pub struct ServingCost {Show 25 fields
pub card: &'static str,
pub batch: u32,
pub host_sequences: u32,
pub rows_per_sequence: u32,
pub emitted_per_sequence: f64,
pub weight_bytes: f64,
pub sequence_bytes: f64,
pub rewrite_bytes: f64,
pub array_macs: f64,
pub vector_macs: f64,
pub scale_cycles: f64,
pub memory_s: f64,
pub host_s: f64,
pub array_s: f64,
pub vector_s: f64,
pub scale_s: f64,
pub pass_s: f64,
pub binds: ServingBind,
pub held_bytes: f64,
pub capacity_bytes: f64,
pub host_held_bytes: f64,
pub host_capacity_bytes: f64,
pub on_chip_bytes: f64,
pub on_chip_capacity: f64,
pub provenance: Provenance,
}Expand description
One pass of a serving on a card.
Fields§
§card: &'static str§batch: u32§host_sequences: u32Of the batch, the sequences whose state and caches live in host memory.
rows_per_sequence: u32§emitted_per_sequence: f64§weight_bytes: f64Weight bytes a pass streams, shared by the batch, embedding rows included.
sequence_bytes: f64Bytes a pass moves for each sequence: its state, its caches, its log, its conv ring.
rewrite_bytes: f64Of sequence_bytes, the recurrent states’ bytes read and written back in place.
array_macs: f64§vector_macs: f64§scale_cycles: f64§memory_s: f64§host_s: f64§array_s: f64§vector_s: f64§scale_s: f64§pass_s: f64The largest of the five: the units and the memories overlap, so no pass is faster.
binds: ServingBind§held_bytes: f64§capacity_bytes: f64§host_held_bytes: f64§host_capacity_bytes: f64§on_chip_bytes: f64§on_chip_capacity: f64§provenance: ProvenanceThe weakest fact the price rests on.
Implementations§
Source§impl ServingCost
impl ServingCost
pub fn tokens_per_s(&self) -> f64
Trait Implementations§
Source§impl Clone for ServingCost
impl Clone for ServingCost
Source§fn clone(&self) -> ServingCost
fn clone(&self) -> ServingCost
Returns a duplicate of the value. Read more
1.0.0 · Source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
Performs copy-assignment from
source. Read moreSource§impl Debug for ServingCost
impl Debug for ServingCost
Source§impl PartialEq for ServingCost
impl PartialEq for ServingCost
Source§impl Serialize for ServingCost
impl Serialize for ServingCost
impl StructuralPartialEq for ServingCost
Auto Trait Implementations§
impl Freeze for ServingCost
impl RefUnwindSafe for ServingCost
impl Send for ServingCost
impl Sync for ServingCost
impl Unpin for ServingCost
impl UnsafeUnpin for ServingCost
impl UnwindSafe for ServingCost
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
Source§impl<T> CloneToUninit for Twhere
T: Clone,
impl<T> CloneToUninit for Twhere
T: Clone,
§impl<T> Downcast for Twhere
T: Any,
impl<T> Downcast for Twhere
T: Any,
§fn into_any(self: Box<T>) -> Box<dyn Any>
fn into_any(self: Box<T>) -> Box<dyn Any>
Convert
Box<dyn Trait> (where Trait: Downcast) to Box<dyn Any>. Box<dyn Any> can
then be further downcast into Box<ConcreteType> where ConcreteType implements Trait.§fn into_any_rc(self: Rc<T>) -> Rc<dyn Any>
fn into_any_rc(self: Rc<T>) -> Rc<dyn Any>
Convert
Rc<Trait> (where Trait: Downcast) to Rc<Any>. Rc<Any> can then be
further downcast into Rc<ConcreteType> where ConcreteType implements Trait.§fn as_any(&self) -> &(dyn Any + 'static)
fn as_any(&self) -> &(dyn Any + 'static)
Convert
&Trait (where Trait: Downcast) to &Any. This is needed since Rust cannot
generate &Any’s vtable from &Trait’s.§fn as_any_mut(&mut self) -> &mut (dyn Any + 'static)
fn as_any_mut(&mut self) -> &mut (dyn Any + 'static)
Convert
&mut Trait (where Trait: Downcast) to &Any. This is needed since Rust cannot
generate &mut Any’s vtable from &mut Trait’s.