speculation-0.5.1: A framework for safe, programmable, speculative parallelism

Control.Concurrent.Speculation

Contents

Synopsis

Speculative application

spec :: Eq a => a -> (a -> b) -> a -> bSource

spec g f a evaluates f g while forcing a, if g == a then f g is returned. Otherwise f a is evaluated.

Furthermore, if the argument has already been evaluated, we avoid sparking the parallel computation at all.

If a good guess at the value of a is available, this is one way to induce parallelism in an otherwise sequential task.

However, if the guess isn't available more cheaply than the actual answer, then this saves no work and if the guess is wrong, you risk evaluating the function twice.

 spec a f a = f $! a

The best-case timeline looks like:

 [---- f g ----]
    [----- a -----]
 [-- spec g f a --]

The worst-case timeline looks like:

 [---- f g ----]
    [----- a -----]
                  [---- f a ----]
 [------- spec g f a -----------]

Compare these to the timeline of f $! a:

 [---- a -----]
              [---- f a ----]

spec' :: Eq a => a -> (a -> b) -> a -> bSource

Unlike spec, this version does not check to see if the argument has already been evaluated. This can save a small amount of work when you know the argument will always require computation.

specBy :: (a -> a -> Bool) -> a -> (a -> b) -> a -> bSource

spec with a user defined comparison function

specBy' :: (a -> a -> Bool) -> a -> (a -> b) -> a -> bSource

spec' with a user defined comparison function

specOn :: Eq c => (a -> c) -> a -> (a -> b) -> a -> bSource

spec comparing by projection onto another type

specOn' :: Eq c => (a -> c) -> a -> (a -> b) -> a -> bSource

spec' comparing by projection onto another type

Speculative application with transactional rollback

specSTM :: Eq a => STM a -> (a -> STM b) -> a -> STM bSource

specSTM g f a evaluates f g while forcing a, if g == a then f g is returned. Otherwise the side-effects of the current STM transaction are rolled back and f a is evaluated.

If the argument a is already evaluated, we don't bother to perform f g at all.

If a good guess at the value of a is available, this is one way to induce parallelism in an otherwise sequential task.

However, if the guess isn't available more cheaply than the actual answer then this saves no work, and if the guess is wrong, you risk evaluating the function twice.

 specSTM a f a = f $! a

The best-case timeline looks like:

 [------ f g ------]
     [------- a -------]
 [--- specSTM g f a ---]

The worst-case timeline looks like:

 [------ f g ------] 
     [------- a -------]
                       [-- rollback --]
                                      [------ f a ------]     
 [------------------ spec g f a ------------------------]

Compare these to the timeline of f $! a:

 [------- a -------]
                   [------ f a ------]

specSTM' :: Eq a => STM a -> (a -> STM b) -> a -> STM bSource

Unlike specSTM, specSTM' doesn't check if the argument has already been evaluated.

specOnSTM :: Eq c => (a -> c) -> STM a -> (a -> STM b) -> a -> STM bSource

specBySTM . on (==)

specOnSTM' :: Eq c => (a -> c) -> STM a -> (a -> STM b) -> a -> STM bSource

specBySTM' . on (==)

specBySTM :: (a -> a -> Bool) -> STM a -> (a -> STM b) -> a -> STM bSource

specSTM using a user defined comparison function

specBySTM' :: (a -> a -> Bool) -> STM a -> (a -> STM b) -> a -> STM bSource

specSTM' using a user defined comparison function

Determining if a closure is evaluated

unsafeGetTagBits :: a -> IntSource

Inspect the dynamic pointer tagging bits of a closure. This is an impure function that relies on GHC internals and may falsely return 0, but never give the wrong tag number if it returns a non-0 value.

unsafeIsEvaluated :: a -> BoolSource

Returns a guess as to whether or not a value has been evaluated. This is an impure function that relies on GHC internals and will return false negatives, but no false positives. This is unsafe as the value of this function will vary (from False to True) over the course of otherwise pure invocations!