Intel® Dynamic Platform and Thermal Framework (Intel® DPTF) is a platform level hardware/software solution for power and thermal management. As a container for multiple power/thermal technologies, Intel DPTF provides a coordinated approach for different policies to effect the hardware state of a device. The high level architecture of Intel DPTF is depicted in the figure below.
Intel DPTF relies on BIOS ACPI implementation on a specific device/system. The BIOS Reference Guide provides implementation details for Intel DPTF and all the supported technologies. Please refer to the Documentation section for BIOS Reference Guide document (to be published soon).
Technologies that are currently supported as part of Intel DPTF are mainly focused on thermal management as specificied in the ACPI 5.0 specification. The thermal management technology is called Intel® Dynamic Power Performance Management (Intel® DPPM). While Intel DPPM is an implementation of the ACPI thermal model, Intel DPPM algorithms for thermal management are different than a OSPM thermal management implementation.
DPPM is an intelligent closed-loop hardware and software solution for managing platform thermals in a constrained design. By allowing an OEM design team to control platform skin and component temperatures, it is possible design for a “typical” rather than “worst case” power scenario while minimizing the impact to performance.
On a non-constrained system, Intel DPTF software works as a replacement for the typical thermal management.
The Intel DPPM technology implements a series of Policies that can be used to better control platform behavior at thermal event trigger. The following policies exist:
The Passive Policy algorithm depends upon the platform thermal characteristics defined in the ACPI _TRT and _TRM objects. Using the details in the _TRT, the Passive Policy is responsible for limiting power and performance of components in response to a participant’s temperature rising above the platform defined _PSV trip point. Please see the latest ACPI specification for details on the objects.
The Active Policy algorithm cools a platform through the removal of heat instead of limiting the power or performance of the device. In many of these cases, Active Policy enables a platform fan at various speeds in response to the thermal conditions defined in the APCI _ART and _ACx objects trip points. Please see the latest ACPI specification for details on these objects.
In the event the platform reaches a critical temperature, the Critical Policy is responsible for gracefully shutting down the system. This is accomplished through the use of the ACPI _CRT and _HOT objects, as defined in the ACPI specification. Depending upon the critical event, the system can be moved to a hibernate (S4) state or shutdown (S5).