Safe Haskell	None

Data.Yarr.Eval

Contents

Aliases for common parameters
Load classes
Common load types
Utility

Description

Loading and computing arrays

Synopsis

Aliases for common parameters

type Threads = IO Int Source

There are 2 common ways to parameterize parallelism: a) to say "split this work between n threads" or b) to say "split this work between maximum reasonable number of threads", that is capabilities. Since getNumCapabilities function is monadic, we need always pass IO Int as thread number parameter in order not to multiply number of functions in this module (there are already too many).

caps :: Threads Source

Alias to getNumCapabilities.

threads :: Int -> Threads Source

Alias to return.

type Fill sh aSource

Arguments

= (sh -> IO a)	Get
-> (sh -> a -> IO ())	Write
-> sh	Start
-> sh	End
-> IO ()

Alias to frequently used get-write-from-to arguments combo.

Passed as 1st parameter of all Loading functions from Data.Yarr.Eval module.

Load classes

class (USource r l sh a, UTarget tr tl sh a, Shape (LoadIndex l tl sh)) => Load r l tr tl sh a whereSource

This class abstracts pair of array types, which could be loaded one to another.

Parameters:

r - source representation. Instance of USource class. Typically one of fused representations: D, (SE D) or CV.
l - source load type
tr - target representation. Instance of UTarget class.
tl - target load type
sh - shape of arrays
a - array element type

Counterpart for arrays of vectors: VecLoad.

TODO: this class seems to be overengineered, normally it should have only 3 parameters: Load l tl sh. But Convoluted (CV) representation is tightly connected with it's load type.

Associated Types

type LoadIndex l tl sh Source

Used in fill parameter function. There are two options for this type to be: sh itself or Int. Don't confuse this type with load type indexes: r and l. There are 2 different meanings of word "index": data type index (haskell term) and array index (linear, shape).

Methods

loadP Source

Arguments

:: Fill (LoadIndex l tl sh) a	Filling (real worker) function
-> Threads	Number of threads to parallelize loading on
-> UArray r l sh a	Source array
-> UArray tr tl sh a	Target array
-> IO ()

O(n) Entirely loads source to target in parallel.

First parameter is used to parameterize loop unrolling to maximize performance. Default choice is fill -- vanilla not unrolled looping.

Examples:

 tarr <- new (extent arr)
 loadP fill caps arr tarr
 loadP (dim2BlockFill n2 n2 touch) (threads 2) arr tarr

loadS Source

Arguments

:: Fill (LoadIndex l tl sh) a	Filling (real worker) function
-> UArray r l sh a	Source array
-> UArray tr tl sh a	Target array
-> IO ()

O(n) Sequential analog of loadP function. Loads source to target entirely.

Example:

loadS (unrolledFill n4 noTouch) caps arr tarr

Instances

(USource r SH sh a, UTarget tr SH sh a) => Load r SH tr SH sh a
(USource r L sh a, UTarget tr SH sh a) => Load r L tr SH sh a
(USource r SH sh a, UTarget tr L sh a) => Load r SH tr L sh a
(USource r L sh a, UTarget tr L sh a) => Load r L tr L sh a
(BlockShape sh, UTarget tr tl sh a) => Load CV CVL tr tl sh a

class (Load r l tr tl sh a, LoadIndex l tl sh ~ sh) => RangeLoad r l tr tl sh a whereSource

Class abstracts pair of arrays which could be loaded in just specified range of indices.

"Range" is a multidimensional segment: segment for Dim1 arrays, square for Dim2 arrays and cube for Dim3. Thus, it is specified by pair of indices: "top-left" (minimum is zero) and "bottom-right" (maximum is (entire arr tarr)) corners.

Methods

rangeLoadP Source

Arguments

:: Fill sh a	Filling (real worker) function
-> Threads	Number of threads to parallelize loading on
-> UArray r l sh a	Source array
-> UArray tr tl sh a	Target array
-> sh	Top-left
-> sh	and bottom-right corners of range to load
-> IO ()

O(n) Loads elements from source to target in specified range in parallel.

