When we dive into per-core power loading, we get the following:
The big chip’s power distribution seems to go up in that 3-4 core loading before coming back down again. But as we load up the second chiplet moving from 8 to 9 core loading, it is worth noting that the second chipset is reporting lower core power, despite showing the same core frequency. AMD is able to supply the two chiplets different amounts of voltage and power, and we might be seeing this play out in real time.
Perhaps very important is that single core power consumption when we are at 5050 MHz of 20.6 W. Going back to our previous generation data, on Zen 2 we were only seeing a peak of 18.3 W, and a slightly higher voltage reported (1.45 V for Zen 2 vs 1.42 V for Zen 3). This means that from the perspective of our two chips, Zen 3 cores scale better in frequency, and even though the power increases as expected, the voltage simultaneously decreases (Note that there can be some silicon variability to also account for some of this.)
Moving down the stack, the 12-core Ryzen 9 5900X doesn’t show any surprises – we’re seeing the same drop off as we load up the cores, this time as we go beyond eight cores. As this processor uses two chiplets, each with six cores, that second set of six cores seem to be consuming lower power per core as we add in additional load.
Some users might be scratching their heads – why is the second chiplet in both of these chips using less power, and therefore being more efficient? Wouldn’t it be better to use that chiplet as the first chiplet for lower power consumption at low loads? I suspect the answer here is nuanced – this first chipet likely has cores that enable a higher leakage profile, and then could arguably hit the higher frequencies at the expense of the power.
Moving down to a single chiplet but will the full power budget, and there is some power savings by not having the communications of a second chiplet. However, at 8-core load, the 5800X is showing 4450 MHz: the Ryzen 9 processors are showing 4475 MHz and 4500 MHz, indicating that there is still some product differentiation going on with this sort of performance. With this chip we still saw 140 W peak power consumption, however it wasn’t on this benchmark (our peak numbers can come from a number of benchmarks we monitor, not just our power-loading benchmark set).