This is an implementation of the 1BRC parsing / aggregation challenge is Rust, using mainly the standard library (and
memchr to find newlines).
It runs a single reader thread, which fills up buffers that are read from by parser threads. The parsing process is generally zero-copy: after data is read from the file by the reader thread, the data is only copied once per unique station name (per thread, to fill up the HashMap). This is accomplished by using a wrapper around the standard HashMap that allows for using the Entry API with a reference type.
On my machine (M2 Macbook Air), it runs in read time: it takes ~8s to read in the file without parsing, and the same amount of time to also parse the file. With faster SSD's, your mileage may vary. This implementation runs around twice as fast as the reference implementation (~20s) on my machine.