Parallel_scheduler (p.parallel.v0.18~preview.130.106+341.doc.parallel.scheduler.Parallel

Source

val with_parallel : 
  ?max_workers:Base.int ->
  (Parallel_kernel.t @ local -> 'a) @ once ->
  'a

with_parallel ?max_workers f creates a scheduler that uses up to max_workers worker threads, spawns f into it, and blocks the current thread until f is done executing. Returns the result of f.

Creating a scheduler is an expensive operation, and should ideally only happen at process startup. Most users should use the functions in the Parallel_command library to create their scheduler.

Source

val with_concurrent : 
  ?max_workers:Base.int ->
  (Parallel_kernel.t Concurrent.t @ local portable -> 'a) @ local once unyielding ->
  'a

with_concurrent ?max_workers f creates a scheduler that uses up to max_workers worker threads, passes it to f, and blocks the current thread until f is done executing. Returns the result of f.

Concurrent tasks are executed in order of submission, so later tasks are not guaranteed to run until the current task returns or yields.

Creating a scheduler is an expensive operation, and should ideally only happen at process startup. Most users should use the functions in the Parallel_command library to create their scheduler.

Source

val scheduler : 
  ?max_workers:Base.int ->
  ?on_root:_ Concurrent.Scheduler.t @ local ->
  Base.unit ->
  Parallel_kernel.t Concurrent.Scheduler.t @ portable

scheduler ?max_workers ?on_root () creates a concurrent scheduler that spawns up to max_workers worker tasks.

If on_root is None and Multicore.max_domains () > 1, no worker is spawned on the root domain. If on_root is Some sched, one worker will be spawned onto sched with affinity zero. sched is required to run this worker on the root domain, where it runs concurrently with the existing program.

Creating a scheduler is an expensive operation, and should ideally only happen at process startup. Most users should use the functions in the Parallel_command library to create their scheduler.

Source

val parallel : 
  Parallel_kernel.t Concurrent.Scheduler.t @ portable ->
  (Parallel_kernel.t @ local ->
    (Parallel_kernel.t Concurrent.t @ local portable ->
    'a) @ local once) @ once ->
  'a @@ portable

parallel scheduler f spawns f into scheduler and blocks the current thread until f is done executing. Returns the result of f.

Concurrent tasks are executed in order of submission, so later tasks are not guaranteed to run until the current task returns or yields.

Source

val concurrent : 
  Parallel_kernel.t Concurrent.Scheduler.t @ portable ->
  Await.Await.t @ local ->
  (Parallel_kernel.t Concurrent.t @ local portable -> 'a) @ local once ->
  'a @@ portable

concurrent scheduler await f spawns f into scheduler and uses await to wait for the result.

Concurrent tasks are executed in order of submission, so later tasks are not guaranteed to run until the current task returns or yields.