19 releases (10 stable)

new 1.5.0 Oct 17, 2020
1.4.2 Jun 24, 2020
1.4.1 Mar 28, 2020
1.0.0 Dec 23, 2019
0.9.0 Mar 24, 2019

#34 in Network programming

Download history 123/week @ 2020-07-02 74/week @ 2020-07-09 46/week @ 2020-07-16 47/week @ 2020-07-23 28/week @ 2020-07-30 92/week @ 2020-08-06 126/week @ 2020-08-13 83/week @ 2020-08-20 116/week @ 2020-08-27 188/week @ 2020-09-03 89/week @ 2020-09-10 89/week @ 2020-09-17 73/week @ 2020-09-24 120/week @ 2020-10-01 174/week @ 2020-10-08 287/week @ 2020-10-15

449 downloads per month
Used in 9 crates (via solana-core)

Apache-2.0

245KB
6K SLoC

raptorq

Build Status Crates Documentation PyPI

Rust implementation of RaptorQ (RFC6330)

Recovery properties: Reconstruction probability after receiving K + h packets = 1 - 1/256^(h + 1). Where K is the number of packets in the original message, and h is the number of additional packets received. See "RaptorQ Technical Overview" by Qualcomm

This crate requires Rust 1.46 or newer.

Examples

See the examples/ directory for usage.

Benchmarks

The following were run on an Intel Core i5-6600K @ 3.50GHz

Symbol size: 1280 bytes (without pre-built plan)
symbol count = 10, encoded 127 MB in 0.532secs, throughput: 1924.7Mbit/s
symbol count = 100, encoded 127 MB in 0.590secs, throughput: 1734.6Mbit/s
symbol count = 250, encoded 127 MB in 0.572secs, throughput: 1788.4Mbit/s
symbol count = 500, encoded 127 MB in 0.549secs, throughput: 1858.8Mbit/s
symbol count = 1000, encoded 126 MB in 0.599secs, throughput: 1695.5Mbit/s
symbol count = 2000, encoded 126 MB in 0.673secs, throughput: 1509.1Mbit/s
symbol count = 5000, encoded 122 MB in 0.758secs, throughput: 1288.3Mbit/s
symbol count = 10000, encoded 122 MB in 0.953secs, throughput: 1024.7Mbit/s
symbol count = 20000, encoded 122 MB in 1.383secs, throughput: 706.1Mbit/s
symbol count = 50000, encoded 122 MB in 2.041secs, throughput: 478.5Mbit/s

Symbol size: 1280 bytes (with pre-built plan)
symbol count = 10, encoded 127 MB in 0.241secs, throughput: 4248.7Mbit/s
symbol count = 100, encoded 127 MB in 0.160secs, throughput: 6396.5Mbit/s
symbol count = 250, encoded 127 MB in 0.173secs, throughput: 5913.0Mbit/s
symbol count = 500, encoded 127 MB in 0.176secs, throughput: 5798.3Mbit/s
symbol count = 1000, encoded 126 MB in 0.200secs, throughput: 5078.1Mbit/s
symbol count = 2000, encoded 126 MB in 0.208secs, throughput: 4882.8Mbit/s
symbol count = 5000, encoded 122 MB in 0.280secs, throughput: 3487.7Mbit/s
symbol count = 10000, encoded 122 MB in 0.400secs, throughput: 2441.4Mbit/s
symbol count = 20000, encoded 122 MB in 0.494secs, throughput: 1976.8Mbit/s
symbol count = 50000, encoded 122 MB in 0.656secs, throughput: 1488.7Mbit/s

Symbol size: 1280 bytes
symbol count = 10, decoded 127 MB in 0.723secs using 0.0% overhead, throughput: 1416.2Mbit/s
symbol count = 100, decoded 127 MB in 0.701secs using 0.0% overhead, throughput: 1460.0Mbit/s
symbol count = 250, decoded 127 MB in 0.650secs using 0.0% overhead, throughput: 1573.8Mbit/s
symbol count = 500, decoded 127 MB in 0.638secs using 0.0% overhead, throughput: 1599.5Mbit/s
symbol count = 1000, decoded 126 MB in 0.676secs using 0.0% overhead, throughput: 1502.4Mbit/s
symbol count = 2000, decoded 126 MB in 0.764secs using 0.0% overhead, throughput: 1329.4Mbit/s
symbol count = 5000, decoded 122 MB in 0.896secs using 0.0% overhead, throughput: 1089.9Mbit/s
symbol count = 10000, decoded 122 MB in 1.176secs using 0.0% overhead, throughput: 830.4Mbit/s
symbol count = 20000, decoded 122 MB in 1.489secs using 0.0% overhead, throughput: 655.9Mbit/s
symbol count = 50000, decoded 122 MB in 2.633secs using 0.0% overhead, throughput: 370.9Mbit/s

symbol count = 10, decoded 127 MB in 0.713secs using 5.0% overhead, throughput: 1436.1Mbit/s
symbol count = 100, decoded 127 MB in 0.702secs using 5.0% overhead, throughput: 1457.9Mbit/s
symbol count = 250, decoded 127 MB in 0.637secs using 5.0% overhead, throughput: 1605.9Mbit/s
symbol count = 500, decoded 127 MB in 0.613secs using 5.0% overhead, throughput: 1664.8Mbit/s
symbol count = 1000, decoded 126 MB in 0.643secs using 5.0% overhead, throughput: 1579.5Mbit/s
symbol count = 2000, decoded 126 MB in 0.701secs using 5.0% overhead, throughput: 1448.8Mbit/s
symbol count = 5000, decoded 122 MB in 0.826secs using 5.0% overhead, throughput: 1182.3Mbit/s
symbol count = 10000, decoded 122 MB in 1.061secs using 5.0% overhead, throughput: 920.4Mbit/s
symbol count = 20000, decoded 122 MB in 1.380secs using 5.0% overhead, throughput: 707.7Mbit/s
symbol count = 50000, decoded 122 MB in 2.341secs using 5.0% overhead, throughput: 417.2Mbit/s

Public API

Note that the additional classes exported by the benchmarking feature flag are not considered part of this crate's public API. Breaking changes to those classes may occur without warning. The flag is only provided so that internal classes can be used in this crate's benchmarks.

Python bindings

The Python bindings are generated using pyo3.

Some operating systems require additional packages to be installed.

$ sudo apt install python3-dev

maturin is recommended for building the Python bindings in this crate.

$ pip install maturin
$ maturin build --cargo-extra-args="--features python"

Alternatively, refer to the Building and Distribution section in the pyo3 user guide. Note, you must pass the --cargo-extra-args="--features python" argument to Maturin when building this crate to enable the Python binding features.

License

Licensed under

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you shall be licensed as above, without any additional terms or conditions.

Dependencies

~0–530KB
~12K SLoC