Latency is the end-to-end time taken to perform a single computation. If Compute1 represents a single computation that starts at t1 and ends at t2, then we can say the following:
Latency = t2-t1
Latency is the end-to-end time taken to perform a single computation. If Compute1 represents a single computation that starts at t1 and ends at t2, then we can say the following:
Latency = t2-t1