HyperBuf

A dynamic and highly optimized buffer with atomic locking mechanisms and asynchronous memory management (async/await ready; use nightly!)

HyperBuf is about 40% faster than BytesMut, and 11% faster than std::vec!

For memory retrieval, instead of spin-waiting and blocking the thread, the system uses an atomically-backed and asynchronous model that has the capacity to treat the data internally as any arbitrary type (Type system pending). This is especially useful for writing a stream of network bytes to a custom Packet type.

There are three ways to interact with the data, and it is up to the programmer to make the wisest decisions:

Direct treatment of the system as a u8 buffer, or;
Asynchronous casting of type to an immutable yet readable version (via ReadVisitors), or;
Asynchronous casting of type to a mutable thus writable version (via WriteVisitor's)

The rule for consistency is simple: if you choose to treat the type as a buffer, you should NOT use Write/Read Visitors. However, keep in mind that, for performance reasons, this check is NOT made programatically! It is up to YOU to design your program correctly around this model

When you use a WriteVisitor, you should specify the amount of bytes you plan on writing when calling visit(). If you don't plan on making the type grow, you can simply enter None.

When using indexes or the put_u8-like methods, HyperVec is faster than the std::vec, and faster than BytesMut:


[System Information]
    Operating System: Windows 10 Enterprise Insider Preview 64-bit (10.0, Build 18917) (18917.rs_prerelease.190607-1942)
    BIOS: 1301 (type: UEFI)
    Processor: Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz (8 CPUs), ~4.2GHz
    Memory: 16384MB RAM
    Available OS Memory: 16326MB RAM
    Page File: 15402MB used, 8859MB available

Vec benches/std vec     time:   [19.311 ns 19.317 ns 19.322 ns]
                        change: [-0.0808% -0.0337% +0.0096%] (p = 0.15 > 0.05)
                        No change in performance detected.
Found 4 outliers among 100 measurements (4.00%)
  1 (1.00%) low mild
  3 (3.00%) high mild
Vec benches/HyperVec    time:   [17.464 ns 17.480 ns 17.498 ns]
                        change: [+0.4541% +0.6362% +0.7789%] (p = 0.00 < 0.05)
                        Change within noise threshold.
Found 12 outliers among 100 measurements (12.00%)
  4 (4.00%) high mild
  8 (8.00%) high severe
Vec benches/BytesMut    time:   [24.262 ns 24.276 ns 24.290 ns]
                        change: [-0.3091% -0.2340% -0.1585%] (p = 0.00 < 0.05)
                        Change within noise threshold.
Found 3 outliers among 100 measurements (3.00%)
  1 (1.00%) low mild
  2 (2.00%) high severe

tbraun96/HyperBuf

HyperBuf