Power Consumption, Peaks and technical Details: Intel’s upcoming Alder Lake S and Raptor Lake S CPUs in an exclusive comparison

Disclaimer: The following article is machine translated from the original German, and has not been edited or checked for errors. Thank you for understanding!

Now let’s get to the interesting part, but you can only get it right if you’ve read the brief introduction to IMPV9.1 on the previous page. To protect my sources, I’ve transferred the heavily watermarked slides into Excel and recreated them. The only difference is that I’ve now separated the TDP stages separately and it’s no longer all lined up next to each other. For the colors I used the original.

Let’s first look at the 125-watt class, where PL1 is well known. PL2 is 188 watts at Alder Lake S (ADL S) and can increase to 241 watts in the performance scenario, depending on the situation resulting from the power telemetry. For Raptor Lake S (RPL S) the PL2 in Perf (Performance) even rises up to 253 watts, the baseline again stays at 188 watts. And what does the PL4 do as an interval with a maximum duration of 10 ms? The baseline is 238 watts for both CPU generations and can go up to 359 watts for ADL S! RPL S, on the other hand, is somewhat more frugal with 314 watts. You can see that the PL2 is higher and the PL4 is somewhat lower with RPL S than with ADL S. This is interesting in any case.

In the power class below this case is repeated, because the PL1 is the same for both, but here the values for PL2 and PL4 differ more. Because the baseline is also different now. For PL1 and PL2, baseline and perf are generally higher for RPL S than for ADL S, while PL4 is generally lower. So the extreme power peaks become much more moderate with RPL S, while the rest increases.

In the 35-watt range, on the other hand, RPL S is virtually on par with the PL2 and far below the values of ADL S in the PL4, which raises interesting questions about frequencies and also the possible number of cores. But we can save that speculation for now, because a lot can change between now and then.

Summary and conclusion

So power supply recommendations are always such a thing, as long as the underlying details are not known. Also the losses in the voltage transformers will never be so high that e.g. with ADL S the threatened 40 amperes are really reached. Of course, everything is possible, but you should never ignore that such lists for power supply manufacturers are always guidelines with a lot of reserves, because Intel probably also includes rather inferior power supplies in their calculations.

But load peaks are load peaks and in combination with those of the current and upcoming graphics hardware they are often unpredictable. Then, if the OCP/OPP reacts too quickly, there will be emergency shutdowns that really shouldn’t have been necessary. Here it applies then once more to read before genuine net part tests and to make oneself really clever. Because wattage or amperage numbers for individual rails don’t say anything, unfortunately, and I’ve been missing trigger intervals in the specs for the customer for a long time.

It will certainly be interesting when ADL S enters the market and it will also be important to check these load peaks and to measure the behavior of the CPUs in concrete situations also in high resolution. Unnecessary panic in advance is not advisable, however, because even Intel will certainly know what they are doing and, above all, why.