computer-architecturelow-powermicroprocessor
There's something I don't get – I am always under the impression that the more transistors we pack in, the more energy it consumes and the hotter it gets (assuming we haven't shrunk the die). However, the Apple A8 has around 3 billion transistors while the Intel Haswell has half the transistor count, yet the A8 is able to run cool and takes much less energy to compute.
Can anyone explain the difference?
A lot of things that give you more power just require more transistors to build them. Wider buses scale the transistor count up in almost all processor components. High speed caches add transistors according to cache size. If you lengthen a pipeline you need to add stages and more complex control units. If you add execution units to help mitigate a bottleneck in the pipeline, each of those requires more transistors, and then the controls to keep the execution units allocated adds still more transistors.
The thing is, in an electronic circuit, everything happens in parallel. In the software world, the default is for things to be sequential, and software designers go to great pains to get parallelism built into the software so that it can take advantage of the parallel nature of hardware. Parallelism just means more stuff happening at the same time, so roughly equates to speed; the more things that can be done in parallel, the faster you can get things done. The only real parallelism is what you get when you have more transistors on the job.
Most micros support a low-power 32.768 kHz watch crystal oscillator with some kind of prescaler and timer interrupt. Set the prescaler so the timer is counting slowly and the interrupt happens at the period you desire.
Some micros also have a built-in low-power RC timer if exact timing isn't critical.
The datasheet for any low-power micro will list the power with 32.768 oscillator (and nothing else) running. It's pretty close to zero. You can do the math to see if this is acceptable, and compare it to the current drawn by the watchdog.
OK, for example on the msp430f2013, let's look at power in the datasheet.
0.5 μA is almost zero, although it is five times the true OFF mode.
For more detail, we can look inside the datasheet.
Going from LPM4 (everything off) to LPM3 (running the oscillator) is the difference between 0.5 μA and 1 μA.
Suppose the battery is CR2032 with 225 mAh capacity. Then standby in LPM4 is about 50 years and in LPM3 is about 25 years. 25 years is long enough for many applications, because the ON-current (during the measurement itself) dominates the consumption.
Best Answer
It is the clock frequency, i.e. frequency of the switching, and the number of transistors actually being switched at the same time that consume the power, and consequently, generate the heat, and not the number of transistors.
For example there could be a large amount of transistors in a area of the die that aren't being used, and so, if they are in the off (closed) state, then they will not draw any current, so less energy and less heat.
However, if some active area is switching at a higher clock frequency, then (on average) they are open for more of the time, and so draw more current, thus energy and heat.
Hope that helps.