warp.fem.TemporaryStore#

class warp.fem.TemporaryStore[source]#

Shared pool of temporary arrays that will be persisted and reused across invocations of warp.fem functions.

A TemporaryStore instance may either be passed explicitly to warp.fem functions that accept such an argument, for instance integrate.integrate(), or can be set globally as the default store using set_default_temporary_store().

By default, there is no default temporary store, so that temporary allocations are not persisted.

__init__()[source]#

Methods

`__init__`()
`add_temporary_convenience_methods`(temporary)	Attach `release` and `detach` convenience methods to a temporary.
`borrow`(shape, dtype[, pinned, device, ...])	Borrow a temporary array from the pool.
`clear`()	Clear all cached temporary pools.

class Pool(dtype, device, pinned)[source]#

Parameters:: pinned (bool)

class Deleter(pool)[source]#

Parameters:: pool (TemporaryStore.Pool)

detach(temporary)[source]#

Parameters:: temporary (array)

borrow(shape, dtype, requires_grad)[source]#

Parameters:: requires_grad (bool)

redeem(ptr)[source]#

Parameters:: ptr (int)

detach(array)[source]#

Parameters:: array (array)

clear()[source]#: Clear all cached temporary pools.

borrow( shape, dtype, pinned=False, device=None, requires_grad=False, )[source]#

Borrow a temporary array from the pool.

Parameters:

shape – Desired shape of the temporary.
dtype – Desired data type of the temporary.
pinned (bool) – Whether to allocate pinned host memory.
device – Device on which to allocate the temporary.
requires_grad (bool) – Whether to allocate a gradient array.

Return type:

array

static add_temporary_convenience_methods(temporary)[source]#

Attach release and detach convenience methods to a temporary.

Parameters:: temporary (array)
Return type:: array