Monday, October 11th 2021

PCIe Gen5 "12VHPWR" Connector to Deliver Up to 600 Watts of Power for Next-Generation Graphics Cards

The upcoming graphics cards based on PCIe Gen5 standard will utilize the latest PCIe connector with double bandwidth of the previous Gen4 that we use today and bring a new power connector that the next generation of GPUs brings. According to the information exclusive to Igor's Lab, the new connector will be called the 12VHPWR and will carry as many as 16 pins with it. The reason it is called 12VHPWR is that it features 12 pins for power, while the remaining four are signal transmission connectors to coordinate the delivery. This power connector is supposed to carry as much as 600 Watts of power with its 16 pins.

The new 12VHPWR connector should work exclusively with PCIe Gen5 graphics cards and not be backward compatible with anything else. It is said to replace three standard 8-pin power connectors found on some high-end graphics cards and will likely result in power supply manufacturers adopting the new standard. The official PCI-SIG specification defines each pin capable of sustaining up to 9.2 Amps, translating to a total of 55.2 Amps at 12 Volts. Theoretically, this translates to 662 Watts; however, Igor's Lab notes that the connector is limited to 600 Watts. Additionally, the 12VHPWR connector power pins have a 3.00 mm pitch, while the contacts in a legacy 2×3 (6-pin) and 2×4 (8-pin) connector lie on a larger 4.20 mm pitch.
There are already implementations of this connector, and one comes from Amphenol ICC. The company has designed a 12VHPWR connector and listed it ready for sale. You can check that out on the company website.
Source: Igor's Lab
Add your own comment

97 Comments on PCIe Gen5 "12VHPWR" Connector to Deliver Up to 600 Watts of Power for Next-Generation Graphics Cards

#77
looniam
TotallyIt's not carrying power, it's just for signaling using a thicker wire would just cause/add unnecessary cost issues.
yeah, look up. (the last post.)
Posted on Reply
#78
Totally
looniamyeah, look up. (the last post.)
Did you look at the article? Because PCIe gen 5 uses double the freq, and to my knowledge when it comes to signaling thicker wires = worse. I don't know why you are bring up current PSUs, because when those beging designed how would Gen 5 considerations be made? I'm pretty sure that's expressly why this isn't backwards compatible. And why would you want to jumper off of ground if you don't have to? I mean shit dude, not only would the noise on the PCIe have to be dealt with add all the noise coming of the ground too for shit and giggles? Anyway, besides that at the end of the day THICKER wire costs MORE.
Posted on Reply
#79
looniam
TotallyDid you look at the article? Because PCIe gen 5 uses double the freq, and to my knowledge when it comes to signaling thicker wires = worse. I don't know why you are bring up current PSUs, because when those beging designed how would Gen 5 considerations be made? I'm pretty sure that's expressly why this isn't backwards compatible. And why would you want to jumper off of ground if you don't have to? I mean shit dude, not only would the noise on the PCIe have to be dealt with add all the noise coming of the ground too for shit and giggles?
wait wut?
i don't know what article you read but this is the power connector to add in cards. there is 12v and ground; any signal is a simple open/closed circuit. closed/connection at one/two sensor(s) is 75/150 watts additional power. (

section 4 explains most of it and fwiw, pci-sig standards are backwards compatible - its all over their documentation..
Posted on Reply
#80
InVasMani
Does it come bundled with a climate controlled room?
Posted on Reply
#81
Totally
looniamwait wut?
i don't know what article you read but this is the power connector to add in cards. there is 12v and ground; any signal is a simple open/closed circuit. closed/connection at one/two sensor(s) is 75/150 watts additional power. (

section 4 explains most of it and fwiw, pci-sig standards are backwards compatible - its all over their documentation..
The doc you linked is for a 300W spec from 15years ago.
Posted on Reply
#82
looniam
TotallyThe doc you linked is for a 300W spec from 15years ago.
and that was the last change. :slap:
Posted on Reply
#83
Totally
looniamand that was the last change. :slap:
really? according to pcisig there have been 4 additonal changes since then, meaning we're on Rev. 5.0 not Rev. 1.0, and the last one was dated June 18, 2021
Posted on Reply
#84
looniam
Totallyreally? according to pcisig there have been 4 additonal changes since then and the last one was dated June 18, 2021
ok mr. smarty pants. tell me how any of those are add in power specs?

all those deal with data signaling as you were yapping aboput before. the 12v power specs are entirely different; and gets refereed to in those docs.

do some reading before judging buddy. this is not my first rodeo, though serves me right for saying something related to PSUs; because there is always that special person . .
Posted on Reply
#85
jonnyGURU
ZoneDymoso this is not the same as that new standard Nvidia is trying to push?

and also....great....lets just embrace gpu's using more and more power....
It is.

