fpc is a Go implementation of Burtscher and Ratanaworabhan's 'FPC' algorithm for compressing a stream of floating point data.
The FPC algorithm can losslessly encode and decode huge amounts of floating-point data very quickly. It scales well to gigabyte-per-second streams. Compression ratios are better than just about any generic compressor like gzip or bzip, and compression and decompression throughput are much better (like, 8x to 300x faster) than other algorithms. For more on this, the paper introducing FPC is really readable - I highly recommend it!
fpc provides a Writer
and a Reader
, following the pattern set by
the Go standard library's compression packages. The Writer wraps an
io.Writer that you want to write compressed data into, and the Reader
wraps an io.Reader that you want to read compressed data out of.
Since FPC encodes streams of float64s, they impose some additional
expectations on callers: when calling Reader.Read(p []byte)
or
Writer.Write(p []byte)
, the length of p
must be a multiple of 8,
to match the expectation that the bytes represent a stream of 8-byte
float64s.
In addition, utility methods are provided: Reader
has a
ReadFloats(fs []float64) (int, error)
method which will read bytes
from its underlying source, parse them as float64s, put them in fs
,
and return the number of float64s it placed in fs. When it reaches the
end of the compressed stream, it will return 0, io.EOF
.
Similarly, Writer
has a WriteFloat(f float64) error
method which
writes a single float64 to the compressed stream.
In benchmarks on a fairly vanilla laptop, reading or writing from an
in-memory stream, fpc
is able to encode at about 1.2 gigabytes per
second, and it can decode at about 0.9 gigabytes per
second. Benchmarks can be run on your own hardware with go test -bench "Read|Write" .
.