up a whole new international standard just to save a few bucks on PSUs for their office PCs?
Besides, the original claim was that Intel was trying to shift costs from PSU makers to motherboard makers (for whatever reason), the overall system cost wouldn’t change that much. I just think that wasn’t their motivation at all, it just so happens that this was necessary to achieve higher efficiency.
What? Do you think Intel would conspire to set up a whole new international standard just to save a few bucks on PSUs for their office PCs?
Why do you keep making malicious assumptions? Stop putting words in my mouth.
Of course it's not the only reason. It is simply one of a list of reasons.
Sorry, but you are not making much sense. We are talking about whether Intel has any motivation to shift costs from PSU makers to motherboard makers.
How much Intel spends on their own ATX PCs is irrelevant.
If we are talking about 12VO in general, Intel wants to make the (Intel CPU powered) PC platform more efficient (and therefore more competitive). That motivation is obvious.
They just don’t care if it shifts costs from one place to another if the overall operating costs are lower.
Saving some money on their own PCs, while technically true, is about the least relevant motivation for this I could imagine, it is a negligible fraction of their expenses. And more importantly, it doesn’t relate to my original argument at all.
OEM have been using 12V-only PSUs for a long time on some systems, before ATX12VO ever launched, tbh.
OEMs have realised the cost & efficiency benefits of 12V-only PSUs, but as there was no ATX standard, they're all proprietary. It's why some $400 OEM PCs have a 80+ Gold or 80+ Platinum PSU out of nowhere.
IIRC, ATX12VO was (partially) to make a standardized ATX platform for OEMs. I'm not sure how many have switched or they're still using their proprietary platforms & designs. But the idea is with one standard, OEMs get lower costs, more competition, and a little more flexibility.
48V while keeping everything else the same would let you deliver 2400W at the same heat output we have today. That’s a machine that can pop 120V 20A and 230V 10A circuit breakers all by itself
I mean the EPS12V connector is known to be extremely robust and should easily be able to handle over 200W on a single 8-pin (can't remember the exact specs). Two 8-pins are more than enough to cover all use cases on a consumer platform, probably even overkill most of the time. There's no good reason to change it.
Seeing "optimized for cost and space" included in those slides sure doesn't inspire confidence when this is the exact mentality that led to the 12VHPWR fiasco. Maybe the connector specs are good enough on paper, but the smaller you make it, the more manufacturing tolerances start to matter.
I want a connector design that just works, damnit. Once everyone starts worrying whether not buying the exact right brand of the cable or looking at it the wrong way during installation will cause your stuff to melt down, you've already screwed up.
From a cursory look, this is a drastic departure even from existing ATX12VO v2. The pin distance is now 3mm also for the 8-pin CPU and 8-pin/12-pin mainboard connector. This means it's the same as the infamous 12V-2x6 GPU connector. Also they remove 12VSB and add this PMBUS communication. So it's completely incompatible to existing ATX12VO systems. Shows you how this is purely some OEM thing since they can just sell new systems.
There's a new 8-pin CPU and 8-pin/12-pin mainboard connector with a pin distance of 3mm now (from 4.2mm like EPS/PCIe 6/8-pin). It's the same as the infamous 12V-2x6 GPU connector.
oh, so they are bringing EXTREMELY FRAGILE possible fire risk pin sizes to mass oem adoption.
but i'm sure the criminals are getting away just fine with it, once the first people die in a house fire.
that is so disgusting.
are they trying to make it always a fire hazard to buy a prebuild or any computer in the future to push for more cloud computing subscriptions? 😃
and it is worth adding here, that this has nothing inherently to do with size, because those shits could have made very small and very safe connectors, that use 2 connections, instead of 8 then with big contacts and thicker cables, but why improve things, when you can just make them vastly worse right?
The previous versions of ATX12VO had a reserved "Telemetry Pin" to communicate total usage, but maybe PMBus is more widely adopted? PMBus cancommunicate a lot more than just total usage: input / output voltages, current, fan speeds, internal temps (3 sensors), fault reporting, etc.
Unfortunately this ATX12VO v3 spec makes the PMBus a tiered optional thing (you can have 1 pin, 2 pins, 3 pins or 4 pins for PMBus). So it's gonna be annoying comparing things depending on what you want when looking to buy a PSU.
Unfortunate. PMBus is usually pretty un-standardised, but it would've been neat for Intel to make PMBus + a standard consumer comms. protocol and standardized pins. Currently, some higher-end PSUs push out data via USB that HWiNFO64 can capture.
