Benchmark Exact Fast Paths — benchmark_fast

Runs reproducible synthetic benchmarks comparing baseline and exact fast paths for task PLS permutation and bootstrap workloads.

Usage

benchmark_fast_paths(
  operations = c("bootstrap", "permutation"),
  method = 1L,
  num_subj_lst = c(16L, 16L),
  num_cond = 2L,
  n_features = 5000L,
  num_boot = 48L,
  num_perm = 48L,
  reps = 3L,
  workers = 2L,
  seed = 1L
)

Arguments

operations: Character vector containing "bootstrap" and/or "permutation".
method: Task method to benchmark: 1L (rotated task PLS) or 2L (non-rotated task PLS).
num_subj_lst: Balanced subject counts by group.
num_cond: Number of conditions.
n_features: Number of features/voxels.
num_boot: Number of bootstrap samples for bootstrap benchmarks.
num_perm: Number of permutation samples for permutation benchmarks.
reps: Number of repeated timing runs per scenario.
workers: Number of workers for the parallel fast-path scenario.
seed: Random seed for synthetic data generation.

Value

A data frame with timing results by operation and scenario.