1
41%
master: 41%

Ran 05 Mar 2019 01:52AM UTC

Files 10

Run time 2s

Badge

Embed ▾

Committed 05 Mar 2019 01:50AM UTC coverage: 42.627% (-0.2%) from 42.799%

Job # GO111MODULE=on

Build Type

push

travis-ci

Committed by

kixelated

Commit Message

Buffer all reads instead of message passing

This avoids channel/lock coordination in favor of performing a copy and
always adding the packet to a queue. Obviously I hate performing copies,
but I think this is the fastest approach with the current API.

I wrote an implementation similar to the old code for comparison sake.

```
name                old time/op    new time/op     delta
PacketBuffer14-8       859ns ± 5%      129ns ± 4%   -84.93%
PacketBuffer140-8      832ns ± 4%      154ns ± 4%   -81.43%
PacketBuffer1400-8     825ns ± 8%      351ns ± 4%   -57.49%

name                old speed      new speed       delta
PacketBuffer14-8    16.3MB/s ± 5%  107.8MB/s ± 3%  +561.08%
PacketBuffer140-8    168MB/s ± 4%    904MB/s ± 4%  +436.70%
PacketBuffer1400-8  1.70GB/s ± 8%   3.99GB/s ± 4%  +134.74%
```

Run Details

318 of 746 relevant lines covered (42.63%)

10.9 hits per line

pions / srtp / 71 / 1
41%
master: 41%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 71.1 (GO111MODULE=on)

pions / srtp / 71 / 1 41% master: 41%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 71.1 (GO111MODULE=on)

pions / srtp / 71 / 1
41%
master: 41%

README BADGES
x