Intel ATX continues its tradition of, to put it lightly, "interesting" changes.
If California & EU regs cannot allow prebuilt 400w gaming PCs with ATX 3.1, due to idle power, the SI will have to stop using ATX parts that DIY uses. DIY market can't support production itself. If prebuild switches to a standard like atx12v v3, the DIY market can follow. But if SI do proprietary stuff like Dell does, the DIY market gets too expensive and dies. atx12vo might actually be very important, if you don't want to buy black box systems.
mduell@reddit
Does anything actually use 12V native? Rather than stepping down from there?
If we’re going to one voltage, let’s make it 48V and stop pushing huge currents.
VenditatioDelendaEst@reddit
12V is more efficiently converted to ~1V than 48V is. Server 48V systems are actually 48 -> 12 -> load or 48 -> 7-ish -> load on the board.
See the time I dug into this deeply and bookmarked it two or three "desktop PCs should use 48V power!"s ago.
mduell@reddit
Yea, for CPU/GPU/RAM I figured they’d have to double convert via something like 7V.
VenditatioDelendaEst@reddit
And seeing as CPU/GPU/RAM consume the lion's share of the power in a typical desktop computer (the exception is people with lots of 3.5" HDDs... and those run their motors off 12V), double-conversion makes the whole system less efficient.
48V is a win only when you are supplying many kW of 48 V to an entire rack of computers at once, from ganged AC-DC converters.
mduell@reddit
I hear you, but the consumer space has a problem today, for years, with GPU plugs melting…
asdfzzz2@reddit
I dont think you could convert 48V directly to 0.8-1.5V with acceptable performance/efficiency. So using 48V would require all separate components in your PC use two-staged conversion, adding cost, complexity and heat dissipation on boards.
UmaThurmish@reddit
you are already going down from 170 DC (max, depends on the load really) to 12 in the powersupply. so the efficiency from the wall should be the same
VenditatioDelendaEst@reddit
The conversion from 170 DC needs to use a transformer for safety isolation anyway, so you can freely choose PSU output voltage without much impact on efficiency of the PSU.
asdfzzz2@reddit
Wall > PSU 12V > 1V has one less conversion step compared to Wall > PSU 48V > 5V or 12V intermediate > 1V.
Every conversion step is extra losses, even if small and in easy-to-cool places.
Slasher1738@reddit
I mean, I think a PSU maker would just build it with 4x12V rails and let the modular cables get it to 48V where needed
UmaThurmish@reddit
yeah 24 or 48v would really simplify connector design and improve efficiency at the board level.
Moscato359@reddit
48v would actually be a problem where above 40v actually can cause regulatory problems with stricter restrictions and proofs of safety
But 24 or 40v is generally treated as easier to get certified
Berengal@reddit
Where is 48V regulation stricter? It has been a de-facto standard worldwide for over a century because it's at the limit of what regulations typically consider low-voltage.
dingo_xd@reddit
Aren't telephone lines 48V?
Berengal@reddit
Yeah, as is most legacy telecom equipment, because 48V was readily available and easy to be compliant with in telecom facilities, which then spread to other applications that either interoperated with telecom equipment or that used multi-purpose telecom equipment and components for other uses, like general audio equipment. It's not an accident that my audio interface has a 48V power-supply built-in.
reddit_equals_censor@reddit
please do some very basic research.
you just described a connector and cable choice problem and not a 12v vs 48v problem.
your graphics and cpu power could be a single xt90 or xt120 power connector with just 2 cables per connector, which would be a bit thicker, but not much and thus be much easier to manage. much much easier actually.
any proper high power connector uses 2 connections (one source + 1 ground)
so if you want simple cable management there is your solution.
it has NOTHING AT ALL to do with 12v or 48v.
and cpu + graphics card power cables are already purely 12v connections anyways.
so again you are asking for the completely wrong thing, because you don't understand the actual solution, that would make your life easier, while also NOT breaking all compatibility with all hardware ever made for pc as it all depends on 12v pretty much.
UmaThurmish@reddit
thats why I said we should have 24 or 48v.
