1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
|
name: "Performance bug report"
description: "It works, but it could/should be faster…"
labels: ["performance"]
assignees: []
body:
- type: markdown
attributes:
value: |
The goal of these issues is to fix performance "mistakes," instances where a fix would make the majority of applications several times faster or more, not fine-tuning an application or trading performance in one case for another (unless the former is a very rare or unusual case).
To prove that something is a performance mistake, it needs to have a reproducible metric and a demonstration that shows how fast it could be in bare metal, such as equivalent C or Numba code. If the comparison is truly equivalent (i.e. a general-purpose function is not compared with a highly specialized one), we'll try to optimize the metric within a factor of 2 or so of the baseline.
Alternatively, if you've found a mistake in the code that would always be faster if fixed, we can fix it without tests. Some bugs are obvious.
- type: input
id: version
attributes:
label: "Version of Awkward Array"
description: |
```python
>>> import awkward as ak
>>> ak.__version__
```
(Your issue might have already been fixed; see [latest version](https://pypi.org/project/awkward/).)
placeholder: "XX.YY.ZZ"
validations:
required: true
- type: textarea
id: description
attributes:
label: "Description and code to reproduce"
description: |
If your code example requires data, be sure to include it in a way that is publicly accessible, such as **clicking on the rectangle below to highlight it** and then dragging the file in. (You can [pickle](https://docs.python.org/3/library/pickle.html) Awkward Arrays, but verify that the issue applies to pickled arrays as well as the original arrays.)
placeholder: "Tell us what should be faster."
validations:
required: true
|
