Power capping must have done differently among these two kinds of applications. Hello, i’m looking at performance variations of my application under a range of power caps. It is easy enough to monitor the actual unhalted processor cycles to determine the average frequency. Skip to main content. Where can i find official information?

Uploader: Jusida
Date Added: 20 May 2009
File Size: 8.67 Mb
Operating Systems: Windows NT/2000/XP/2003/2003/7/8/10 MacOS 10/X
Downloads: 30329
Price: Free* [*Free Regsitration Required]

Added to drivers build bitops: I often find it difficult to figure out which parts of Chapter 14 of Volume cappimg of the Intel Architectures SW Developer’s Manual actually apply to my systems. I checked duty cycling and clock modulation.

Power Capping Framework and RAPL Driver []

Soon it is very likely that other vendors are also adding or considering such implementation. Fri, 4 Oct With a low power cap, the perfromance is very bad. Oower in retired instruction rate may indicate hardware clock cycle modulation. There are several use cases for such technologies: Log in to post comments.

RAPL power capping: how does it work

Each device can report its power consumption. Setting power limits on the devices allows users to guard against platform reaching max system power level. My questions is how RAPL caps the power.

Depending on your processor, another place to look for DVFS is in the “uncore” clock. Leave a Comment Please sign in to add a comment. Add class driver PowerCap: While staying below a power limit, it allows devices to automatically adjust performance to meet demands – Dynamic control and re-budgeting: Intel is slowly adding many devices under RAPL control.


Someone told me, if the power cap is high, it does DVFS. If the power limitation is low, it manipulates clock duty cycles.

RAPL power capping: how does it work

At least some processors will support frequencies below the “maximum efficiency” frequency, but I don’t know if the Power Control Unit will use these or switch to duty cycle modulation Section Also there are other technologies available, for power capping various devices.

I read through the two sections in the manual, still I’m not very clear what is the difference between duty cycle modulation and clock cycle modulation? Setting DVFS to 1. In one instance, I noticed that when running a power-limited application on a Xeon Platinum system, the uncore frequency was reduced by a much larger percentage than the core frequency.

I seem to recall one case with severe thermal throttling that was running very slowly as if duty cycle modulation was being employedbut I never saw any changes to the corresponding MSRs.

If the power limit is still exceeded at the “maximum efficiency” frequency typically 1. Where can i find official information? I can conform your observation with compute-bound applications.


One possible interpretation is that the MSRs show software-controlled duty-cycle modulation, but may not show hardware-initiated duty-cycle modulation. Power Capping framework is an effort to have a uniform interface available to Linux drivers, which will enable – A uniform sysfs interface for all devices which can offer power capping – A common API for drivers, which will avoid code duplication jntel easy implementation of client drivers.

Power Capping Framework and RAPL Driver

With a high power cap, the performance is reasonable. Setting power cap directly to 75 watts no DVFS and turbo boost is onthe frequency is 1.

If power cap is set to 45 watts no DVFS and turbo boost is onthe observed frequency is more or less 1. For more complete information about compiler optimizations, see our Optimization Notice.

If each device can be constrained to cappnig power, extra power can redistributed to other devices, which needs additional performance.