Confused between Temporal and Spatial locality in real life code

Question

I was reading this question, I wanted to ask more about the code that he showed i.e

for(i = 0; i < 20; i++)
    for(j = 0; j < 10; j++)
        a[i] = a[i]         


        

Answer 1

Writing this answer as I didn't get it even after reading the other answers on this question, a few other questions and wikipedia (that's more confusing.)

I think we spend a lot of time and energy to understand the terminology which is bit confusing/complex in this case. I found it easier to understand when I didn't pay any heed to the terms 'spacial' and 'temporal'.

Let's start with the basics.

Let's try to understand what the cache is - a place which is quicker to access than the main memory. That's cool. But this place is limited and expensive, so one should use it wisely. But how would you (or OS) decide what to put in cache and what not to put? There should be some way to know what would we need in the future.. ah future predictions! ( Minority Report! Ring some bells?).

There should be some way to determine what would the program need in future. Using common sense and the code, we can say that some parts of the code are repetitive in nature - example - loops! If there is a variable i inside a loop you know it's going to get accessed in near future again and again. This is the principle behind temporal locality. i can be brought into cache as it is temporally local.

In another area if the code is using any linear data structure (example: an Array) and that too in a loop with an increment in the index, then it's easy to see that although the current need is only 3rd location (for example) of this data structure, very soon the next locations would also be needed because the index increases by 1 for that linear data structure. So it would be great if we bring in the data in the next few locations as well. This is the principle behind spacial locality. Next few locations can be brought into cache as they are spacially local.

The concept of locality is basically to identify the data and instructions to bring in cache so that we can reduce the cache misses and utilize this special place effectively.