UmaThurmish@reddit
you are already going down from 170 DC (max, depends on the load really) to 12 in the powersupply. so the efficiency from the wall should be the same
Nuck_Chorris_Stache@reddit
I don't think so. And also, there's probably nothing that uses 5V or 3.3V native either.
reddit_equals_censor@reddit
the nvidia 12 pin fire hazard has nothing to do with what voltage gets used.
it has 0 safety margins, is extremely fragile, is designed to be a melting and fire risk at low and high power.
and it crucially is NOT smaller than other power connectors. xt90 or xt120 connectors are about the same size and actually safely carry 720watts (xt120 at 12v 60 amps).
this is nothing new. nvidia decided to push a known fire hazard, instead of going with safe connectors like xt90 or xt120.
Moscato359@reddit
If gpus had 4x voltage they run at now, they would need 1/4 amperage
And the lack of margins are caused by too much amperage
reddit_equals_censor@reddit
safety margins for the nvidia 12 pin fire hazard were considered a big reason for the melting, but it is AMONG the reasons.
as got shown by 9070 xt cards and 5070 cards and others melting at vastly lower max power.
that is why increasing the voltage at the same power and thus dropping amps a lot can NOT fix an inherent fire hazard.
it could fix a missing safety margin, but that is NOT the only issue, but again one in a long list of reasons the nvidia 12 pin fire hazard melts and catches on fire.
Also have you ever tried to unplug an xt90 or xt120? Lol
https://youtu.be/4rcCdCVBXAI?si=VNruGSlKzLQ_5m_1&t=29
???
are you saying, that safe reliable tightly fitting connectors are a problem now?
a connector, that the person in the video has no problem pulling apart in one go?
like wtf.
and as another reminder, IF the insane move would get made to jump to 48v, then the connectors, that you'd wanna use would still be xt90/xt120 style connectors with 2 big contacts and massive surfaces areas REGARDLESS, because that is just how you design proper connectors, although maybe you go down to an xt60 connector with guess what the same design just scaled down..
reddit_equals_censor@reddit
oh so you want to break ALL COMPATIBILITY for all hardware ever made in the past? interesting...
your graphics card? nah that won't work anymore. your storage? nope.
but hey at least you can get 0 benefit from the change, because we're talking about a few cents of copper, so there is that right?
and there are no "huge currents" in desktop systems.
even if you have a 720 watt graphics card, that could all be powered by a single small xt120 power connector with proper safety margins and LESS cables than today, because it would be 2 thicker cables, rather than lots of smaller cables.
there is no problem with 12v and i am sick of people being so willing to set EVERY SINGLE DEVICE, that uses 12v on fire, because they feel like, that 48v is a nicer number.
Moscato359@reddit
You make a 48v rail that can step down to 12v for required components
Gpus actually run at 1v lo
Slasher1738@reddit
I mean everything steps down from 12. GPU, CPUs, Memory, pcie cards. But I agree, 48V would be best if we're moving to a new standard.
Moscato359@reddit
40v is significantly easier for regulatory reasons than 48v
JtheNinja@reddit
HDD motors maybe? Fans? Some LEDs, USB ports doing USB-PD charging. Not any actual chips though.
zopiac@reddit
Sort of an aside: this makes me wonder how USB-PD works when going through a motherboard. Does the USB controller for PD-capable ports get fed 12V and then buck/boost it to the requested voltage? Or does it only support 5 and 12v modes and just switches which rail gets connected to the port?
For the most part I just pretend all motherboard ports are still only capable of 5v 500mA and live in blissful ignorance.
mrheosuper@reddit
There will be a dc-dc for usb C port. PPS mode supports dynamic voltage, not only fixed voltages.
mduell@reddit
Good call on HDD motors and fans.
reddit_user42252@reddit
Yeah there's been a big push for 48V in servers i think. Weird how desktops are still stuck at only 12v, laptops use 20v. But the inertia with 12v is so strong i doubt it will change.
fixminer@reddit
CPU and GPU, which consume the vast majority of the power for most systems, run at around 1V.
Vitosi4ek@reddit
I feel from the very introduction of ATX12VO in like 2019(?) most consumers saw it for what it was - an effort to shift costs from PSU makers to motherboard makers, since that's where all the conversion to lower voltages will now have to take place. Might make sense for OEMs though, but that depends on everyone aside from California tightening efficiency standards enough that OEMs consider it worth the cost of transition.
VenditatioDelendaEst@reddit
OEMs use it because it makes the whole computer cheaper if you have to pay for both the PSU and the motherboard.
Guess who else has to pay for both the PSU and the motherboard.
Consoomers are fucking idiots.
oscardssmith@reddit
IMO it makes sense. Basically nothing uses 5v or 3.3v anyway so for most systems the cost is just removed entirely.
