xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 05:25:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

[ci][distributed] disable ray dag tests (#7099)

Browse Source
This commit is contained in:
youkaichao 2024-08-02 22:32:04 -07:00 committed by GitHub
parent 04e5583425
commit a0d164567c
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 18 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
tests/distributed/test_pipeline_parallel.py
View File

@ -14,36 +14,29 @@ from ..utils import compare_two_settings, fork_new_process_for_each_test
VLLM_MULTI_NODE = os.getenv("VLLM_MULTI_NODE", "0") == "1" VLLM_MULTI_NODE = os.getenv("VLLM_MULTI_NODE", "0") == "1"
@pytest.mark.parametrize( @pytest.mark.parametrize(("TP_SIZE, PP_SIZE, EAGER_MODE, CHUNKED_PREFILL, "
("TP_SIZE, PP_SIZE, EAGER_MODE, CHUNKED_PREFILL, " "MODEL_NAME, DIST_BACKEND"),
"MODEL_NAME, DIST_BACKEND, USE_RAY_ADAG, USE_RAY_ADAG_NCCL"), [ [
(2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", False, False), (2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray"),
(2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", False, False), (2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray"),
(1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray", False, False), (1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray"),
(1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", False, False), (1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray"),
(1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", False, False), (1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray"),
(2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", True, False), (2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "mp"),
(2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, False), (2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "mp"),
(1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, False), (1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "mp"),
(1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", True, False), (1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "mp"),
(1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, False), (1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "mp"),
(2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", True, True), ])
(2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, True),
(1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, True),
(1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", True, True),
(1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, True),
(2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
(2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
(1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
(1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
(1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
])
def test_compare_tp(TP_SIZE, PP_SIZE, EAGER_MODE, CHUNKED_PREFILL, MODEL_NAME, def test_compare_tp(TP_SIZE, PP_SIZE, EAGER_MODE, CHUNKED_PREFILL, MODEL_NAME,
DIST_BACKEND, USE_RAY_ADAG, USE_RAY_ADAG_NCCL): DIST_BACKEND):
if VLLM_MULTI_NODE and DIST_BACKEND == "mp": if VLLM_MULTI_NODE and DIST_BACKEND == "mp":
pytest.skip("Skipping multi-node pipeline parallel test for " pytest.skip("Skipping multi-node pipeline parallel test for "
"multiprocessing distributed backend") "multiprocessing distributed backend")
USE_RAY_ADAG_NCCL = 0
USE_RAY_ADAG = 0
pp_args = [ pp_args = [
# use half precision for speed and memory savings in CI environment # use half precision for speed and memory savings in CI environment
"--dtype", "--dtype",