Array#

class vulkpy.Array#

Bases: _GPUArray

GPU Array for float (32bit)

Methods Summary

`abs`([inplace])	Element-wise Abs
`acos`([inplace])	Element-wise acos()
`acosh`([inplace])	Element-wise acosh()
`asin`([inplace])	Element-wise asin()
`asinh`([inplace])	Element-wise asinh()
`atan`([inplace])	Element-wise atan()
`atanh`([inplace])	Element-wise atanh()
`broadcast_to`(shape)	Broadcast to new shape
`clamp`(min, max[, inplace])	Element-wise clamp()
`cos`([inplace])	Element-wise cos()
`cosh`([inplace])	Element-wise cosh()
`exp`([inplace])	Element-wise exp()
`exp2`([inplace])	Element-wise exp2()
`flush`()	Flush Buffer to GPU
`gather`(indices[, axis])	Gather values of indices
`invsqrt`([inplace])	Element-wise 1/sqrt()
`log`([inplace])	Element-wise log()
`log2`([inplace])	Element-wise log2()
`max`(other[, inplace])	Element-wise Max
`maximum`([axis, keepdims, rebroadcast])	Get Maximum Value
`mean`([axis, keepdims, rebroadcast])	Calculate Mean Value
`min`(other[, inplace])	Element-wise Min
`minimum`([axis, keepdims, rebroadcast])	Get Minimum Value
`prod`([axis, keepdims, rebroadcast])	Calculate Product of Elements
`reshape`(shape)	Reshape of this array
`sign`([inplace])	Element-wise sign
`sin`([inplace])	Element-wise sin()
`sinh`([inplace])	Element-wise sinh()
`sqrt`([inplace])	Element-wise sqrt()
`sum`([axis, keepdims, rebroadcast])	Calculate Sum of Elements
`tan`([inplace])	Element-wise tan()
`tanh`([inplace])	Element-wise tanh()
`wait`()	Wait Last Job

Methods Documentation

abs(inplace: bool = False) → Array#

Element-wise Abs

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: abs array
Return type:: Array

acos(inplace: bool = False) → Array#

Element-wise acos()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: acos array
Return type:: Array

acosh(inplace: bool = False) → Array#

Element-wise acosh()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: acosh array
Return type:: Array

asin(inplace: bool = False) → Array#

Element-wise asin()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: asin array
Return type:: Array

asinh(inplace: bool = False) → Array#

Element-wise asinh()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: asinh array
Return type:: Array

atan(inplace: bool = False) → Array#

Element-wise atan()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: atan array
Return type:: Array

atanh(inplace: bool = False) → Array#

Element-wise atanh()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: atanh array
Return type:: Array

broadcast_to(shape: Iterable[int]) → Array#

Broadcast to new shape

Parameters:: shape (iterable of ints) – Shape of broadcast target
Returns:: Broadcasted array
Return type:: vulkpy.Array
Raises:: ValueError – If shape is not compatible.

clamp(min: Array | float, max: Array | float, inplace: bool = False) → Array#

Element-wise clamp()

Parameters:

min (Array or float) – Minimum value
max (Array or float) – Maximum value
inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.

Returns:

clamped array

Return type:

Array

cos(inplace: bool = False) → Array#

Element-wise cos()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: cos array
Return type:: Array

cosh(inplace: bool = False) → Array#

Element-wise cosh()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: cosh array
Return type:: Array

exp(inplace: bool = False) → Array#

Element-wise exp()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: exp array
Return type:: Array

exp2(inplace: bool = False) → Array#

Element-wise exp2()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: exp2 array
Return type:: Array

flush()#: Flush Buffer to GPU

gather(indices: U32Array, axis: int | None = None) → Array#

Gather values of indices

Parameters:

indices (vulkpy.U32Array) – Indices
axis (int, optional) – Axis of gather. If None (default), array is flattened beforehand.

