The digestion time is not related to the size of the DNA. Your supervisor probably suggested a longer digestion just to ensure that no undigested PCR product remains.

If you look up the manual for the restriction enzyme that you are using it will tell you recommended digestion times. Usually 1 hour is the minimum, but I regularly digest overnight if I want to make sure there is no undigested product. The downside of digesting for longer is that it increases the chance of star activity (where the restriction enzyme cuts at an unexpected site), but for most restriction enzymes from commercial sources there is a low chance of star activity.

I think that the logic is that 1 unit of RE is defined by the time that it takes to cut a defined amount of dna. This dna is often human dna. If the enzyme is a 6 cutter and if we assume that the distribution of bases in human dna is random then a 6 cutter cuts every 4 exp6 or every 4096 bases but in your pcr product which is 500bp there may be one cut site in every molecule so the enzyme is having to cut 8 times as many cut sites for any given amount of dna so it is wise to give the enzyme longer so that it can cut all of the RE sites. In genomic dna partials are not very obvious but partial digests of a linear pcr product just look uncut which may lead you to think that the molecule has not cut therefore the cut site does not exist so you will charaterise the pcr product as having the wrong genotype

One other issue to add to the prior responses is that if the restriction site is near the end of the PCR product (such as on the primer) then there will be a reduction in efficiency of cleavage. Restriction sites near the end of a fragment are much less efficiently digested.