Intel Intros new PSU standard ATX12VO

GiGaBiTe

[H]ard|Gawd
Joined
Apr 26, 2013
Messages
1,607
They already do this on NUC, Thin-ITX and Mini-STX boards. Those things are miniscule and I haven't heard of this being a serious issue, the same applies to some degree to most every laptop on the planet. Whats that? A billion or so examples. God,they must exploding every second.

You're really good at creating non sequitur and straw man fallacies, you should be a politician ;)

Your concerns, look more like pointless FUD.

When you're calling a widely documented phenomenon about Apple laptops dying for no reason FUD, you become an Apple shill.

This is in fact a smart move from Intel. Will the need of 3.3V and 5V disappear ? No, but it won't be the PSU but the motherboard which will provide the current converted from 13V.
The sùart move is that the motherboards will be become more expensive and fragile and the PSU simpler and cheaper. Intel doesn't care about the PSU manufacturers but cares about the motherboard manufacturers and the loss of Intel speedy update pace they may resent. So Intel is taking from the PSU manufacturers and giving to the motherboard manufacturers what he took and in between he adds another brick to the planned obsolescence of your system.

Yeah, it seems like Intel is doing everything in their power to make desktop PCs less reliable and more rapidly obsolete. I remember when Intel had multiple core revisions of a CPU architecture used the same socket, those were good times. Seems like AMD is the only one doing that now.
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
11,267
You're really good at creating non sequitur and straw man fallacies, you should be a politician ;)



When you're calling a widely documented phenomenon about Apple laptops dying for no reason FUD, you become an Apple shill.



Yeah, it seems like Intel is doing everything in their power to make desktop PCs less reliable and more rapidly obsolete. I remember when Intel had multiple core revisions of a CPU architecture used the same socket, those were good times. Seems like AMD is the only one doing that now.


I think you need to examine what a non sequitur is.

Every laptop, NUC, Mini-STX, and Thin-ITX board all run on a single voltage push Voltage conversions to the Motherboard.

This is over a billion devices with similar setup, to the tech under discussion. Nothing could be more germane to the topic under discussion.

But you think it's a problem to do this, because some Apple laptops had an issue?

The concern here is whether this kind of technology is a real problem in general. A billion+ devices using it says no.

If you want to rant about Apple problems, well that actually is a non sequiter in the context of using this technology on more motherboards.
 

BinarySynapse

[H]F Junkie
Joined
Feb 6, 2006
Messages
15,111
You're really good at creating non sequitur and straw man fallacies, you should be a politician ;)



When you're calling a widely documented phenomenon about Apple laptops dying for no reason FUD, you become an Apple shill.

Apple laptops dying for no reason have nothing to do with the power rails. 99.99% of laptops are fed with 16-20v supplies that are dropped down to four or five different voltages. The problem with Apple is they cut costs so close to the bone that their designs tend to be running on the edge of self-destruction.



Yeah, it seems like Intel is doing everything in their power to make desktop PCs less reliable and more rapidly obsolete. I remember when Intel had multiple core revisions of a CPU architecture used the same socket, those were good times. Seems like AMD is the only one doing that now.

There has never been a time when multiple generations of CPUs work on the same socket for Intel. I know you're going to say socket 775 was one, but it really wasn't and was a complete mess throughout its production life.

multicore models didn't work on i925,
Netburst as a whole was dropped on P35 and P45.
Conroe didn't work on i955, i875, or i865, and many i945 boards were incompatible without a BIOS update (which may not have been provided).
Penryn didn't work on P965.



Pentium 4 had three different sockets, one of which required a VRM revision to support Prescott.
P6 had four (Socket 8, Slot I, socket 370 Coppermine, and socket 370 Tualatin).
Pentium had three (socket 4, socket 5, socket 7).
80486 had three. (168, Socket 2, and Socket 6), four if you count Socket 1 for Overdrive processors.
 

GiGaBiTe

[H]ard|Gawd
Joined
Apr 26, 2013
Messages
1,607
The problem with Apple is they cut costs so close to the bone that their designs tend to be running on the edge of self-destruction.

Design faults and cost cutting too far aren't strangers in motherboard design, no matter the manufacturer.

There has never been a time when multiple generations of CPUs work on the same socket for Intel.

