Previously we looked at dynamic clock gating, but we can also turn off power supply to regions of a chip with fine or coarse gain, creating so-called power islands. We use power gating cells in series with supply rails. These are large, slow, low-leakage transistors. (Best to disconnect the ground supply since an N-channel transistor can be used which has smaller area for same resistance.)
Signal isolation and retention cells (t-latches) on nets that cross in and out of the region are needed. There is no register and RAM data retention in a block while the power is off.
This technique is suitable at coarse grain for complete sub-systems of a chip that are not in use on a particular product or for quite a long time, such as a bluetooth tranceiver or audio input ADC.
It can also be used on a fine grain with automated control similar to clock gating.
However, power gating requires some sequencing to activate the enables to the isolation cells in the correct order and hence several clock cycles or more are needed to power up/down a region. Additionally, gradual turn on over tens of milli-seconds avoids creating noise on the global power rails. Originally, power off/on was controlled by software or top-level input pads to the SoC. Today, dedicated microsequencer hardware might control a hundred power islands within a single subsystem.
A common practice is to power off a whole chip except for a one or two RAMs and register files. This was particularly common before FLASH memory was invented, when a small battery is/was used to retain contents using a lower supply (CMOS RAM data holding voltage). Today, most mobile phones, laptops and PC motherboards have a second, tiny battery that maintains a small amount of running logic when the main power is off or battery removed. This runs the real-time clock (RTC).
|19: (C) 2008-13, DJ Greaves, University of Cambridge, Computer Laboratory.|