Struct AutoDiffAttrs

pub struct AutoDiffAttrs {
    pub mode: DiffMode,
    pub width: u32,
    pub ret_activity: DiffActivity,
    pub input_activity: Vec<DiffActivity>,
}

Fields

mode: DiffMode

Conceptually either forward or reverse mode AD, as described in various autodiff papers and e.g. in the JAX Documentation.

width: u32

A user-provided, batching width. If not given, we will default to 1 (no batching). Calling a differentiated, non-batched function through a loop 100 times is equivalent to:

• Calling the function 50 times with a batch size of 2
• Calling the function 25 times with a batch size of 4, etc. A batched function takes more (or longer) arguments, and might be able to benefit from cache locality, better re-usal of primal values, and other optimizations. We will (before LLVM’s vectorizer runs) just generate most LLVM-IR instructions width times, so this massively increases code size. As such, values like 1024 are unlikely to work. We should consider limiting this to u8 or u16, but will leave it at u32 for experiments for now and focus on documenting the implications of a large width.

ret_activity: DiffActivity

input_activity: Vec<DiffActivity>

Implementations

impl AutoDiffAttrs

pub fn has_primal_ret(&self) -> bool

impl AutoDiffAttrs

pub fn has_ret_activity(&self) -> bool

pub fn has_active_only_ret(&self) -> bool

pub const fn error() -> Self

pub fn source() -> Self

pub fn is_active(&self) -> bool

pub fn is_source(&self) -> bool

pub fn apply_autodiff(&self) -> bool

pub fn into_item(self, source: String, target: String) -> AutoDiffItem

Trait Implementations

impl Clone for AutoDiffAttrs

fn clone(&self) -> AutoDiffAttrs

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for AutoDiffAttrs

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<__D: SpanDecoder> Decodable<__D> for AutoDiffAttrs

fn decode(__decoder: &mut __D) -> Self

impl<__E: SpanEncoder> Encodable<__E> for AutoDiffAttrs

fn encode(&self, __encoder: &mut __E)

impl<__CTX> HashStable<__CTX> for AutoDiffAttrs

where __CTX: HashStableContext,

fn hash_stable(&self, __hcx: &mut __CTX, __hasher: &mut StableHasher)

impl PartialEq for AutoDiffAttrs

fn eq(&self, other: &AutoDiffAttrs) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Eq for AutoDiffAttrs

impl StructuralPartialEq for AutoDiffAttrs

Layout

Note: Most layout information is completely unstable and may even differ between compilations. The only exception is types with certain repr(...) attributes. Please see the Rust Reference's “Type Layout” chapter for details on type layout guarantees.

Size: 40 bytes