That's great and all, but that's not what I said. I said multiple core revisions of an architecture, not multiple architecture generations. But, you also disprove your own argument here just a few lines down, multiple generations of processor architectures did indeed use the LGA775 socket, irrespective of chipset limitations. There were weird motherboards (especially from ASUS) that used black magic to get chipsets to do things they normally wouldn't have, like the P5GC-MX/1333 which could run pretty much the full range of Celeron, Pentium 4 and Core 2 parts up to Wolfdale on an i945 chipset. I do recall some P35 motherboards which still supported Netburst parts via similar witchcraft, even though Intel officially abandoned it.

Netburst had Northwood, Gallatin, Prescott, Prescott-2M and Cedar Mill on LGA775.
Core 2 had Conroe, Allendale and Wolfdale on LGA775. Penryn is a mobile part.
Going further back; Coppermine, Coppermine-T and Tualatin had PGA370 parts.
Socket 7 was backwards compatible with Socket 5 CPUs
Super Socket 7 was backwards compatible with both Socket 7 and Socket 5 CPUs, my FIC PA-2013 will run all three, even though Intel didn't use the socket.

My point was obviously multiple, I didn't say "all".
 

BinarySynapse

[H]F Junkie
Joined
Feb 6, 2006
Messages
15,111
I'm not sure how you quoted my words with someone else's name.



Design faults and cost cutting too far aren't strangers in
motherboard design, no matter the manufacturer.

If Gigabyte, Asus, or MSI decide to skimp on the 5v and 3.3v circuits and end up with bunk boards and the support headaches that go with them because of it, that's on them. It's not a reason to believe the specification they're abusing was flawed or a bad idea.


That's great and all, but that's not what I said. I said multiple core revisions of an architecture, not multiple architecture generations. But, you also disprove your own argument here just a few lines down, multiple generations of processor architectures did indeed use the LGA775 socket, irrespective of chipset limitations. There were weird motherboards (especially from ASUS) that used black magic to get chipsets to do things they normally wouldn't have, like the P5GC-MX/1333 which could run pretty much the full range of Celeron, Pentium 4 and Core 2 parts up to Wolfdale on an i945 chipset. I do recall some P35 motherboards which still supported Netburst parts via similar witchcraft, even though Intel officially abandoned it.

Netburst had Northwood, Gallatin, Prescott, Prescott-2M and Cedar Mill on LGA775.
Core 2 had Conroe, Allendale and Wolfdale on LGA775. Penryn is a mobile part.
Going further back; Coppermine, Coppermine-T and Tualatin had PGA370 parts.
Socket 7 was backwards compatible with Socket 5 CPUs
Super Socket 7 was backwards compatible with both Socket 7 and Socket 5 CPUs, my FIC PA-2013 will run all three, even though Intel didn't use the socket.

My point was obviously multiple, I didn't say "all".

Your list is wrong.

Northwood was never on 775. It was only on 423 and 478, There was also Willamette on 423 .
Gallatin was only on socket 603 (and was a server/workstation part).
Penryn was used to reference the architecture before release in addition to the code name of the mobile part.


Secondly, I wasn't saying you said "all". I was pointing out how badly you're remembering the good ole days by listing multiple examples of when Intel changed things mid-architecture in ways that broke compatibility, particularly on the same socket. From 1992 through 2019 they have broken forward socket compatibility at best every other year and often every year.

775
-You couldn't buy a Pentium 4 530 and a i925 board in 2004 and upgrade to a Pentium D 930 a year later.
-You couldn't buy a Pentium D 930 and a 955X board in 2005 and upgrade to an E6300 in 2006.
-You couldn't buy a Core 2 Duo E6600 (Conroe) and a P965 board in 2006 and upgrade to a Core 2 Duo E8400 (Wolfdale) in 2008.

478
-You couldn't buy a Pentium 4 2.4C (Northwood) and i865 board in 2003 and upgrade to a Pentium 4 3.2E (Prescott) in 2004.

370
-You couldn't buy a 1GHz Pentium III (Coppermine) in 2000 and be able to upgrade to a 1.2GHz Pentium III (Tualatin) in 2001 without a socket adapter (extra cost).

On that front, Intel is doing nothing different today than they ever have, except at least now it's easier to see when they're breaking compatibility.


And yes, some funky stuff was out there that let you use newer stuff on older platforms, but it was never available when the older platform was new and people were buying it. Plus those products often had stability and compatibility issues. sometimes very serious one. They were almost never worth the money.