The problem is that Intel half-assed the standard. To do it properly, they needed to get a whole new standard that got rid of the legacy 3.3v and 5v connections on motherboards (e.g. SATA and PCIE both have 3.3 and 5v pins that no one really uses (and ideally would have gotten rid of on board pciex16 since GPUs hanging off motherboards is just a really bad idea).
Drachentier@reddit
5 V and 3.3 V are still very much used. Notably, USB needs 5 V and M.2 SSDs run on 3.3 V. A lot of the small chips on the board doing various things (e.g. on board audio) also usually don't run on 12 V.
I still see a few advantages, though. Having the voltage conversion on board allows you to avoid way overbuilt rails for those voltages and cuts down on transmission distance, as transmitting power with such low voltage is really inefficient over longish cables.
zopiac@reddit
One potential problem I foresee is cheaper boards having bugger all current capacity, at least without risking your motherboard's life. As it stands, they can just pass the PSU's power through to a handful of USB ports without worrying about you trying to run a webcam, audio interface, stream deck, capture card, and fancy USB lighting rig while also charging a few devices through a bus-powdered hub.
My worry is that they'll slap on a bargain basement voltage regulator that's effectively spec'd for only mouse/keyboard type power draw whereas I'd expect even a cheap PSU wouldn't have much issues supporting that mess reliably. I'm mostly thinking of the -DS3H and -HDV etc. class of boards being cost cut in this way, and it's not as if you can trust consumers to know (or care) what the technical limits of each port is.
I more-or-less am rooting for 12vo (even just to finally nix that ancient 24pin) but I've no doubt these companies would gladly neuter their products to save a buck in components and tell consumers, "Well you should have bought the $250+ model if you expected to actually be able to really use it!"
mrheosuper@reddit
Cheap as they are, they still have to follow the spec. If the sink device follow the spec and the port still fail, they have to warranty it.
BatteryPoweredFriend@reddit
M.2 uses 3.3v
TwoCylToilet@reddit
Any component that's directly connected to the motherboard with no power cables from the PSU can receive 3.3 and 5V from the motherboard in the 12VO standards, so it's irrelevant that m.2 uses 3.3V.
fixminer@reddit
I don’t see why Intel would have any incentive to reduce costs for PSU makers, that’s just a nice side effect.
12VO removes legacy voltages and makes systems considerably more energy efficient.
Nuck_Chorris_Stache@reddit
Well, if power supplies are cheaper, it means they don't have to pay as much money for them. That's the incentive.
fixminer@reddit
Intel doesn’t sell ATX PCs, so they don’t buy PSUs. (Other than for the PCs for their own offices, I guess, but that is hardly relevant)
Nuck_Chorris_Stache@reddit
Intel is a chip making company. They have and they use computers. They don't need to be selling computers to benefit from lower costs.
fixminer@reddit
What? Do you think Intel would conspire to set up a whole new international standard just to save a few bucks on PSUs for their office PCs?
Besides, the original claim was that Intel was trying to shift costs from PSU makers to motherboard makers (for whatever reason), the overall system cost wouldn’t change that much. I just think that wasn’t their motivation at all, it just so happens that this was necessary to achieve higher efficiency.
Nuck_Chorris_Stache@reddit
Why do you keep making malicious assumptions? Stop putting words in my mouth.
Of course it's not the only reason. It is simply one of a list of reasons.
fixminer@reddit
Sorry, but you are not making much sense. We are talking about whether Intel has any motivation to shift costs from PSU makers to motherboard makers.
How much Intel spends on their own ATX PCs is irrelevant.
If we are talking about 12VO in general, Intel wants to make the (Intel CPU powered) PC platform more efficient (and therefore more competitive). That motivation is obvious.
They just don’t care if it shifts costs from one place to another if the overall operating costs are lower.
Saving some money on their own PCs, while technically true, is about the least relevant motivation for this I could imagine, it is a negligible fraction of their expenses. And more importantly, it doesn’t relate to my original argument at all.
Nuck_Chorris_Stache@reddit
I wasn't.
Companies like saving money. Sure it's not the only reason, but it is still a reason.
-protonsandneutrons-@reddit
OEM have been using 12V-only PSUs for a long time on some systems, before ATX12VO ever launched, tbh.
OEMs have realised the cost & efficiency benefits of 12V-only PSUs, but as there was no ATX standard, they're all proprietary. It's why some $400 OEM PCs have a 80+ Gold or 80+ Platinum PSU out of nowhere.
