[BE] Adding aliases for CUDA and XPU API documentation (#162984)

This PR reorganizes CUDA and XPU API documentation with additional aliases pages. Multiple entries of APIs under torch.cuda are thus removed.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/162984
Approved by: https://github.com/janeyx99

docs/source/conf.py

@@ -509,10 +509,6 @@ coverage_ignore_functions = [
     "custom_fwd",
     # torch.cuda.amp.common
     "amp_definitely_not_available",
-    # torch.cuda.graphs
-    "graph_pool_handle",
-    "is_current_stream_capturing",
-    "make_graphed_callables",
     # torch.mtia.memory
     "reset_peak_memory_stats",
     # torch.cuda.nccl
@@ -524,25 +520,11 @@ coverage_ignore_functions = [
     "reduce_scatter",
     "unique_id",
     "version",
-    # torch.cuda.nvtx
-    "range",
-    "range_end",
-    "range_start",
     # torch.cuda.profiler
     "init",
     "profile",
     "start",
     "stop",
-    # torch.cuda.random
-    "get_rng_state",
-    "get_rng_state_all",
-    "initial_seed",
-    "manual_seed",
-    "manual_seed_all",
-    "seed",
-    "seed_all",
-    "set_rng_state",
-    "set_rng_state_all",
     # torch.distributed.algorithms.ddp_comm_hooks.ddp_zero_hook
     "hook_with_zero_step",
     "hook_with_zero_step_interleaved",
@@ -2172,8 +2154,6 @@ coverage_ignore_classes = [
     "EventHandler",
     "SynchronizationError",
     "UnsynchronizedAccessError",
-    # torch.cuda.memory
-    "MemPool",
     # torch.distributed.elastic.multiprocessing.errors
     "ChildFailedError",
     "ProcessFailure",
@@ -2479,10 +2459,6 @@ coverage_ignore_classes = [
     # torch.amp.grad_scaler
     "GradScaler",
     "OptState",
-    # torch.cuda.graphs
-    "CUDAGraph",
-    # torch.cuda.streams
-    "Event",
     # torch.distributed.algorithms.ddp_comm_hooks.post_localSGD_hook
     "PostLocalSGDState",
     # torch.distributed.algorithms.ddp_comm_hooks.powerSGD_hook

docs/source/cuda.aliases.md

@@ -0,0 +1,47 @@
+# Aliases in torch.cuda
+
+The following are aliases to their counterparts in ``torch.cuda`` in the nested namespaces in which they are defined. For any of these APIs, feel free to use the top-level version in ``torch.cuda`` like ``torch.cuda.seed`` or the nested version ``torch.cuda.random.seed``.
+
+```{eval-rst}
+.. automodule:: torch.cuda.random
+.. currentmodule:: torch.cuda.random
+.. autosummary::
+    :toctree: generated
+    :nosignatures:
+
+    get_rng_state
+    get_rng_state_all
+    set_rng_state
+    set_rng_state_all
+    manual_seed
+    manual_seed_all
+    seed
+    seed_all
+    initial_seed
+```
+
+```{eval-rst}
+.. automodule:: torch.cuda.graphs
+.. currentmodule:: torch.cuda.graphs
+.. autosummary::
+    :toctree: generated
+    :nosignatures:
+
+    is_current_stream_capturing
+    graph_pool_handle
+    CUDAGraph
+    graph
+    make_graphed_callables
+```
+
+```{eval-rst}
+.. automodule:: torch.cuda.streams
+.. currentmodule:: torch.cuda.streams
+.. autosummary::
+    :toctree: generated
+    :nosignatures:
+
+    Stream
+    ExternalStream
+    Event
+```

docs/source/cuda.md

@@ -274,10 +274,6 @@ See the docs for {class}`~torch.cuda.gds.GdsFile` for an example of how to use t
 .. py:module:: torch.cuda.gds
 ```
 
-```{eval-rst}
-.. py:module:: torch.cuda.graphs
-```
-
 ```{eval-rst}
 .. py:module:: torch.cuda.jiterator
 ```
@@ -294,14 +290,13 @@ See the docs for {class}`~torch.cuda.gds.GdsFile` for an example of how to use t
 .. py:module:: torch.cuda.profiler
 ```
 
-```{eval-rst}
-.. py:module:: torch.cuda.random
-```
-
 ```{eval-rst}
 .. py:module:: torch.cuda.sparse
 ```
 
 ```{eval-rst}
-.. py:module:: torch.cuda.streams
+.. toctree::
+    :hidden:
+
+    cuda.aliases.md
 ```

docs/source/xpu.aliases.md

@@ -0,0 +1,32 @@
+# Aliases in torch.xpu
+
+The following are aliases to their counterparts in ``torch.xpu`` in the nested namespaces in which they are defined. For any of these APIs, feel free to use the top-level version in ``torch.xpu`` like ``torch.xpu.seed`` or the nested version ``torch.xpu.random.seed``.
+
+```{eval-rst}
+.. automodule:: torch.xpu.random
+.. currentmodule:: torch.xpu.random
+.. autosummary::
+    :toctree: generated
+    :nosignatures:
+
+    get_rng_state
+    get_rng_state_all
+    initial_seed
+    manual_seed
+    manual_seed_all
+    seed
+    seed_all
+    set_rng_state
+    set_rng_state_all
+```
+
+```{eval-rst}
+.. automodule:: torch.xpu.streams
+.. currentmodule:: torch.xpu.streams
+.. autosummary::
+    :toctree: generated
+    :nosignatures:
+
+    Event
+    Stream
+```

docs/source/xpu.md

@@ -86,9 +86,9 @@
      reset_peak_memory_stats
 ```
 
-<!-- This module needs to be documented. Adding here in the meantime
-for tracking purposes -->
 ```{eval-rst}
-.. py:module:: torch.xpu.random
-.. py:module:: torch.xpu.streams
+.. toctree::
+    :hidden:
+
+    xpu.aliases.md
 ```