Skip to content

Commit 768f9c5

Browse files
committed
doc: minor updates to docstring removing mention of dynamic not implemented with cuda
1 parent b666567 commit 768f9c5

File tree

2 files changed

+8
-8
lines changed

2 files changed

+8
-8
lines changed

pylops/waveeqprocessing/_kirchhoff_cuda.py

Lines changed: 6 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -7,10 +7,10 @@
77

88

99
class _KirchhoffCudaHelper:
10-
"""A helper class to perform Kirchhoff demigration/migration using Numba CUDA.
10+
"""Helper class for Kirchhoff operator using Numba CUDA.
1111
12-
This class provides methods to compute the forward and adjoint operations for the
13-
Kirchhoff operator, utilizing GPU acceleration through Numba's CUDA capabilities.
12+
This class provides methods to compute the forward and adjoint operations of the
13+
Kirchhoff operator utilizing GPU acceleration through Numba's CUDA capabilities.
1414
1515
Parameters
1616
----------
@@ -39,7 +39,8 @@ def _grid_setup(self):
3939
This method configures the number of blocks and threads per block for
4040
CUDA kernels, depending on the number of sources and receivers.
4141
"""
42-
# use warp size as number of threads per block
42+
# use half of warp size as number of threads per block
43+
# although this will be ideally lifted up to the user
4344
current_device = cuda.get_current_device()
4445
warp = current_device.WARP_SIZE // 2
4546
self.num_threads_per_blocks = (warp, warp)
@@ -339,7 +340,7 @@ def _call_kinematic(self, opt, *inputs):
339340
return y_d
340341

341342
def _call_dynamic(self, opt, *inputs):
342-
"""Synamic computations using CUDA.
343+
"""Dynamic computations using CUDA.
343344
344345
This method handles data preparation and execution of CUDA kernels
345346
for both forward and adjoint operations of dynamic operator.

pylops/waveeqprocessing/kirchhoff.py

Lines changed: 2 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -111,8 +111,7 @@ class Kirchhoff(LinearOperator):
111111
Deprecated, will be removed in v3.0.0. Simply kept for back-compatibility with previous implementation,
112112
but effectively not affecting the behaviour of the operator.
113113
engine : :obj:`str`, optional
114-
Engine used for computations (``numpy``, ``numba`` or ``cuda``). Note that the ``cuda`` engine
115-
currently supports only ``dynamic=False``.
114+
Engine used for computations (``numpy``, ``numba`` or ``cuda``).
116115
dtype : :obj:`str`, optional
117116
Type of elements in input array.
118117
name : :obj:`str`, optional
@@ -134,7 +133,7 @@ class Kirchhoff(LinearOperator):
134133
If ``mode`` is neither ``analytic``, ``eikonal``, or ``byot``.
135134
136135
NotImplementedError
137-
If ``dynamic=True`` and ``engine="cuda"``
136+
If ``engine="cuda"`` and ``trav`` is provided as a single table
138137
139138
Notes
140139
-----

0 commit comments

Comments
 (0)