19 releases (5 stable)
1.1.2 | Dec 20, 2023 |
---|---|
1.1.0 | Nov 11, 2023 |
1.0.1 | Oct 18, 2023 |
1.0.0 | May 5, 2023 |
0.2.3 | Sep 20, 2022 |
#167 in Concurrency
87 downloads per month
Used in blaze-rs
145KB
2.5K
SLoC
Atomic Utils
Simple crate with various miscelaneous atomic-related utils
FillQueue
An atomic queue intended for use cases where taking the full contents of the queue is needed.
The queue is, basically, an atomic singly-linked list, where nodes are first allocated and then the list's tail is atomically updated.
When the queue is "chopped", the list's tail is swaped to null, and it's previous tail is used as the base of the ChopIter
Performance
The performance of pushing elements is expected to be similar to pushing elements to a SegQueue
or Mutex<Vec<_>>
,
but "chopping" elements is expected to be arround 2 times faster than with a Mutex<Vec<_>>
, and 3 times faster than a SegQueue
You can see the benchmark results here
Use FillQueue
when:
- You want a queue that's updateable by shared reference
- You want to retreive all elements of the queue at once
- There is no specifically desired order for the elements to be retreived on, or that order is LIFO (Last In First Out)
Don't use FillQueue
when:
- You don't need a queue updateable by shared reference
- You want to retreive the elements of the queue one by one (see
SegQueue
) - You require the elements in a specific order that isn't LIFO
Features
Name | Description | Enables | Default |
---|---|---|---|
std |
Enables libstd functionality | alloc |
Yes |
alloc |
Enables liballoc functionality | Yes (via std ) |
|
alloc_api |
Enables allocator_api functionality |
alloc & nightly |
No |
futures |
Enables async/await functionality | No | |
const |
Enables constant trait implementations | No | |
nightly |
Enables the use of nightly features | Yes |
Dependencies
~1.7–2.3MB
~48K SLoC