Struct PartitionConfig 

pub struct PartitionConfig {
    pub placement: Placement,
    pub bind_memory: bool,
    pub threads_per_partition: Option<usize>,
}

Placement plus the knobs the partitioned engine reads alongside it.

Fields

placement: Placement

How partitions map onto the machine.

bind_memory: bool

Whether each pool’s workers bind their allocations to the partition’s NUMA node (Python’s pin_memory=).

threads_per_partition: Option<usize>

Workers in every partition’s pool (at least one) on the CPU set the placement resolved, or None for the placement’s own count.

A measurement knob for holding the total thread count fixed across partition counts; it overrides a device partition’s small host pool too.

Implementations

impl PartitionConfig

pub fn num_partitions(&self) -> Result<usize, TopologyError>

The partition count the placement resolves to on this machine, without building any pool.

Errors

TopologyError::NotPowerOfTwo if an explicit or unpinned partition count is not a power of two, TopologyError::EmptySet if an explicit set is empty, and TopologyError::CpuNotAllowed if an explicit set names a CPU outside the affinity mask.

Trait Implementations

impl Clone for PartitionConfig

fn clone(&self) -> PartitionConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for PartitionConfig

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

Formats the value using the given formatter.

impl Default for PartitionConfig

fn default() -> Self

One partition per NUMA node, memory bound to it.

impl PartialEq for PartitionConfig

fn eq(&self, other: &PartitionConfig) -> 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 PartitionConfig

impl StructuralPartialEq for PartitionConfig