It's the same 12-pin part, with a +4 added to the bottom for additional sense wires. The FE's 12-pin fits in this 12+4-pin. The rub is if the GPU manufacturer pushes that 450W limit and requires the sense pin to "prove" that a "approved" connector is in play. That's when you can no longer use the 12-pin.
Posted on Reply
#86
R-T-B
looniamwww.amphenol-icc.com/product-series/minitek-pwr-cem-5-pcie.html


ick. why not the same gauge? - yeah whatever; this needs to die.

there is time.
Because it's a signal pin carrying no more than 1A?
Vayra86Then we have different sorts of reading comprehension, because what I'm reading in the quoted post is "Mining and playing games both use energy, so if you want to regulate mining, you need to regulate everything else that uses energy, including things that serve the purpose of having fun / playing"

Which is an utterly ridiculous train of thought, underlined by my link up there.
"Go play outside" is an easy counter.

The point is legislating what people do with energy for their own personal use is folly.
Posted on Reply
#87
comtek
What does VHPWR stand for? Very High Power?
The successor will be VHSPWRFTW, VeryHighSuperPowerFTW
Posted on Reply
#88
TheLostSwede
News Editor
comtekWhat does VHPWR stand for? Very High Power?
The successor will be VHSPWRFTW, VeryHighSuperPowerFTW
I think it stands for 12V High Power, but if you disregard the 12, I guess you could make it stand for anything...
Posted on Reply
#89
looniam
R-T-BBecause it's a signal pin carrying no more than 1A?
it maybe rated for 1 amp but it carries no load. afaik it not a signal (mis-label in the manufacturer's spec sheet?) as much as a sensor that jumpered to ground.
page 31 of the pci-e specs i attached in aprevious post:


when it comes down to it pci-sig needs to update the power - anything with more that two 6 pin OR one 8 pin and one 6 pin is outta spec and won't be pci-e certified. yes, that means anything with two 8 pins are not.

imo, 600 watts for one connector is overkill; the 450 watts connector nvidia tried out is much more appropriate - not because it was nvidia but that would be 75/150/450 instead of 75/150/600 - which leaves a big gap.
Posted on Reply
#90
Tatty_Two
Gone Fishing
Grateful if we can slip back to the topic in hand ...... thank you.
Posted on Reply
#91
svan71
CPU power increasing while energy requirement decreasing, why are GPUs getting more power hungry ?
Posted on Reply
#92
cst1992
svan71CPU power increasing while energy requirement decreasing, why are GPUs getting more power hungry ?
It's a race between performance targets and fab improvements. A few years ago the fabs were winning, but now they aren't. And they shouldn't - you can only make a transistor so small. We're already at the 3nm point - I'll bet in only a few years we'll reach the point of "that's it, we can't make the manufacturing any smaller than that".
Posted on Reply
#93
LabRat 891
cst1992It's a race between performance targets and fab improvements. A few years ago the fabs were winning, but now they aren't. And they shouldn't - you can only make a transistor so small. We're already at the 3nm point - I'll bet in only a few years we'll reach the point of "that's it, we can't make the manufacturing any smaller than that".
We'll see how much truth there is in it, but Intel, publicly, and other companies less publically, have 'angstrom-scale' lithography process nodes on 'roadmap'.
I expect photonics and/or spintronics to start being 'hybridized' into the 'classical' computing architectures we use today.
Please correct my ignorance if so, but I'm pretty sure the science(s) associated with spintronics are already being factored today to mitigate Electron Migration and other (near)quantum-scale engineering challenges.
Posted on Reply
#94
cst1992
If they're indeed talking about literally the angstrom scale, then we're talking about components made out of individual atoms. Like the seven-atom transistor that was made back in 2010. Even if you reach that scale, will it be possible to make components that behave correctly at reasonable cost? I doubt it. Either the components will be very expensive, or have defects that can't be easily fixed, or take a long time to make.
Posted on Reply
Add your own comment
Jun 12th, 2024 08:59 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts