Intel Alder Lake ATX12V Peak Current Recommendation is Allegedly Higher Compared to Rocket Lake

[AleksandarK]

Intel's upcoming Alder Lake family of processors, more specifically the desktop ALD-S SKUs, are supposedly going to require a higher peak current for the upcoming processors. In the table provided by the Chinese tech media outlet, FCPOWERUP, we are seeing that ALD-S processors have different power requirements for the ATX12V rails on their power supplies. The listed table shows the previous generation of Intel processors, the 10th and 11th generation, as available in 165 Watt variants. Even though there are no 165 Watt Comet Lake and Rocket Lake SKUs, this is only a placeholder for their PSU recommendations in case those SKUs were to be released.

According to the table, the peak current recommendation for the upcoming Alder Lake is higher at least 5 Amps across all SKUs. The 165 Watt SKUs have the requirement of 45 Amps (compared to the 40 A of Comet Lake and Rocket Lake), while the 125 Watt SKUs require 39 Amps, which is higher than the previous 34 Amp requirement. For 65 Watt models, the new peak recommendation is 38.5 Amps, a jump from the previous 30 Amp choice. The lowest rated 35 Watt SKUs are recommended to use 20.5 A, while the previous generations used 16.5 A current. It should be noted that the continuous rating has not been changed (new generation 35 Watt models actually use less current), which indicates that Alder Lake could have higher peak usages of power, meaning that PSU choice should be made with a 50-100 Watt higher rating.


View at TechPowerUp Main Site

 

[Crackong]

165W SKUs ?

 

[londiste]

Maybe a preparation for ATX12VO support?

Edit:
Also, what is the context of this table? It looks like a system-wide power recommendation, think similar to Nvidia/AMD PSU wattage recommendations, although in this case probably for system integrators or OEMs.
For example with a 65W TDP CPU - 276W sustained and peak at 360W or now 462W? This is even beyond what Intel's CPUs consume when unlimited.

 

[docnorth]

To be honest, I didn't pay much attention to peaks and spikes until the recent generation of GPUs. We can't know if ADL-S will have higher median consumption than RKL-S (unlikely imo) or CML-S or, like @londiste said, it has something to do with the ATX12VO standards etc. Maybe it's not right to draw conclusions about CPUs from GPU consumption, but Big Navi GPUs tend to produce higher spikes than Ampere, despite being usually more efficient.
Still those peaks could alter the PSU choice.

 

[_Flare]

If i am not mistaking, the manufacturers are searching for ways to use more drive current (Amps) for the transistors.
Intel may have found a way to do things differently from 14nm+++ to 10nmSF, now Intel7 process with a little less voltage but more Amps.

The 12V is not directly related to the 45A here, just because 12V x 45A = 540W wich is no number directly tied to anything CPU related.

Would be very interesting to have numbers from a similar document but related to TigerLake Amps.

 

[DeathtoGnomes]

I wonder if they are upping the amps to boost single core performance higher

 

[eidairaman1]

Sounds like conductors would need to be physically thicker then, we are limited by the laws of physics

 

[londiste]

I wonder if they are upping the amps to boost single core performance higher

No.
Unless you think 200W+ to a single core is feasible.

 

[Tomorrow]

Maybe a preparation for ATX12VO support?

Edit:
Also, what is the context of this table? It looks like a system-wide power recommendation, think similar to Nvidia/AMD PSU wattage recommendations, although in this case probably for system integrators or OEMs.
For example with a 65W TDP CPU - 276W sustained and peak at 360W or now 462W? This is even beyond what Intel's CPUs consume when unlimited.

ATX12VO wont be a thing on most DIY motherboards. Only a handful on models will be using it.

Motherboard Manufacturers Balk at Intel's New ATX12VO Power Standard, Alder Lake Z690 Motherboards to Retain 24-Pin Connectors

Something that Intel is keen on introducing as part of its upcoming Intel Z690 motherboards for 12th Gen Core "Alder Lake" processors is support for a new power standard called ATX12VO that promises reduced idle desktop power for lower energy use.

www.thefpsreview.com

 

[Punkenjoy]

It look to me that ADL-S might boost harder and quicker so PSU will need to support that. But they should already do because driving a modern high end GPU is already harder.

 

[TumbleGeorge]

Is this peaks dangerous for current models power supplies? Not to grill it but for to turn off the power supply's due to too sensitive protection systems with a low response threshold?

 

[TheGuruStud]

Is this peaks dangerous for current models power supplies? Not to grill it but for to turn off the power supply's due to too sensitive protection systems with a low response threshold?

They'll be super throttled by bios in OEM crap (hard 125W cap, I bet). And you know the rest will be reduced to laughable perf at 65 and 95W.

All hail the 300W monster of deficiency....and inefficiency

 

[Richards]

Its gonna hit 5.6ghz on the big cores.. so hit needs more juice

 

[Punkenjoy]

Is this peaks dangerous for current models power supplies? Not to grill it but for to turn off the power supply's due to too sensitive protection systems with a low response threshold?

If you have enough power headroom, it wouldn't be a problem, but let say don't yeah, you might get computer shutdown when the load spike too quickly.

By example, if you have a 300w gfx card and a 850w power supply with few devices, there are already enough power for the spike.

But if you have a gfx card that pump 450+ watts and you get the top end CPU, a 850w powersupply might cause problem in some scenario.

The nature of boosting (increasing the power consumption shortly to increase frequency) is not news, but intel look like they want to go even more savage on that front.

 

[watzupken]

Higher power draw on Alder Lake is not going to be surprising for me. I feel Intel is just taking 10nm and going on the same path as their 14nm. Given that their 7nm is unlikely to show up till 2023, they can only pump more power into 10nm and try to push it harder with each passing year. Intel's strategy to go with big/little core config I believe is not because they want to be more efficient. Rather, they can't put too many high power cores either due to insufficient density to fit up to double the number of cores, and likely also how power hungry these cores will actually be. Imagine 16 Golden cove cores @ 5+ Ghz will require too much power and produce too much heat for it to be practical for most users. Even with 8 efficient cores supporting the 8 Golden Cove cores, the power consumption is already expected to be higher based on all the rumors so far. So to me, the primary reason for using the efficient cores is not exactly for efficiency reasons.

 

[zlobby]

But, but intel run cooler and with less power! /s

 

[Jism]

I wonder if they are upping the amps to boost single core performance higher

CPU's just get wider, and thus needing more amps yes for high performance computing tasks.

I mean we already have a CPU clock limit at around 4 to 5Ghz.

 

[dragontamer5788]

The 12V is not directly related to the 45A here, just because 12V x 45A = 540W wich is no number directly tied to anything CPU related.

540W spike, but everything on small-time scales get kinda funny. If they draw 45A for 1-millisecond, then its 540W for that millisecond. Averaged out over the full second, they only draw 160W on the average.

This seems to indicate that Intel's processors have an "instantaneous power draw", requiring the PSUs to be more carefully built to handle these massive spikes