Struct CpuContext 

pub struct CpuContext { /* private fields */ }

Reusable CPU execution context carrying CPU parallelism policy.

CpuContext stores the requested thread count as a kernel-level parallelism hint and owns the Rayon pool used by multi-threaded CPU work.

Examples

use tenferro_cpu::CpuContext;

let ctx = CpuContext::with_threads(1).unwrap();
let value = ctx.install(|| 1 + 1);
assert_eq!(value, 2);
assert_eq!(ctx.num_threads(), 1);

Implementations

impl CpuContext

pub fn from_env() -> Self

Create a CPU context from RAYON_NUM_THREADS, or fall back to a single-threaded context with a stderr warning when validation fails.

Examples

use tenferro_cpu::CpuContext;

let ctx = CpuContext::from_env();
assert!(ctx.num_threads() >= 1);

pub fn try_from_env() -> Result<Self>

Try to create a CPU context from RAYON_NUM_THREADS.

Examples

use tenferro_cpu::CpuContext;

let ctx = CpuContext::try_from_env()
    .unwrap_or_else(|_| CpuContext::with_threads(1).unwrap());
assert!(ctx.num_threads() >= 1);

Errors

Returns CpuContextError when RAYON_NUM_THREADS is malformed or requests an invalid worker count.

pub fn with_threads(num_threads: usize) -> Result<Self>

Create a CPU context with a fixed parallelism hint.

Examples

use tenferro_cpu::CpuContext;

let ctx = CpuContext::with_threads(2).unwrap();
assert_eq!(ctx.num_threads(), 2);

Errors

Returns CpuContextError::InvalidThreadCount through Error::Validation when num_threads is zero, or Error::BackendSource when Rayon rejects the thread pool.

pub fn with_pinned_cpus( cpus: CpuSet, num_threads: usize, ) -> Result<Self, CpuContextError>

Create a Rayon context whose workers are pinned to assigned logical CPUs.

A real Rayon pool is constructed even when num_threads is one. The worker count cannot exceed the assigned CPU count.

Examples

use tenferro_cpu::{process_cpu_affinity, CpuContext};

if let Some(allowed) = process_cpu_affinity() {
    let one_cpu = tenferro_cpu::CpuSet::new([allowed.as_slice()[0]])?;
    let context = CpuContext::with_pinned_cpus(one_cpu.clone(), 1)?;
    assert_eq!(context.pinned_cpus(), Some(&one_cpu));
}

Errors

Returns CpuContextError::InvalidThreadCount for zero workers, CpuContextError::TooManyWorkers when the request exceeds the CPU set, or an affinity error when workers cannot be pinned.

pub fn num_threads(&self) -> usize

Return this context’s CPU parallelism hint.

Examples

use tenferro_cpu::CpuContext;

let ctx = CpuContext::with_threads(2).unwrap();
assert_eq!(ctx.num_threads(), 2);

pub fn pinned_cpus(&self) -> Option<&CpuSet>

Return the worker CPU domain for a pinned context.

Legacy thread-count-only contexts return None because they do not own worker affinity.

Examples

use tenferro_cpu::CpuContext;

assert_eq!(CpuContext::with_threads(1)?.pinned_cpus(), None);

pub fn install<R: Send>(&self, op: impl FnOnce() -> R + Send) -> R

Run a closure inside this context’s CPU execution scope.

Examples

use tenferro_cpu::CpuContext;

let ctx = CpuContext::with_threads(1).unwrap();
let value = ctx.install(|| 1 + 1);
assert_eq!(value, 2);

Trait Implementations

impl Clone for CpuContext

fn clone(&self) -> CpuContext

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl CpuDomainExecutor for CpuContext

fn capabilities(&self) -> CpuDomainExecutorCapabilities

Return immutable construction-time executor capabilities.

fn submit(&self, jobs: &dyn ScopedCpuJobs) -> Result<(), CpuDomainExecutorError>

Submit all indexed jobs as one synchronous fork/join operation.

fn install( &self, job: &mut dyn ScopedCpuJob, ) -> Result<(), CpuDomainExecutorError>

Enter one synchronous provider-owned inner parallel region.

impl Debug for CpuContext

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

Formats the value using the given formatter.