Expand description
Safely typed views over slices with static alignment, architecture dispatch, reference typestates, and execution mode.
Re-exports§
pub use tile::TileMatrixMultiply;pub use tile::TileView;pub use mask_reg::Mask;pub use vector_reg::Vector;
Modules§
- gather
- Vectorized indirect load (gather) operations.
- mask_
reg - Module containing the SIMD mask register wrappers. Monomorphized SIMD mask register wrapper.
- masked
- Lane-masked math and compaction/expansion operations on SIMD views.
- ops
- Standard elementwise and accumulation operations on SIMD views.
Elementwise and transform operations over
SimdView. - ops_mut
- Standard exclusive mutable elementwise operations on SIMD views.
Mutable elementwise SIMD operations on views backed by
&'a mut [T]. - reduce
- Unrolled generic horizontal reductions on SIMD views.
Horizontal and pairwise SIMD reductions over
SimdView. - scan
- Inclusive/exclusive prefix scans and running min/max.
- select
- Lane-wise conditional select and masked-negate.
Conditional SIMD blend/select operations on
SimdView. - tile
- 2D matrix tile views and operations. 2D matrix tile views for high-performance tiled and register-blocked matrix multiply kernels.
- unary
- Unary mapping operations on SIMD views.
- vector_
ops - Module containing operator overload implementations for SIMD vectors.
Operator overloads for SIMD vectors.
Operator overload implementations for the
Vectorregister wrapper. - vector_
reg - Module containing the generic SIMD vector register wrappers. Monomorphized SIMD vector register wrapper.
Structs§
- Simd
View - A zero-copy, typed slice view parameterized by architecture, alignment, execution mode, and reference typestates.
Enums§
- Simd
Error - Error types for SIMD view operations.