As far as this spec goes, they're not forcing the change. They're offering an alternative to OEMs to reduce costs and complexity while increasing flexibility in the systems they offer. If there's seen a benefit in it, it'll take of like ATX and PCIe did; if not then it'll die like BTX and Rambus did.

It could even benefit the DIY crowd. Need more 5v power, get a stronger 5v converter. Don't need 5v at all, you won't have a bunch of unused 5v wiring. You're not saddled with a PSU with a strong 12v rail but undersized 5v.
 

Factum

2[H]4U
Joined
Dec 24, 2014
Messages
2,466
I like the idea of simplifying the voltages/connections to the mainboard. I'm not so certain about having SATA/etc. power coming off the mainboard to the drives, though it's certainly less of an issue these days than in the past (e.g., the proliferation of m.2, disappearing optical drives). Mainboards are already crowded enough without adding more connectors to them.

I do wish they'd rethink the form-factor of the PSU. The large, bulky cube of the ATX standard always takes up so much space in a case and is in the way of a sleek, low-profile build. Maybe with the PSU being simplified SFX will take off, which will help in that respect.

Not throwing physics out the window (I have never looked at my PSU since installing it btw so I don't get the "form thing?) what "form" would you like to see?
 
Joined
Dec 1, 2011
Messages
988
Not throwing physics out the window (I have never looked at my PSU since installing it btw so I don't get the "form thing?) what "form" would you like to see?

I'd like to see something much lower in height, basically designed to unobstructively sit at the bottom of the case. Similar to the PSUs used in the various cheese-grater PowerMacs/Mac Pros or some HP workstations I've seen. It should also be implicitly designed for passive cooling, but with the option for quiet fan cooling for higher-capacity units (maybe something akin to the blowers used in some GPUs). Gold-rated at a minimum.

Like I said, SFX helps. But it still takes too much from ATX I feel (e.g., fan placement) and is still fairly tall. TFX has the same basic issues. FlexATX and 1U come close, but suffer from overall size restraints (width, length) and the typical 40mm fan(s).

Yeah, the basics of component sizes (e.g., transformers) can present problems. But if 1U PSUs can be made to support 700+ W, then it's certainly possible. And the proposed simplification to eliminate the 3.3 and 5 V circuits only makes it easier.
 

RamonGTP

Supreme [H]ardness
Joined
Nov 9, 2005
Messages
8,150
It's doesn't matter if they have 200k ppl, as most of them are doing jack. Intel is failing bad at their core competency. 10K ppl at AMD (vs 100k at Intel), is showing Intel how shit is to be done right. Lmao, Intel cannot even incorporate pcie4. I don't have an issue with changing up the psu standards, but imo they have some major core competency issues to deal with first.

How do you know what they're doing???
 

Shadowarez

Limp Gawd
Joined
Jul 8, 2019
Messages
410
so guessing that spec wont make it into this seasonic psu they previewed ces 2019 the Seasonic Connect,
 

Attachments

  • Connect-transp.png
    Connect-transp.png
    1.5 MB · Views: 0

IdiotInCharge

NVIDIA SHILL
Joined
Jun 13, 2003
Messages
14,710
I'd like to see something much lower in height, basically designed to unobstructively sit at the bottom of the case. Similar to the PSUs used in the various cheese-grater PowerMacs/Mac Pros or some HP workstations I've seen. It should also be implicitly designed for passive cooling, but with the option for quiet fan cooling for higher-capacity units (maybe something akin to the blowers used in some GPUs). Gold-rated at a minimum.

Like I said, SFX helps. But it still takes too much from ATX I feel (e.g., fan placement) and is still fairly tall. TFX has the same basic issues. FlexATX and 1U come close, but suffer from overall size restraints (width, length) and the typical 40mm fan(s).

Yeah, the basics of component sizes (e.g., transformers) can present problems. But if 1U PSUs can be made to support 700+ W, then it's certainly possible. And the proposed simplification to eliminate the 3.3 and 5 V circuits only makes it easier.

Biggest issue I think is noise. With current ATX, SFX, etc. sizes, PSUs can be essentially silent most of the time, but also remain one of the quietest components even under heavy load for the size.

While living around 40mm screamers, honestly, I'm open to suggestions that do not raise the dB / watt threshold that's been achieved with current form factors. I really do like that many ATX PSUs ship with 140mm fans.
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
11,267
Biggest issue I think is noise. With current ATX, SFX, etc. sizes, PSUs can be essentially silent most of the time, but also remain one of the quietest components even under heavy load for the size.

