Preßy,

ls -h is available only on GNU ls. So that is not a very general answer. Besides, the real thrust of the question is the mental conversion of bytes to megabytes.

3499990308 as displayed by ls represents bytes. To get it to megabytes, I want to drop the last 6 digits. 3499990308 is sort of how I "see" the number. Now I do evaluate that 990308 as I throw it away. I do that really by looking just at the leading nine. Since I am throwing away a "large" number, I add one to the number that I keep. 3499 + 1 = 3500 MB. Whether or not to add that one depends on the leading digit of what I threw away. If it's 5,6,7,8, or 9 then I add the one.

This is a compromise, I would use a more accurate algorithm if I were writing a program. But I can do this instantly and effortlessly.

Effectively, what I have done is to divide the number by 1,000,000. Perfect accuracy would have required dividing by 1,048,576. Using the bc program I see that the real answer is 3338 MB. With a file this big, I don't care about a lousy 162 MB.

29994 is too small. Well, ok, it is 0 MB. I simply never want to convert tiny sizes to megabytes. But kilobytes would work in this case. The rule is similiar to the above, but here I just drop 3 digits rather than 6. So I estimate that file at 30 KB. bc says the real answer is 29 KB. Again, close enough.

Notice that both of my estimates were high. That is guaranteed to be the case.