I have two groups of data: In every group there are 29 subgroups, in every subgroup, there are 1-15 numbers ranged from 0 to 80. The subgroups are DEPENDENT of each other.
I wanna compare whether these two groups are sgnificantly different
without mixing all numbers in subgroups together. How should I do the analysis?

To make one example of this question, suppose:
2 groups, in every group there are only 3 subgroups, in every subgroup
there are at most 5 numbers ranged from 0 to 80:

group 1:
subgroup a: 0.1, 23, 44, 12.5, 5.0
subgroup b: 24
subgroup c: 23, 44

group 2:
subgroup a: 32, 45, 5.5, 6.7
subgroup b: 22.2, 45, 56, 52, 10
subgroup c: 2.2, 4.5, 6.1, 32

In every group, the 3 subgroups are not independent of each other. How can I compare whether group 1 and 2 are significantly different from each other, without mixing the 8 numbers in group 1 together and mixing the 13 numbers in group 2 together?

You are using "group" as a synonim for "set" or "collection."
Correct?

(Be aware: "group" and "subgroup" has a very precise, *technical*
sense which is likely to confuse the hell out of most mathematician
reading your post unless you mean that technical sense, in which case
you will confuse *all* mathematicians reading your post because I
cannot fathom how you could possibly mean that without specifying the
operation).

Also, it appears that you are using the word "independent" in a way
that is not standard in mathematical writing. Do you mean that the
sets are not pairwise disjoint? that is, that there may be a number
that is in two or more of the sets in one of the two collections?

I second Arturo's suggestion that you not use the words "group"
and "subgroup" when you don't intend them with the usual meanings
given to them by mathematicians. You'll note I've used "collection"
and "set" instead.