While living around 40mm screamers, honestly, I'm open to suggestions that do not raise the dB / watt threshold that's been achieved with current form factors. I really do like that many ATX PSUs ship with 140mm fans.

Another area where the PC could use a Form Factor rethink. The modern PSU is something that really doesn't produce a lot of waste heat. You should be able to simply have a smaller PSU with good open venting that relies on positive case pressure to push some air through and cool it.
 

BinarySynapse

[H]F Junkie
Joined
Feb 6, 2006
Messages
15,111
Another area where the PC could use a Form Factor rethink. The modern PSU is something that really doesn't produce a lot of waste heat. You should be able to simply have a smaller PSU with good open venting that relies on positive case pressure to push some air through and cool it.
You don't want it too open. There are dangerous voltage levels in there. And even the best efficiency PSUs are still putting out 50-100w of heat.
 

IdiotInCharge

NVIDIA SHILL
Joined
Jun 13, 2003
Messages
14,710
Another area where the PC could use a Form Factor rethink. The modern PSU is something that really doesn't produce a lot of waste heat. You should be able to simply have a smaller PSU with good open venting that relies on positive case pressure to push some air through and cool it.

I share the desire for smaller and quieter, but I balance that with the lowest common denominator -- if you take the fan out completely, enough system builders and end users will install them with insufficient airflow to cause significant support issues.

And on the other hand, even with ITX, we're still looking at the majority of hardware arrangements supporting ATX PSU dimensions, and with the higher power requirements (Threadripper, top-end GPUs, overclocks, AICs...) available, at least the volume provided by ATX dimensions is actually needed.
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
11,267
I share the desire for smaller and quieter, but I balance that with the lowest common denominator -- if you take the fan out completely, enough system builders and end users will install them with insufficient airflow to cause significant support issues.

And on the other hand, even with ITX, we're still looking at the majority of hardware arrangements supporting ATX PSU dimensions, and with the higher power requirements (Threadripper, top-end GPUs, overclocks, AICs...) available, at least the volume provided by ATX dimensions is actually needed.

It would need to be part of wider PC form factor rethink. Not just taking fans away from an existing PSU designs.
 

plugwash

[H]ard|Gawd
Joined
Sep 17, 2010
Messages
1,547
As for the PSU manufacturers, it matters even less. Today's PSUs are fully capable of not needing a load on their 5v and 3.3v rails. A modular power supply can just leave the motherboard connector disconnected and supply a cable that conforms to the new standard using their existing pinout.
Unfortunately in addition to getting rid of the 5V and 3.3V rails they also changed the standby rail from 5V to 12V.

So you can't just use a passive adapter on an existing PSU for this.

I'm not concerned about power draw, I'm concerned about implementation and the fact that multiple buck converters are going to be sprinkled around the board.
That ship sailed years ago.

Modern electronics needs a whole bunch of voltages, most of them very low, some of them variable to control the trade-off between speed and power consumption. I see no reason to belive that 3.3V and 5V buck converters would be any more problematic than the bunch of converters that motherboards already contain.
 

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,465
Unfortunately in addition to getting rid of the 5V and 3.3V rails they also changed the standby rail from 5V to 12V.

So you can't just use a passive adapter on an existing PSU for this.


That ship sailed years ago.

Modern electronics needs a whole bunch of voltages, most of them very low, some of them variable to control the trade-off between speed and power consumption. I see no reason to belive that 3.3V and 5V buck converters would be any more problematic than the bunch of converters that motherboards already contain.

While true, I don't see why you can't have an active adapter that doesn't take up all that much space and is fairly inexpensive. Something like this built into the cable.
 

Wooshoo

Weaksauce
Joined
May 27, 2019
Messages
110
Same here. My first thought was "Did they apply it to new form factors"?. They didn't, But it should still help streamline things if it takes off.

I really feel like we are overdue for a complete form factor overhaul, for card, PSU, everything... ATX is has poor space efficiency, and cooling for GPUs is really sub optimal. The card cage model was designed for VERY low power cards, not 300W GPUs.
Lol My grumpy father in law said essentially the same thing just last week when I was talking to him about “my computer” that I am building for him. I was trying to get an idea on the case and discussing cooling details went a bit off topic.
 
Top