For any particular port, increasing drop stats is often a great indication the port is being overwhelmed. (Also noted by @MHM Cisco World)
If CPU performance is very high to maxed out, that often indicates some issue beyond traffic volume, itself, as switches have dedicated hardware for switching.
Head-of-line blocking, or fabric blocking, can cause poor performance, but troubleshooting those is more involved than I want to delve into here, beyond . . .
Head of line blocking architectures is rare, generally, with modern switches.
Fabric blocking happens when the aggregate volume of traffic, across multiple switch ports, exceeds the fabric's bandwidth capacity. Most modern Enterprise grade switch fabrics have sufficient bandwidth for all ports to operate at full capacity, concurrently.
