You need to consult the specifications. There is absolutely no other way unless you buy the micro, code it and run tests.
So first, you need to define what exactly processing means. Best way to do it is to write the program that does the processing, in assembly. Let's take a hypothetical program in a hypothetical assembly:
load R1, portA
add R1, R1, 1
store portB, R1
First thing you need to know is how many cycles it takes the microcontroller to execute this code. Again, you need to consult the data sheet of the micro. Hypothetical numbers:
- Load takes 3 cycles
- Load from PortA adds 2 cycles overhead
- Add takes 1 cycle
- Store takes 2 cycles
- Store to PortB adds 4 cycles overhead
So you would know that for each piece of data you read (hypothetically, 1 byte) you need 12 cycles to process them, assuming the micro is stupid enough to not use pipelines, prefetching of any sort etc.
If you read the specifications well, you will now get a good estimate of the speed of the micro for your task. In our hypothetical example, 1 byte per 12 cycles.
So to address your first question, how to know if the micro is good enough for your task, it's quite simple. You want
X bytes of data per second so it takes
12X cycles to do that. All you need to know is how many cycles per second the micro runs (let's say
Y). So if
Y > 12X you're good.
Your second question is also simple to answer. You already have the speed in terms of "per cycle". To calculate the speed in terms of "per second", you just multiply the per-cycle speed by cycles/second of the microcontroller (again, found in the data sheet). In our hypothetical example:
speed (byte/cycle) = 1/12
speed (byte/second) = 1/12 * Y
Once you define your task, so it would be clear how heavy the computation is, probably higher grade students or even vendors would be able to help you pick up a suitable micro.