pub struct PermutePlan {
pub fused_dims: Vec<usize>,
pub src_strides: Vec<isize>,
pub dst_strides: Vec<isize>,
pub root: Option<ComputeNode>,
pub mode: ExecMode,
pub lda_inner: isize,
pub ldb_inner: isize,
pub block: usize,
}Expand description
Complete permutation plan.
Fields§
§fused_dims: Vec<usize>Fused dimensions (after bilateral fusion).
src_strides: Vec<isize>Fused source strides.
dst_strides: Vec<isize>Fused destination strides.
root: Option<ComputeNode>Root of the recursive loop structure (None for Scalar mode).
mode: ExecModeExecution mode.
lda_inner: isizeSource stride along dim_B — the “lda” for macro_kernel. (In the 2D view for the macro-kernel, this is the stride that steps between columns of the source tile.)
ldb_inner: isizeDest stride along dim_A — the “ldb” for macro_kernel.
block: usizeMacro-kernel tile size (= BLOCK, e.g. 16 for f64).
Trait Implementations§
Auto Trait Implementations§
impl Freeze for PermutePlan
impl RefUnwindSafe for PermutePlan
impl Send for PermutePlan
impl Sync for PermutePlan
impl Unpin for PermutePlan
impl UnsafeUnpin for PermutePlan
impl UnwindSafe for PermutePlan
Blanket Implementations§
Source§impl<T> BorrowMut<T> for Twhere
T: ?Sized,
impl<T> BorrowMut<T> for Twhere
T: ?Sized,
Source§fn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Mutably borrows from an owned value. Read more