As the other posters have already noted, individual switch can vary greatly on their interior switch hardware supporting their ports and can have different resources supported "normal" edge ports vs. "uplink" ports. I.e. switch X with 48 ports and switch Y also with 48 ports can be very different in how well they can support their 48 ports. Basically, for the same number of "like" ports, if switch X is much more expensive than switch Y (from same vendor), there's usually internal hardware reasons for the increased cost (although more expensive devices also often have a larger profit margin in their pricing too).
What capacity you need depends on the nature of the hosts, and their topology. For instance, user hosts often tend to have less traffic then server hosts, but even that can vary based on the "kind" of user or server hosts. (For example, years ago I supported an engineering department [with CADD operators], that generated more traffic on our local/department LAN than the company [a fortune 100] had on their LAN backbone.)
As you noted, a "cascade" type of topology could more easily fill transit links with traffic, as traffic can "accumulate" as your transit the switches. (BTW, this type of topology also often increases latency, as there can be more switches to transit between hosts.)
From a topology viewpoint a "star" or "mesh" topology is often "better" for performance, but without knowing your traffic "needs", low-end switches, even in a "cascade" (daisy-chain) topology, might be perfectly fine, for you.
