torchmonarch-nightly 2025.6.27__cp313-cp313-manylinux2014_x86_64.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (165) hide show
  1. monarch/__init__.py +189 -0
  2. monarch/_monarch/__init__.py +5 -0
  3. monarch/_monarch/hyperactor/__init__.py +58 -0
  4. monarch/_monarch/selection/__init__.py +13 -0
  5. monarch/_monarch/worker/__init__.py +0 -0
  6. monarch/_monarch/worker/debugger.py +117 -0
  7. monarch/_monarch/worker/logging.py +107 -0
  8. monarch/_rust_bindings.so +0 -0
  9. monarch/_testing.py +230 -0
  10. monarch/actor_mesh.py +761 -0
  11. monarch/allocator.py +220 -0
  12. monarch/bootstrap_main.py +59 -0
  13. monarch/builtins/__init__.py +14 -0
  14. monarch/builtins/log.py +22 -0
  15. monarch/builtins/random.py +68 -0
  16. monarch/cached_remote_function.py +257 -0
  17. monarch/code_sync.py +10 -0
  18. monarch/common/_C.pyi +11 -0
  19. monarch/common/_C.so +0 -0
  20. monarch/common/__init__.py +0 -0
  21. monarch/common/_coalescing.py +308 -0
  22. monarch/common/_device_utils.py +18 -0
  23. monarch/common/_tensor_to_table.py +172 -0
  24. monarch/common/base_tensor.py +28 -0
  25. monarch/common/borrows.py +143 -0
  26. monarch/common/client.py +690 -0
  27. monarch/common/constants.py +10 -0
  28. monarch/common/context_manager.py +40 -0
  29. monarch/common/controller_api.py +104 -0
  30. monarch/common/device_mesh.py +417 -0
  31. monarch/common/fake.py +55 -0
  32. monarch/common/function.py +160 -0
  33. monarch/common/function_caching.py +164 -0
  34. monarch/common/future.py +168 -0
  35. monarch/common/invocation.py +125 -0
  36. monarch/common/mast.py +221 -0
  37. monarch/common/messages.py +573 -0
  38. monarch/common/mock_cuda.py +41 -0
  39. monarch/common/opaque_ref.py +98 -0
  40. monarch/common/pickle_flatten.py +48 -0
  41. monarch/common/pipe.py +152 -0
  42. monarch/common/process_group.py +55 -0
  43. monarch/common/recording.py +127 -0
  44. monarch/common/reference.py +33 -0
  45. monarch/common/remote.py +297 -0
  46. monarch/common/selection.py +9 -0
  47. monarch/common/shape.py +229 -0
  48. monarch/common/stream.py +114 -0
  49. monarch/common/tensor.py +814 -0
  50. monarch/common/tensor_factory.py +31 -0
  51. monarch/common/tree.py +73 -0
  52. monarch/controller/__init__.py +7 -0
  53. monarch/controller/backend.py +223 -0
  54. monarch/controller/controller.py +223 -0
  55. monarch/controller/debugger.py +47 -0
  56. monarch/controller/history.py +90 -0
  57. monarch/controller/rust_backend/__init__.py +7 -0
  58. monarch/controller/rust_backend/controller.py +245 -0
  59. monarch/debugger.py +379 -0
  60. monarch/fetch.py +55 -0
  61. monarch/future.py +76 -0
  62. monarch/gradient/__init__.py +11 -0
  63. monarch/gradient/_gradient_generator.pyi +22 -0
  64. monarch/gradient/_gradient_generator.so +0 -0
  65. monarch/gradient_generator.py +185 -0
  66. monarch/memory.py +43 -0
  67. monarch/mesh_controller.py +271 -0
  68. monarch/monarch_controller +0 -0
  69. monarch/notebook.py +761 -0
  70. monarch/opaque_module.py +235 -0
  71. monarch/opaque_object.py +88 -0
  72. monarch/parallel/__init__.py +9 -0
  73. monarch/parallel/pipelining/__init__.py +7 -0
  74. monarch/parallel/pipelining/runtime.py +847 -0
  75. monarch/parallel/pipelining/schedule_ir.py +692 -0
  76. monarch/parallel/pipelining/scheduler.py +249 -0
  77. monarch/pdb_wrapper.py +135 -0
  78. monarch/proc_mesh.py +299 -0
  79. monarch/profiler.py +160 -0
  80. monarch/python_local_mesh.py +107 -0
  81. monarch/random.py +61 -0
  82. monarch/rdma.py +162 -0
  83. monarch/remote_class.py +114 -0
  84. monarch/rust_backend_mesh.py +280 -0
  85. monarch/rust_local_mesh.py +1402 -0
  86. monarch/sim_mesh.py +359 -0
  87. monarch/simulator/__init__.py +7 -0
  88. monarch/simulator/command_history.py +424 -0
  89. monarch/simulator/config.py +21 -0
  90. monarch/simulator/interface.py +59 -0
  91. monarch/simulator/ir.py +770 -0
  92. monarch/simulator/mock_controller.py +214 -0
  93. monarch/simulator/profiling.py +424 -0
  94. monarch/simulator/simulator.py +1052 -0
  95. monarch/simulator/task.py +255 -0
  96. monarch/simulator/tensor.py +373 -0
  97. monarch/simulator/trace.py +395 -0