Example:

 let ext = extent convolved
 res <- new ext
 rangeLoadP fill caps convolved res (5, 5) (ext `minus` (5, 5))

rangeLoadS Source

Arguments

:: Fill sh a	Filling (real worker) function
-> UArray r l sh a	Source array
-> UArray tr tl sh a	Target array
-> sh	Top-left
-> sh	and bottom-right corners of range to load
-> IO ()

O(n) Sequentially loads elements from source to target in specified range.

Instances

(USource r SH sh a, UTarget tr SH sh a) => RangeLoad r SH tr SH sh a
(USource r L sh a, UTarget tr SH sh a) => RangeLoad r L tr SH sh a
(USource r SH sh a, UTarget tr L sh a) => RangeLoad r SH tr L sh a
(BlockShape sh, UTarget tr tl sh a) => RangeLoad CV CVL tr tl sh a

class (UVecSource r slr l sh v e, UVecTarget tr tslr tl sh v2 e, Load slr l tslr tl sh e, Shape (LoadIndex l tl sh), Dim v ~ Dim v2) => VecLoad r slr l tr tslr tl sh v v2 e whereSource

Class abstracts separated in time and space loading slices of one array type to another. Result of running functions with -Slices- infix is always identical to result of running corresponding function from Load class. VecLoad and RangeVecLoad are just about performance. If target representation is separate (ex. (SE F)), using loadSlicesP may be faster than loadP because of per-thread memory locality.

Parameters:

r - source representation
slr - source slice representation
l - source load type
tr - target representation
tslr - target slice representation
tl - target load type
sh - shape of arrays
v - source vector type
v2 - target vector type
e - vector element type, common for source and target arrays

Methods

loadSlicesP Source

Arguments

:: Fill (LoadIndex l tl sh) e	Fill function to work on slices
-> Threads	Number of threads to parallelize loading on
-> UArray r l sh (v e)	Source array of vectors
-> UArray tr tl sh (v2 e)	Target array of vectors
-> IO ()

O(n) Entirely, slice-wise loads vectors from source to target in parallel.

Example:

 -- blurred and delayedBlurred are arrays of color components.
 loadSlicesP fill caps delayedBlurred blurred

loadSlicesS Source

Arguments

:: Fill (LoadIndex l tl sh) e	Fill function to work on slices
-> UArray r l sh (v e)	Source array of vectors
-> UArray tr tl sh (v2 e)	Target array of vectors
-> IO ()

O(n) Sequentially loads vectors from source to target, slice by slice.

Instances

(UVecSource r slr SH sh v e, UVecTarget tr tslr SH sh v2 e, Load slr SH tslr SH sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr SH tr tslr SH sh v v2 e
(UVecSource r slr L sh v e, UVecTarget tr tslr SH sh v2 e, Load slr L tslr SH sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr L tr tslr SH sh v v2 e
(UVecSource r slr SH sh v e, UVecTarget tr tslr L sh v2 e, Load slr SH tslr L sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr SH tr tslr L sh v v2 e
(UVecSource r slr L sh v e, UVecTarget tr tslr L sh v2 e, Load slr L tslr L sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr L tr tslr L sh v v2 e
(BlockShape sh, Vector v e, UVecTarget tr tslr tl sh v2 e, ~ * (Dim v) (Dim v2), InlinableArity (Dim v)) => VecLoad (SE CV) CV CVL tr tslr tl sh v v2 e

class (VecLoad r slr l tr tslr tl sh v v2 e, LoadIndex l tl sh ~ sh) => RangeVecLoad r slr l tr tslr tl sh v v2 e whereSource

This class extends VecLoad just like RangeLoad extends Load. It abstracts slice-wise loading from one array type to another in specified range.

Methods

rangeLoadSlicesP Source

Arguments

:: Fill sh e	Fill function to work on slices
-> Threads	Number of threads to parallelize loading on
-> UArray r l sh (v e)	Source array of vectors
-> UArray tr tl sh (v2 e)	Target array of vectors
-> sh	Top-left
-> sh	and bottom-right corners of range to load
-> IO ()

O(n) Loads vectors from source to target in specified range, slice-wise, in parallel.

rangeLoadSlicesS Source

Arguments

:: Fill sh e	Fill function to work on slices
-> UArray r l sh (v e)	Source array of vectors
-> UArray tr tl sh (v2 e)	Target array of vectors
-> sh	Top-left
-> sh	and bottom-right corners of range to load
-> IO ()

O(n) Sequentially loads vector elements from source to target in specified range, slice by slice.

