Skip to content
This repository has been archived by the owner on Oct 19, 2024. It is now read-only.

task 801 - launch physical meshes after compilation #938

Open
wants to merge 2 commits into
base: main
Choose a base branch
from

Conversation

haifaksh
Copy link

No description provided.

alpa/pipeline_parallel/runtime_emitter.py Outdated Show resolved Hide resolved
alpa/device_mesh.py Outdated Show resolved Hide resolved
mesh_id: int = None
):
super().__init__(host_ids, host_info, num_devices_per_host, parent, devices)
self.host_ips = []
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This member seems never used. If so, please remove it

alpa/pipeline_parallel/compile_executable.py Outdated Show resolved Hide resolved
alpa/pipeline_parallel/compile_executable.py Outdated Show resolved Hide resolved
@@ -195,6 +195,8 @@ def __init__(self, task_spec, collective_group, src_mesh, dst_mesh):
self.send_worker_task_ids = {}
self.recv_worker_task_ids = {}

self.task_dones = []
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

remove unused code. For a series of tasks, we always create a tmp task dones instead of keeping using the same one

alpa/pipeline_parallel/pipeshard_executable.py Outdated Show resolved Hide resolved
alpa/pipeline_parallel/pipeshard_executable.py Outdated Show resolved Hide resolved
alpa/pipeline_parallel/pipeshard_executable.py Outdated Show resolved Hide resolved
@@ -30,12 +30,14 @@ class CreateStateExecutable(PipeshardDriverExecutable):

def __init__(self,
mesh_group: PhysicalDeviceMeshGroup,
#virtual_mesh_group: VirtualMeshGroup,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

remove unused lines. Also applies for some code below

@@ -100,8 +100,8 @@ def test_pipeshard_parallel(self):
def suite():
suite = unittest.TestSuite()
suite.addTest(CreateStateTest("test_shard_parallel"))
suite.addTest(CreateStateTest("test_shard_parallel_grad_acc"))
suite.addTest(CreateStateTest("test_pipeshard_parallel"))
#suite.addTest(CreateStateTest("test_shard_parallel_grad_acc"))
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

recover the unit test

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants