Unless we like to waste money or are very ignorant people who usually leave the lights on, no one likes to waste energy and the same can be said of the other basic elements of daily consumption. In the case of processors, on the other hand, this does not come from the economic and environmental consciousness of the designers, but rather a product of the laws of physics and especially of the mode of transmission of electricity.
Let us not forget that a processor today is only an electrical circuit miniaturized to the maximum. In other words, today we are talking about billions of transistors, each of them having to be powered autonomously and simultaneously with the others.
The most complex part when designing a new CPU or GPU today is not to come up with an architecture that achieves a certain power on paper, but that its physical and actual implementation is able to achieve the performance expected in a specific consumption. One of the tricks to achieve this is the partial deactivation of the processor i.e. some parts of it are turned off while others are kept on and this ability can be achieved on the fly and, therefore, with the fully functional CPU or GPU.
Power consumption of a processor
The formula for the power consumption of each of the transistors in a processor is as follows:
Power = Q * C * F * V2
Or:
- Q is the percentage of time that the transistor remains active.
- It is the electronic capacity, and therefore the level of charge that a semiconductor can withstand.
- F is the frequency or clock speed
- V is the voltage
Until 2005, the rules for designing a processor were the same in terms of power consumption per zone. The idea is that if the area of a processor was halved, the scale ratio “S” behaved similarly, which was calculated with the following formula: S = √2
With this we can already understand the following table:
In the era of Dennard scaling, the voltage consumed by each transistor decreased with each manufacturing node, but the point was reached where the voltage stopped decreasing. The result has been that energy consumption has skyrocketed and not only, and this is where the problem comes from: the amount of wasted energy converted into heat has increased. Suddenly it was no longer possible to increase the clock speed of the processors to increase performance, forcing the creation of multicore systems. As for the graph, it is the consequence of the end of Dennard’s scaling.
What is partial deactivation?
Okay, we already know that processors have the problem that the power consumption after a certain point has exploded and has become impossible to scale. The consequences of this have been the so-called black silicon, which is based on the fact that certain parts of the processor are turned off while performing certain operations to maintain a constant power budget. Due to the phenomenon we talked about in the previous section, the area attributed to black silicon has increased over time.
On the other hand, partial deactivation, although similar in concept, is different, as it relies on deactivating entire and one hundred percent functional blocks of a processor and not specific parts. For example, when a GPU doesn’t decode video through its hardware codec, it turns it off, or when a processor is idle but not completely turned off, it just turns off all but one of the cores.
Partial deactivation is therefore the equivalent of the processor turning off the lights in a room if no one is using it or leaving the water faucets closed. Although it is not only faced with the tension, but also with the heat generated by the active parts. If an element next to another receives energy, it therefore produces heat and can thermally suffocate neighboring components. This is why, when faced with overclocking or increasing the clock of a processor, entire cores are often disabled several times.
How is partial deactivation implemented?
Here we enter the concept of the so-called power or voltage islands, the implementation of which is carried out during the design period of any processor. This idea is to design the circuits of different parts of the processor at different frequencies and voltages independent of each other.
The idea of having different voltages in each part of a processor makes sense because not all of them need the same power to operate and this ends up leading to higher power consumption in areas where it doesn’t. is not necessary. Well, partial deactivation is based on a very simple concept: to make it so that in each of these power islands, the voltage can be changed until it is zero. Likewise, this allows the CPU to increase and decrease its voltage to do the same with the clock speed if necessary.
When we access the BIOS to enable or disable parts of the processor, keep in mind that this is a partial activation and deactivation of it, which today we can do in flight and without having to turn off the processor. At the manufacturing level, many processors are partially disabled compared to models with the same architecture, in this case what is done is to disable the circuits that power this part, which prevents it from functioning.
This is easy to see in Ryzen processors with their Ryzen Master software and now with Intel Alder Lake this is also happening.