Instances

(UVecSource r slr SH sh v e, UVecTarget tr tslr SH sh v2 e, Load slr SH tslr SH sh e, ~ * (Dim v) (Dim v2)) => RangeVecLoad r slr SH tr tslr SH sh v v2 e
(UVecSource r slr L sh v e, UVecTarget tr tslr SH sh v2 e, Load slr L tslr SH sh e, ~ * (Dim v) (Dim v2)) => RangeVecLoad r slr L tr tslr SH sh v v2 e
(UVecSource r slr SH sh v e, UVecTarget tr tslr L sh v2 e, Load slr SH tslr L sh e, ~ * (Dim v) (Dim v2)) => RangeVecLoad r slr SH tr tslr L sh v v2 e
(BlockShape sh, Vector v e, UVecTarget tr tslr tl sh v2 e, ~ * (Dim v) (Dim v2), InlinableArity (Dim v)) => RangeVecLoad (SE CV) CV CVL tr tslr tl sh v v2 e

compute Source

Arguments

:: (USource r l sh a, Manifest tr mtr tl sh b)
=> (UArray r l sh a -> UArray mtr tl sh b -> IO ())	Loading function
-> UArray r l sh a	Source array
-> IO (UArray tr tl sh b)	Entirely loaded from the source, `freeze`d manifest target array

O(n) This function simplifies the most common way of loading arrays.

Instead of

 mTarget <- new (extent source)
 loadP fill caps source mTarget
 target <- freeze mTarget

You can write just

target <- compute (loadP fill caps) source

dComputeP :: (USource r l sh a, Manifest tr mtr tl sh a, Load r l mtr tl sh a) => UArray r l sh a -> IO (UArray tr tl sh a)Source

dComputeS :: (USource r l sh a, Manifest tr mtr tl sh a, Load r l mtr tl sh a) => UArray r l sh a -> IO (UArray tr tl sh a)Source

Common load types

data L Source

Linear load type index. UArrays with L load type index define linearIndex and linearWrite and leave index and write functions defined by default.

Instances

DefaultFusion D D L
DefaultFusion FS D L
DefaultFusion F D L
DefaultFusion MB D L
DefaultFusion B D L
Fusion r D L
(Shape sh, Storable e) => UTarget FS L sh e
(Shape sh, Storable a) => UTarget F L sh a
(Shape sh, NFData a) => UTarget MB L sh a
Shape sh => USource D L sh a
(Shape sh, Storable e) => USource FS L sh e
(Shape sh, Storable a) => USource F L sh a
(Shape sh, NFData a) => USource MB L sh a
(Shape sh, NFData a) => USource B L sh a
Shape sh => Regular D L sh a
Shape sh => Regular FS L sh e
Shape sh => Regular F L sh a
(Shape sh, NFData a) => Regular MB L sh a
(Shape sh, NFData a) => Regular B L sh a
(Shape sh, Storable a) => Manifest F F L sh a
(Shape sh, NFData a) => Manifest B MB L sh a
(Shape sh, Vector v e, Storable e) => UVecTarget F FS L sh v e
(Shape sh, Vector v e) => UVecSource D D L sh v e
(Shape sh, Vector v e, Storable e) => UVecSource F FS L sh v e
(Shape sh, Vector v e) => VecRegular D D L sh v e
(Shape sh, Vector v e, Storable e) => VecRegular F FS L sh v e
(USource r L sh a, UTarget tr SH sh a) => RangeLoad r L tr SH sh a
(USource r SH sh a, UTarget tr L sh a) => RangeLoad r SH tr L sh a
(USource r L sh a, UTarget tr SH sh a) => Load r L tr SH sh a
(USource r SH sh a, UTarget tr L sh a) => Load r SH tr L sh a
(USource r L sh a, UTarget tr L sh a) => Load r L tr L sh a
(UVecSource r slr L sh v e, UVecTarget tr tslr SH sh v2 e, Load slr L tslr SH sh e, ~ * (Dim v) (Dim v2)) => RangeVecLoad r slr L tr tslr SH sh v v2 e
(UVecSource r slr SH sh v e, UVecTarget tr tslr L sh v2 e, Load slr SH tslr L sh e, ~ * (Dim v) (Dim v2)) => RangeVecLoad r slr SH tr tslr L sh v v2 e
(UVecSource r slr L sh v e, UVecTarget tr tslr SH sh v2 e, Load slr L tslr SH sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr L tr tslr SH sh v v2 e
(UVecSource r slr SH sh v e, UVecTarget tr tslr L sh v2 e, Load slr SH tslr L sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr SH tr tslr L sh v v2 e
(UVecSource r slr L sh v e, UVecTarget tr tslr L sh v2 e, Load slr L tslr L sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr L tr tslr L sh v v2 e
(Shape sh, Vector v e, Storable e) => UVecSource (SE F) F L sh v e
(Shape sh, Vector v e, NFData e) => UVecSource (SE MB) MB L sh v e
(Shape sh, Vector v e, NFData e) => UVecSource (SE B) B L sh v e
Shape sh => NFData (UArray D L sh a)
Shape sh => NFData (UArray FS L sh e)
Shape sh => NFData (UArray F L sh a)
(Shape sh, NFData a) => NFData (UArray MB L sh a)
(Shape sh, NFData a) => NFData (UArray B L sh a)