Returns:

Gathered array

Return type:

vulkpy.Array

invsqrt(inplace: bool = False) → Array#

Element-wise 1/sqrt()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: 1/sqrt array
Return type:: Array

log(inplace: bool = False) → Array#

Element-wise log()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: log array
Return type:: Array

log2(inplace: bool = False) → Array#

Element-wise log2()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: log2 array
Return type:: Array

max(other: Array | float, inplace: bool = False) → Array#

Element-wise Max

Parameters:

other (Array or float) – Other value
inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.

Returns:

max array

Return type:

Array

Raises:

ValueError – If shape is not same.

maximum(axis: None | int | Iterable[int] = None, keepdims: bool = False, rebroadcast: bool = False) → Array#

Get Maximum Value

Parameters:

axis (int, optional) – Reduction axis
keepdims (bool, optional) – When True, reduced dimensions are keeped with size one. Default is False.
rebroadcast (bool, optional) – When True, keep shape by re-broadcasting after reduce. Default is False.

Returns:

Maximum array

Return type:

vulkpy.Array

mean(axis: None | int | Iterable[int] = None, keepdims: bool = False, rebroadcast: bool = False) → Array#

Calculate Mean Value

Parameters:

axis (int, optional) – Reduction axis
keepdims (bool, optional) – When True, reduced dimensions are keeped with size one. Default is False.
rebroadcast (bool, optional) – When True, keep shape by re-broadcasting after reduce. Default is False.

Returns:

Mean array

Return type:

vulkpy.Array

min(other: Array | float, inplace: bool = False) → Array#

Element-wise Min

Parameters:

other (Array or float) – Other value
inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.

Returns:

min array

Return type:

Array

Raises:

ValueError – If shape is not same.

minimum(axis: None | int | Iterable[int] = None, keepdims: bool = False, rebroadcast: bool = False) → Array#

Get Minimum Value

Parameters:

axis (int, optional) – Reduction axis
keepdims (bool, optional) – When True, reduced dimensions are keeped with size one. Default is False.
rebroadcast (bool, optional) – When True, keep shape by re-broadcasting after reduce. Default is False.

Returns:

Minimum array

Return type:

vulkpy.Array

prod(axis: None | int | Iterable[int] = None, keepdims: bool = False, rebroadcast: bool = False) → Array#

Calculate Product of Elements

Parameters:

axis (int, optional) – Reduction axis
keepdims (bool, optional) – When True, reduced dimensions are keeped with size one. Default is False.
rebroadcast (bool, optional) – When True, keep shape by re-broadcasting after reduce. Default is False.

Returns:

Producted array

Return type:

vulkpy.Array

reshape(shape: Iterable[int])#

Reshape of this array

Parameters:: shape (iterable of int) – New shape
Raises:: ValueError – If shape is incompatible

sign(inplace: bool = False) → Array#

Element-wise sign

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: sign array
Return type:: Array

sin(inplace: bool = False) → Array#

Element-wise sin()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: sin array
Return type:: Array

sinh(inplace: bool = False) → Array#

Element-wise sinh()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: sinh array
Return type:: Array

sqrt(inplace: bool = False) → Array#

Element-wise sqrt()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: sqrt array
Return type:: Array

sum(axis: None | int | Iterable[int] = None, keepdims: bool = False, rebroadcast: bool = False) → Array#

Calculate Sum of Elements

Parameters:

axis (int, optional) – Reduction axis
keepdims (bool, optional) – When True, reduced dimensions are keeped with size one. Default is False.
rebroadcast (bool, optional) – When True, keep shape by re-broadcasting after reduce. Default is False.

Returns:

Summarized array

Return type:

vulkpy.Array

tan(inplace: bool = False) → Array#

Element-wise tan()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: tan array
Return type:: Array

tanh(inplace: bool = False) → Array#

Element-wise tanh()

Parameters:: inplace (bool) – If True, update inplace, otherwise returns new array. Default value is False.
Returns:: tanh array
Return type:: Array

wait()#: Wait Last Job

__init__(gpu: GPU, *, data=None, shape: Iterable[int] | None = None)#

Initialize Array

Parameters:

gpu (GPU) – GPU instance to allocate at.
data (array_like, optional) – Data which copy to GPU buffer.
shape (array_like, optional) – Array shape

Raises:

ValueError – If both data and shape are None.