IIRC, ATX12VO was (partially) to make a standardized ATX platform for OEMs. I'm not sure how many have switched or they're still using their proprietary platforms & designs. But the idea is with one standard, OEMs get lower costs, more competition, and a little more flexibility.
trololololo2137@reddit
wake me up when we get 48V systems. pretty big efficiency gains on wires and PSUs
VenditatioDelendaEst@reddit
No, 48V sucks for desktop.
TerriersAreAdorable@reddit
I really wish NVIDIA had thrown their weight around to deliver a 48v solution instead of going all-in on 12VHPWR.
jocnews@reddit
48V fixes nothing, the connectors would still be faulty and Nvidia would just use the extra headroom to raise GPU TGPs again.
Unlike the 12V-2x6, using 12V for powering PCs is not broken.
TwoCylToilet@reddit
I'd be quite tickled to see 2400W GPUs.
JtheNinja@reddit
48V while keeping everything else the same would let you deliver 2400W at the same heat output we have today. That’s a machine that can pop 120V 20A and 230V 10A circuit breakers all by itself
jocnews@reddit
Looking at the connector sizing, that is the same connector type as the Nvidia 12VHouseFire. Damn it, and I thought the features looked really good.
Moscato359@reddit
To be fair, the 16 pin connector with melting was for 450w+ and this is far, far less power
AtLeastItsNotCancer@reddit
I mean the EPS12V connector is known to be extremely robust and should easily be able to handle over 200W on a single 8-pin (can't remember the exact specs). Two 8-pins are more than enough to cover all use cases on a consumer platform, probably even overkill most of the time. There's no good reason to change it.
Seeing "optimized for cost and space" included in those slides sure doesn't inspire confidence when this is the exact mentality that led to the 12VHPWR fiasco. Maybe the connector specs are good enough on paper, but the smaller you make it, the more manufacturing tolerances start to matter.
I want a connector design that just works, damnit. Once everyone starts worrying whether not buying the exact right brand of the cable or looking at it the wrong way during installation will cause your stuff to melt down, you've already screwed up.
Constellation16@reddit (OP)
From a cursory look, this is a drastic departure even from existing ATX12VO v2. The pin distance is now 3mm also for the 8-pin CPU and 8-pin/12-pin mainboard connector. This means it's the same as the infamous 12V-2x6 GPU connector. Also they remove 12VSB and add this PMBUS communication. So it's completely incompatible to existing ATX12VO systems. Shows you how this is purely some OEM thing since they can just sell new systems.
Moscato359@reddit
Atx12vo already required custom motherboards due to lack of 5v rail, right?
reddit_equals_censor@reddit
oh, so they are bringing EXTREMELY FRAGILE possible fire risk pin sizes to mass oem adoption.
but i'm sure the criminals are getting away just fine with it, once the first people die in a house fire.
that is so disgusting.
are they trying to make it always a fire hazard to buy a prebuild or any computer in the future to push for more cloud computing subscriptions? 😃
and it is worth adding here, that this has nothing inherently to do with size, because those shits could have made very small and very safe connectors, that use 2 connections, instead of 8 then with big contacts and thicker cables, but why improve things, when you can just make them vastly worse right?
-protonsandneutrons-@reddit
The previous versions of ATX12VO had a reserved "Telemetry Pin" to communicate total usage, but maybe PMBus is more widely adopted? PMBus can communicate a lot more than just total usage: input / output voltages, current, fan speeds, internal temps (3 sensors), fault reporting, etc.
Nicholas-Steel@reddit
Unfortunately this ATX12VO v3 spec makes the PMBus a tiered optional thing (you can have 1 pin, 2 pins, 3 pins or 4 pins for PMBus). So it's gonna be annoying comparing things depending on what you want when looking to buy a PSU.
-protonsandneutrons-@reddit
Unfortunate. PMBus is usually pretty un-standardised, but it would've been neat for Intel to make PMBus + a standard consumer comms. protocol and standardized pins. Currently, some higher-end PSUs push out data via USB that HWiNFO64 can capture.
Intel ATX continues its tradition of, to put it lightly, "interesting" changes.
nittanyofthings@reddit
If California & EU regs cannot allow prebuilt 400w gaming PCs with ATX 3.1, due to idle power, the SI will have to stop using ATX parts that DIY uses. DIY market can't support production itself. If prebuild switches to a standard like atx12v v3, the DIY market can follow. But if SI do proprietary stuff like Dell does, the DIY market gets too expensive and dies. atx12vo might actually be very important, if you don't want to buy black box systems.