![]() 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. The big chip’s power distribution seems to go up in that 3-4 core loading before coming back down again. When we dive into per-core power loading, we get the following: We might go back in future on other boards to see if this is consistent. This might be an odd quirk of our specific chip, our power test, or it might be motherboard or BIOS specific (or a combination of several factors). There’s also a bigger increase in non-core power, up from 16 W to 21 W, which perhaps decreases the power to the cores, reducing the voltage. Looking into the data, the frequency of the active cores drops from 4725 to 4675, which isn’t a big drop, however the voltage decreases from 1.38 V to 1.31 V, which seems to be more sizeable drop than other voltage readouts as we scale the core-to-core loading. The second feature is an odd dip in power moving from 4 to 5 cores loaded. ![]() We see this effect on the 5900X as well, perhaps indicating this is a feature of the dual chiplet design – we’re not seeing it on the 5800X or 5600X. We’re seeing the difference between the two values also increasing slightly, as more data is transferred over those off-chiplet communications. We saw this when we first tested the previous generation 3950X, and is indicative of how the processor has increased current density as it loads up the cores, and as a result there’s a balance between the frequency it can give, delivering the power, and applying the voltage in a consistent way. There are two significant features of this graph.įirst is the hump, and a slow decrease in total package power consumption after 8-10 core loading. The difference between the two covers the IO die as well as any chiplet-to-chiplet communications, PCIe, CPU-to-chipset, and DRAM controller consumption. ![]() Here we are reporting two of the values that we have access to on the chip, which the chip estimates as part of its turbo detection and action algorithms: total package power (for the whole chip), and the power solely used by the sum of cores, which includes the 元 cache. If we look directly at the Ryzen 9 5950X for chip wide power consumption over per-core loading, we get this following graph. For the sole 65 W processor, the PPT value is 88 W, and we’re seeing only 76 W, showing some of the efficiencies on the Ryzen 5 5600X. For these processors, we can see our peak power consumption through our testing matching that value. All turbo modes and power modes above that are not covered by warranty.įor AMD’s new Ryzen 5000 processors, most of them have a 105 W TDP, with a Package Power Tracking (PPT) setting of 142 W. In simple terms, processor manufacturers only ever guarantee two values which are tied together - when all cores are running at base frequency, the processor should be running at or below the TDP rating.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |