Let’s move on to the new ATX 3.0 standard and what is being secretly worked out in the specification backroom with the involvement of CPU and GPU manufacturers. The whole thing really packs a punch and that’s where my criticism comes in. On the one hand, power supplies designed in this way can cause problems with older systems (which would be tolerable collateral damage from the point of view of a certain CPU manufacturer, for example), and on the other hand it only complicates the production (and costs) of really efficient power supplies and complete systems unnecessarily. Which means I have to say a word about the ATX12VO v2.0 propagated by Intel, which violates any logic of sustainability and sense and thus only serves the interest of the protagonist. But more on that in a moment.
Power consumption and load peaks as an extreme mix
Let’s move on to what Intel has given manufacturers for 2022 and first marvel at the numbers. For those who don’t know: CEM means PCI SIG Card Electromechanical, i.e. the plugged-in add-in cards. The CPU manufacturer estimates between 275 and 300 watts for the CPU, which is plenty generous, though. However, I’ve also managed to squeeze well over 300 watts into a Core i9-12900KF, which quickly puts the generous rating into perspective again.
The power consumption values of the graphics cards are also anything but utopian, because even without OC, one can hear rumors for AMD’s and NVIDIA’s new flagships about 500 to 550 watts of real board power. There better not be any overclocking headroom left. The estimation for the rest of the system is rather inconsistent, as the 100 watt approach is already a bit tight with motherboard power consumption of up to 50 watts even at the lower end, while the 300 watts is a bit excessive for the largest system. On balance, though, that should all somehow add up.
The estimated values for the PSU measurement (“PSU Size”) are thus clarified, but what is it about the Total Power? This is exactly what we need to talk about right now, because it’s the part that worries me (and the power supply manufacturers) the most! Intel also has a defined idea for this and at the same time an announcement to the producers of the electricity suppliers. Let’s take a look at this now and get wide-eyed. Even if the length of the highest load peak intervals is limited to 100 μs each, the permissible total share of up to 10% of the total is downright abstruse. This means that the power supply must actually deliver 200% of the power over 10% of the operating time.
This actually turns the 1.2 kW power supply into a 2.4 kW power supply in disguise. This is because the normal operation is defined at just 50% of the time and the rest of the time you are free to overload. Of course, this will all more or less come down to a pure single-rail solution as well, but from a purely electrical standpoint, this is outright insanity. Which supervisor chip should still be able to realize a reliable protection circuit for OCP/OPP in a fixed defined interval? What then is an allowable load and where does the emergency begin?
ATX v3.0 – Soften limits, close eyes and hope for the best!
And as if all this wasn’t enough ecological and circuit nonsense, they’re going one better. Even if some of the manufacturers would be able to realize such power supplies under the current ATX specifications, the bar for the quality of the power supplies and the limits to be met is much too high for most of the cheap manufacturers. Or the products would then at least be so expensive again that the acceptance would go completely towards zero. So they’re bending it a bit and even risking backwards compatibility. But let’s start back at the PCI SIG and see what they’re pretending:
So far, so clear. And now the load peaks (“transients”) come into play. This in turn affects the power supply and the so-called DC output voltage regulation, which is needed to cope with power fluctuations. You know my hobby with tracking down and documenting such transients and if you believe the information trickling out of the R&D of the graphics card manufacturers, then it’s going to be pretty neat in the not too distant future.
The current state with a GeForce RTX 3090 FE, the Core i9-12900KF and a MSI MAG Z690 Unify I have tested is such that already without OC and without other components (except the two SSDs) almost 800 watts add up, which can strike as load peaks. With a well cooled GeForce RTX 3090 and a 450 Watt BIOS, we would be at 1000 Watt and above with a bit of bad luck, because the spikes don’t increase linearly, but almost exponentially.
We can see that the guidelines with the tolerance values for the graphics cards of +5% or -8 percent from the 12 volts nominal voltage by the PCI SIG were already not set very tight, but the danger of undervoltage due to even faster and more frequent load changes (we remember now please up to 50% of the operating time) has unfortunately become even more real with it. Let’s do the math: The -8% of 12 volts is -0.96 volts and thus 11.04 volts. which must not be fallen short of. This is much more generous than the current ATX specifications, which only allow 11.4 volts.
However, this lower value is now to be lowered to 11.2 volts for ATX 3.0 in order to be able to handle the load peaks more generously. Since this is still not enough in reality and one must also manage the balancing act of downwards compatibility, they additionally want to raise the 12 volt nominal voltage moderately to 12.2 volts for ATX 3.0 or ATX12VO v2.0. In turn the 0.6 volts of the old specification would become a whopping 1.0 volt, which is now allowed as a fluctuation. This sounds smart at first, but is dangerous for other hardware that is sensitive to overvoltage. This also includes consumers with various extra-low voltages, which are obtained by several cascaded step-down converters and whose end product can then already deviate significantly from the planned and required voltage window.
ATX12VO v2.0
The reduction of the power supplies to pure 12.2 volts propagated by Intel is not only nonsensical in terms of circuitry, but also very inefficient. I know of several motherboard layouts in this regard where generating the other partial voltages up to the power hungry 5 volt connectors (USB 4 / TB 4) is a real challenge. If one really wanted to implement this consistently, then there would be no motherboard without active cooling and already the Micro-ATX format would be obsolete. What is currently still possible in the ITX area with the smallest boards and low power consumption, quickly becomes a farce in the power ranges of current PCs.
Good power supplies last for several years and warranty periods of 5 to 10 years are also not uncommon. Statistically, however, motherboards are already being replaced much more frequently than power supplies. If you then put all the DC-DC controllers on the motherboard, it would not only become significantly more expensive, but valuable resources would also be wasted senselessly with every failure or replacement. Then quickly disposed of with what may still work. And then there are still the protection circuits that need to be implemented.
The error rate of a motherboard would also increase, as shown by initial statistical surveys and MTBF tests on the boards. This is also one of the reasons why the entire power supply and motherboard industry is against this superfluous connector and rebels loudly. It is amazing, to say the least, that unlike other attempts at standardization, Intel has not somehow managed to get away with it, despite its overwhelming market presence and influence. That’s saying something. Nevertheless, the ATX12VO v2.0 is also raised to 12.2 volts.
Summary and conclusion
I think the approach of adding more and more power just to generate more and more performance is completely the wrong way to go, at least in the consumer space. Breaking this spiral and using the know-how, including technical manufacturing progress, instead for more environmental compatibility and sustainability would certainly be the real order of the day. It’s just that AMD, Intel and NVIDIA are unfortunately companies where electrical power doesn’t seem to matter at their headquarters, because anything that doesn’t cost much doesn’t get the appreciation it deserves.
At the moment we don’t have to explicitly appeal to anyone to avoid such products, since they are not available anyway (or the availability of their successors is still completely up in the air), but it would be time to think about where all this is going and whether we want to submit to this compulsion to consume without resistance. I would like to give this information to the readers, so everyone can decide themselves based on their own priorities. Don’t even get me started about the unspeakable mining, that’s all nonsense from my position anyway.
109 Antworten
Kommentar
Lade neue Kommentare
Urgestein
Mitglied
Urgestein
Urgestein
Urgestein
1
Urgestein
Urgestein
Urgestein
1
Urgestein
Urgestein
Veteran
Veteran
Urgestein
Urgestein
Veteran
Urgestein
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →