What are the differences between GSVA and GSEA?

I got some idea from a toward data science post (link):

"GSVA builds on top of GSEA where a set of genes is characterized between two condition groups defined in the sample. GSEA works on how genes are behaving differently between the two groups defined."

from my own understanding,

• GSEA first analyze the differences between the tumor vs normal group for each gene; then rank them and calculate the enrichment scores.

• GSVA instead, first analyze each gene among all groups (including both tumor and normal) to get a distribution for each genes, then rank them and calculate the GSVA scores.

But what is the goal or usage for each of these methods?