[test] ignore 8 gpu test (#1080)

* [test] ignore 8 gpu test

* polish code

* polish workflow

* polish workflow

tests/test_utils/test_checkpoint/test_checkpoint_1d.py

@@ -67,6 +67,7 @@ def check_checkpoint_1d(rank, world_size, port):
 
 
 @pytest.mark.dist
+@pytest.mark.skip("This test should be invoked with 8 GPUs")
 @rerun_on_exception(exception_type=mp.ProcessRaisedException, pattern=".*Address already in use.*")
 def test_checkpoint_1d():
     world_size = 8

tests/test_utils/test_checkpoint/test_checkpoint_2d.py

@@ -67,6 +67,7 @@ def check_checkpoint_2d(rank, world_size, port):
 
 
 @pytest.mark.dist
+@pytest.mark.skip("This test should be invoked with 8 GPUs")
 @rerun_on_exception(exception_type=mp.ProcessRaisedException, pattern=".*Address already in use.*")
 def test_checkpoint_2d():
     world_size = 8

tests/test_utils/test_checkpoint/test_checkpoint_2p5d.py

@@ -67,6 +67,7 @@ def check_checkpoint_2p5d(rank, world_size, port):
 
 
 @pytest.mark.dist
+@pytest.mark.skip("This test should be invoked with 8 GPUs")
 @rerun_on_exception(exception_type=mp.ProcessRaisedException, pattern=".*Address already in use.*")
 def test_checkpoint_2p5d():
     world_size = 8

tests/test_utils/test_checkpoint/test_checkpoint_3d.py

@@ -67,6 +67,7 @@ def check_checkpoint_3d(rank, world_size, port):
 
 
 @pytest.mark.dist
+@pytest.mark.skip("This test requires 8 GPUs to execute")
 @rerun_on_exception(exception_type=mp.ProcessRaisedException, pattern=".*Address already in use.*")
 def test_checkpoint_3d():
     world_size = 8

tests/test_utils/test_memory.py

@@ -22,7 +22,7 @@ def run_dist(rank, world_size, port):
 
 
 @pytest.mark.dist
-@pytest.mark.parametrize("world_size", [4, 5])
+@pytest.mark.parametrize("world_size", [3, 4])
 def test_memory_utils(world_size):
     run_func = partial(run_dist, world_size=world_size, port=free_port())
     mp.spawn(run_func, nprocs=world_size)