Optimizing CPU Power Consumption Comparison: Choosing the Right HWiNFO Metrics for APUs
In the pursuit of evaluating CPU power efficiency, especially during intensive workloads like Cinebench R24 multi-core tests, selecting the appropriate measurement metric is crucial. With the proliferation of APUs and integrated systems, understanding which HWiNFO metrics best represent actual CPU power consumption can help users and reviewers make fair and accurate comparisons.
Understanding the Available Metrics in HWiNFO
HWiNFO offers several metrics to gauge power usage:
- CPU Cores Power: Reflects the power consumed solely by the CPU cores.
- CPU Cores + SoC Power: Includes the cores and the System on Chip components, such as integrated controllers and other subsystems.
- CPU Package Power: Represents the total power drawn by the entire CPU package, encompassing cores, iGPU, cache, and other integrated components.
The Challenge of Choosing the Appropriate Metric
While many in the benchmarking community tend to rely on the “CPU Package Power” metric, this approach warrants scrutiny—particularly for APUs and integrated processors. The total package power often includes the power draw of components like the iGPU and integrated memory controllers, which may not be active or relevant during certain stress tests aimed solely at evaluating CPU cores.
For example, during a Cinebench R24 multi-core run, an advanced APU such as the AMD Ryzen AI MAX 300 series or the Apple M4 series might report a package power reading of approximately 70W. However, HWiNFO could show only 30-40W for “CPU Cores + SoC” during the same test. The discrepancy arises because the total package power includes additional subsystems not involved in the specific CPU core workload.
Implications for Fair Comparison
If the objective is to evaluate the raw power efficiency of the CPU cores under load—particularly when comparing different architectures or chipsets—it makes sense to focus on metrics that isolate core activity. Using “CPU Cores Power” or “CPU Cores + SoC Power” provides a clearer picture of how much energy the CPU cores alone consume during computational stress.
However, if the goal is to assess overall thermal and power characteristics of the entire package, including integrated graphics and other systems, then “CPU Package Power” remains relevant.
Recommendations for Accurate Testing
- For CPU-Only Load Analysis: Measure “CPU Cores Power” or “CPU Cores + SoC
Share this content:
