A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 4, 02:30 UTC
[CI] Narrow engine.yaml source dependencies (#42055)
Signed-off-by: khluu <khluu000@gmail.com>
Kevin H. Luu committed
May 8, 2026 at 14:55 UTC
d2f22dfc9f3551a19988101d742d08b0b925d813
1 file changed
+47
-5
.buildkite/test_areas/engine.yaml
+47
-5
@@ -7,7 +7,17 @@ steps:
7
timeout_in_minutes: 15
8
device: h200_18gb
9
source_file_dependencies:
10
- - vllm/
10
+ - vllm/compilation/
11
+ - vllm/config/
12
+ - vllm/engine/
13
+ - vllm/entrypoints/logger.py
14
+ - vllm/envs.py
15
+ - vllm/logger.py
16
+ - vllm/logging_utils/
17
+ - vllm/platforms/
18
+ - vllm/sequence.py
19
+ - vllm/triton_utils/
20
+ - vllm/utils/
21
- tests/engine
22
- tests/test_sequence
23
- tests/test_config
@@ -52,8 +62,24 @@ steps:
62
optional: true
63
num_devices: 2
64
source_file_dependencies:
55
- - vllm/
56
- - tests/v1/e2e
65
+ - vllm/compilation/
66
+ - vllm/config/
67
+ - vllm/distributed/
68
+ - vllm/engine/
69
+ - vllm/envs.py
70
+ - vllm/forward_context.py
71
+ - vllm/inputs/
72
+ - vllm/logger.py
73
+ - vllm/logging_utils/
74
+ - vllm/model_executor/
75
+ - vllm/multimodal/
76
+ - vllm/platforms/
77
+ - vllm/sampling_params.py
78
+ - vllm/transformers_utils/
79
+ - vllm/triton_utils/
80
+ - vllm/utils/
81
+ - vllm/v1/
82
+ - tests/v1/e2e/spec_decode
83
commands:
84
# Only run tests that need exactly 2 GPUs
85
- pytest -v -s v1/e2e/spec_decode/test_spec_decode.py -k "tensor_parallelism"
@@ -64,8 +90,24 @@ steps:
90
optional: true
91
num_devices: 4
92
source_file_dependencies:
67
- - vllm/
68
- - tests/v1/e2e
93
+ - vllm/compilation/
94
+ - vllm/config/
95
+ - vllm/distributed/
96
+ - vllm/engine/
97
+ - vllm/envs.py
98
+ - vllm/forward_context.py
99
+ - vllm/inputs/
100
+ - vllm/logger.py
101
+ - vllm/logging_utils/
102
+ - vllm/model_executor/
103
+ - vllm/multimodal/
104
+ - vllm/platforms/
105
+ - vllm/sampling_params.py
106
+ - vllm/transformers_utils/
107
+ - vllm/triton_utils/
108
+ - vllm/utils/
109
+ - vllm/v1/
110
+ - tests/v1/e2e/spec_decode
111
commands:
112
# Only run tests that need 4 GPUs
113
- pytest -v -s v1/e2e/spec_decode/test_spec_decode.py -k "eagle_correctness_heavy"