Skip to content

Commit d57574b

Browse files
committed
Edit docs
1 parent d52e432 commit d57574b

File tree

2 files changed

+14
-6
lines changed

2 files changed

+14
-6
lines changed

README.rst

+7-3
Original file line numberDiff line numberDiff line change
@@ -53,10 +53,14 @@ Status and Lookout
5353

5454
- Performance status:
5555

56-
+ Arithmetic involving arrays of contiguous inner dimension optimizes very well.
57-
+ ``.fold()`` and ``.zip_mut_with()`` are the most efficient ways to
56+
+ Performance of an operation depends on the memory layout of the array
57+
or array view. Especially important if it's a binary operation, which
58+
needs the two arrays to match to be efficient (in most cases).
59+
+ Arithmetic optimizes very well if the arrays are have contiguous inner dimension.
60+
+ The callback based methods like ``.mapv()``, ``.applyv()`` and
61+
``.zip_mut_with()`` are the most efficient ways to
5862
perform single traversal and lock step traversal respectively.
59-
+ ``.iter()`` and ``.iter_mut()`` are efficient for contiguous arrays.
63+
+ ``.iter()`` is efficient for c-contiguous arrays.
6064
+ Can use BLAS in some operations (``dot`` and ``mat_mul``).
6165

6266
Crate Feature Flags

src/lib.rs

+7-3
Original file line numberDiff line numberDiff line change
@@ -43,10 +43,14 @@
4343
//! *will be deprecated* when Rust supports `+=` and similar in Rust 1.8.
4444
//! + We try to introduce more static checking gradually.
4545
//! - Performance status:
46-
//! + Arithmetic involving arrays of contiguous inner dimension optimizes very well.
47-
//! + `.fold()` and `.zip_mut_with()` are the most efficient ways to
46+
//! + Performance of an operation depends on the memory layout of the array
47+
//! or array view. Especially important if it's a binary operation, which
48+
//! needs the two arrays to match to be efficient (in most cases).
49+
//! + Arithmetic optimizes very well if the arrays are have contiguous inner dimension.
50+
//! + The callback based methods like ``.mapv()``, ``.applyv()`` and
51+
//! ``.zip_mut_with()`` are the most efficient ways to
4852
//! perform single traversal and lock step traversal respectively.
49-
//! + `.iter()` and `.iter_mut()` are efficient for contiguous arrays.
53+
//! + ``.iter()`` is efficient for c-contiguous arrays.
5054
//! + Can use BLAS in some operations (`dot` and `mat_mul`).
5155
//!
5256
//! ## Crate Feature Flags

0 commit comments

Comments
 (0)