EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

[Misc] Add extra checking to torchair_graph_config. (#1939)

Browse Source 
### What this PR does / why we need it?

cherry-pick #1675  to main
This PR adds validation checking to torchair_graph_config for better
reliability.

Co-authored-by: whx-sjtu <2952154980@qq.com>

### Does this PR introduce _any_ user-facing change?

No

### How was this patch tested?


- vLLM version: v0.10.0
- vLLM main:
2836dd73f1

Signed-off-by: 22dimensions <waitingwind@foxmail.com>
This commit is contained in:
22dimensions
2025-08-01 09:24:11 +08:00
committed by GitHub
parent 2284289880
commit 8cf97d8310
5 changed files with 100 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

68
tests/ut/test_ascend_config.py
View File

@@ -236,3 +236,71 @@ class TestAscendConfig(TestBase):
for model_type, expected_output in test_cases:
self.assertEqual(_check_torchair_supported(model_type),
expected_output)
@_clean_up_ascend_config
def test_ascend_config_load_error(self):
test_vllm_config = VllmConfig()
# graph_batch_sizes should be list.
with self.assertRaises(TypeError):
test_vllm_config.additional_config = {
"torchair_graph_config": {
"graph_batch_sizes": "fake_size",
},
"refresh": True
}
init_ascend_config(test_vllm_config)
# use_cached_graph should not be enabled without torchair graph mode
with self.assertRaises(RuntimeError):
test_vllm_config.additional_config = {
"torchair_graph_config": {
"enabled": False,
"use_cached_graph": True,
},
"refresh": True
}
init_ascend_config(test_vllm_config)
# graph_batch_sizes_init should not be enabled without torchair graph mode
with self.assertRaises(RuntimeError):
test_vllm_config.additional_config = {
"torchair_graph_config": {
"enabled": False,
"graph_batch_sizes_init": True,
},
"refresh": True
}
init_ascend_config(test_vllm_config)
# enable_multistream_mla should not be enabled without torchair graph mode
with self.assertRaises(RuntimeError):
test_vllm_config.additional_config = {
"torchair_graph_config": {
"enabled": False,
"enable_multistream_mla": True,
},
"refresh": True
}
init_ascend_config(test_vllm_config)
# enable_multistream_moe should not be enabled without torchair graph mode
with self.assertRaises(RuntimeError):
test_vllm_config.additional_config = {
"torchair_graph_config": {
"enabled": False,
"enable_multistream_moe": True,
},
"refresh": True
}
init_ascend_config(test_vllm_config)
# enable_kv_nz should not be enabled without torchair graph mode
with self.assertRaises(RuntimeError):
test_vllm_config.additional_config = {
"torchair_graph_config": {
"enabled": False,
"enable_kv_nz": True,
},
"refresh": True
}
init_ascend_config(test_vllm_config)