  98. monarch/simulator/utils.py +41 -0
  99. monarch/simulator/worker.py +389 -0
  100. monarch/telemetry.py +19 -0
  101. monarch/tensor_worker_main.py +260 -0
  102. monarch/tensorboard.py +84 -0
  103. monarch/timer/__init__.py +21 -0
  104. monarch/timer/example_monarch.py +78 -0
  105. monarch/timer/example_spmd.py +55 -0
  106. monarch/timer/execution_timer.py +199 -0
  107. monarch/timer/execution_timer_test.py +131 -0
  108. monarch/tools/__init__.py +7 -0
  109. monarch/tools/cli.py +167 -0
  110. monarch/tools/commands.py +251 -0
  111. monarch/tools/components/__init__.py +7 -0
  112. monarch/tools/components/hyperactor.py +58 -0
  113. monarch/tools/config/__init__.py +20 -0
  114. monarch/tools/config/defaults.py +54 -0
  115. monarch/tools/mesh_spec.py +165 -0
  116. monarch/tools/network.py +69 -0
  117. monarch/worker/__init__.py +7 -0
  118. monarch/worker/_testing_function.py +481 -0
  119. monarch/worker/compiled_block.py +270 -0
  120. monarch/worker/debugger.py +125 -0
  121. monarch/worker/lines.py +47 -0
  122. monarch/worker/monitor.py +53 -0
  123. monarch/worker/worker.py +1191 -0
  124. monarch/world_mesh.py +34 -0
  125. monarch_supervisor/__init__.py +1044 -0
  126. monarch_supervisor/_testing.py +44 -0
  127. monarch_supervisor/function_call.py +30 -0
  128. monarch_supervisor/host.py +386 -0
  129. monarch_supervisor/launchers.py +145 -0
  130. monarch_supervisor/log_pstree.py +48 -0
  131. monarch_supervisor/logging.py +103 -0
  132. monarch_supervisor/python_executable.py +42 -0
  133. tests/__init__.py +0 -0
  134. tests/dispatch_bench.py +124 -0
  135. tests/dispatch_bench_helper.py +25 -0
  136. tests/error_test_binary.py +180 -0
  137. tests/simulator/__init__.py +0 -0
  138. tests/simulator/test_profiling.py +136 -0
  139. tests/simulator/test_simulator.py +411 -0
  140. tests/simulator/test_task.py +64 -0
  141. tests/simulator/test_worker.py +102 -0
  142. tests/sleep_binary.py +35 -0
  143. tests/test_actor_error.py +240 -0
  144. tests/test_alloc.py +25 -0
  145. tests/test_allocator.py +365 -0
  146. tests/test_coalescing.py +492 -0
  147. tests/test_controller.py +845 -0
  148. tests/test_device_mesh.py +132 -0
  149. tests/test_fault_tolerance.py +398 -0
  150. tests/test_future.py +94 -0
  151. tests/test_grad_generator.py +121 -0
  152. tests/test_mock_cuda.py +74 -0
  153. tests/test_pdb_actor.py +110 -0
  154. tests/test_python_actors.py +736 -0
  155. tests/test_remote_functions.py +1271 -0
  156. tests/test_rust_backend.py +217 -0
  157. tests/test_signal_safe_block_on.py +103 -0
  158. tests/test_sim_backend.py +54 -0
  159. tests/test_tensor_engine.py +52 -0
  160. torchmonarch_nightly-2025.6.27.dist-info/METADATA +94 -0
  161. torchmonarch_nightly-2025.6.27.dist-info/RECORD +165 -0
  162. torchmonarch_nightly-2025.6.27.dist-info/WHEEL +5 -0
  163. torchmonarch_nightly-2025.6.27.dist-info/entry_points.txt +3 -0
  164. torchmonarch_nightly-2025.6.27.dist-info/licenses/LICENSE +29 -0
  165. torchmonarch_nightly-2025.6.27.dist-info/top_level.txt +3 -0
monarch/world_mesh.py ADDED
@@ -0,0 +1,34 @@
1
+ # Copyright (c) Meta Platforms, Inc. and affiliates.
2
+ # All rights reserved.
3
+ #
4
+ # This source code is licensed under the BSD-style license found in the
5
+ # LICENSE file in the root directory of this source tree.
6
+
7
+ # pyre-unsafe
8
+
9
+ from typing import List
10
+
11
+ from monarch.common.client import Client
12
+
13
+ from monarch.common.device_mesh import DeviceMesh
14
+ from monarch.common.shape import NDSlice
15
+
16
+ from monarch.controller.backend import ProcessBackend
17
+
18
+ from monarch.controller.controller import Controller
19
+ from monarch_supervisor import Context, Host
20
+
21
+
22
+ def world_mesh(
23
+ ctx: Context,
24
+ hosts: List[Host],
25
+ gpu_per_host: int,
26
+ _processes=None,
27
+ ) -> DeviceMesh:
28
+ backend = ProcessBackend(ctx, hosts, gpu_per_host, _processes=_processes)
29
+ client = Client(Controller(backend), backend.world_size, backend.gpu_per_host)
30
+ return DeviceMesh(
31
+ client,
32
+ NDSlice(offset=0, sizes=[len(hosts), gpu_per_host], strides=[gpu_per_host, 1]),
33
+ ("host", "gpu"),
34
+ )