| || |
The thing is, we don't all have EXACTLY the same hardware. It varies on a nanometer level. For example, the processor in our phone is based on a 65nm process. That means the smallest thing can can be produced on that process is 65nm. That's about the size of one molecule of a virus (an actual biological one). At that scale, a few extra atoms here or there can make a difference, which is why some phones with the same processor can clock higher than others.
Originally Posted by cpiddy
As an example, consider your car. Every car like yours has the same hardware, so you'd expect them all to break down in the same way at the same time. That simply isn't the case. There may be trends (one type of car may be more reliable than another), but for each individual car some things will be different.
Some cars may be lemons. The same applies to processors. Occasionally there may be a "bad" batch, which simply won't go as fast as others. The manufacturer simply guarantees a processor will go at least XXX Mhz (or Ghz) fast. It may run faster with some tweaking, but anything more you get is a bonus, not a given. In fact, "bad" batches of processors is so common that manufacturers simply take the bad ones and sell them as a cheaper, slower, part even though they are physically identical (in terms of design, architecture, and the process by which they are made) but are different enough on a microscopic level that one can run faster than another. This process is called "product binning" if you want to google it and read up on it.
Having said all that, it is possible for the software to make some difference but the absolute physical limits aren't determined by your kernel or setcpu. They are determined by what your processor can physically handle. In my experience Xionia is well sorted and stable and setcpu as well, so whatever you can hit with those is probably pretty close to the absolute limit.