data SH Source

General shape load type index. UArrays with SH load type index specialize index and write and leave linearIndex and linearWrite functions defined by default.

Type-level distinction between Linear and SHaped arrays is aimed to avoid integral division operations while looping through composite (Dim2, Dim3) indices.

Integral division is very expensive operation even on modern CPUs.

Instances

DefaultFusion D D SH
Fusion r D SH
Shape sh => UTarget DT SH sh a
Shape sh => USource D SH sh a
Shape sh => Regular DT SH sh a
Shape sh => Regular D SH sh a
(Shape sh, Vector v e) => UVecSource D D SH sh v e
(Shape sh, Vector v e) => VecRegular D D SH sh v e
(USource r SH sh a, UTarget tr SH sh a) => RangeLoad r SH tr SH sh a
(USource r L sh a, UTarget tr SH sh a) => RangeLoad r L tr SH sh a
(USource r SH sh a, UTarget tr L sh a) => RangeLoad r SH tr L sh a
(USource r SH sh a, UTarget tr SH sh a) => Load r SH tr SH sh a
(USource r L sh a, UTarget tr SH sh a) => Load r L tr SH sh a
(USource r SH sh a, UTarget tr L sh a) => Load r SH tr L sh a
(UVecSource r slr SH sh v e, UVecTarget tr tslr SH sh v2 e, Load slr SH tslr SH sh e, ~ * (Dim v) (Dim v2)) => RangeVecLoad r slr SH tr tslr SH sh v v2 e
(UVecSource r slr L sh v e, UVecTarget tr tslr SH sh v2 e, Load slr L tslr SH sh e, ~ * (Dim v) (Dim v2)) => RangeVecLoad r slr L tr tslr SH sh v v2 e
(UVecSource r slr SH sh v e, UVecTarget tr tslr L sh v2 e, Load slr SH tslr L sh e, ~ * (Dim v) (Dim v2)) => RangeVecLoad r slr SH tr tslr L sh v v2 e
(UVecSource r slr SH sh v e, UVecTarget tr tslr SH sh v2 e, Load slr SH tslr SH sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr SH tr tslr SH sh v v2 e
(UVecSource r slr L sh v e, UVecTarget tr tslr SH sh v2 e, Load slr L tslr SH sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr L tr tslr SH sh v v2 e
(UVecSource r slr SH sh v e, UVecTarget tr tslr L sh v2 e, Load slr SH tslr L sh e, ~ * (Dim v) (Dim v2)) => VecLoad r slr SH tr tslr L sh v v2 e
Shape sh => NFData (UArray DT SH sh a)
Shape sh => NFData (UArray D SH sh a)

Utility

entire :: (Regular r l sh a, Regular r2 l2 sh b) => UArray r l sh a -> UArray r2 l2 sh b -> shSource

Determines maximum common range of 2 arrays - intersection of their extents.