OLD | NEW |
| (Empty) |
1 Skia Perf | |
2 ========= | |
3 | |
4 [Skia Perf](https://perf.skia.org) is a web based interface for exploring | |
5 performance data produced by `nanobench` and the code size bot. The data | |
6 includes: | |
7 | |
8 * The nanobench test times in ms. | |
9 * Total memory consumed during a nanobench run. | |
10 * Code size for various symbol types in bytes. | |
11 | |
12 All of the data can be plotted and also can be [analyzed using k-means | |
13 clustering](https://perf.skia.org/clusters/). | |
14 | |
15 Calculations | |
16 ------------ | |
17 | |
18 Skia Perf has the ability to perform calculations over the test data | |
19 allowing you to build up interesting queries. | |
20 | |
21 For example, this query displays the [total code size of the library over time](
https://perf.skia.org/#1877): | |
22 | |
23 sum(fill(filter("config=memory&sub_result=bytes"))) | |
24 | |
25 This query displays [the ratio of playback time in ms to the number of ops for d
esk\_wowwiki.skp](https://perf.skia.org/#1876): | |
26 | |
27 ratio( | |
28 ave(fill(filter("name=desk_wowwiki.skp&sub_result=min_ms"))), | |
29 ave(fill(filter("name=desk_wowwiki.skp&sub_result=ops"))) | |
30 ) | |
31 | |
32 You can also use the data to answer questions like [how many tests were run per
commit](https://perf.skia.org/#1878). | |
33 | |
34 count(filter("")) | |
35 | |
36 See Skia Perf for the [full list of functions available](https://perf.skia.org/h
elp). | |
OLD | NEW |