Skip to main content

measure_thread_count_sweep

silicera_lab::threads

Function measure_thread_count_sweep 

Source 
pub fn measure_thread_count_sweep(
    info: &HardwareInfo,
    mcfg: &MeasurementConfig,
    total_iters: u64,
    thread_counts: Option<&[usize]>,
) -> Result<ThreadCountReport>
Expand description

Measure concurrency kernels across thread counts (Silicera-owned; no OS affinity).

Total work is held approximately constant: iters_per_thread * threads ≈ iters_per_thread_at_1 so wall-time comparisons ask whether parallelism helps, not whether doing N× more work is slower.