@tduncan / Nvidia-Optimization-Stack-VLLM / commits / 433f291195

[CI] Right-size test-area timeouts from nightly durations (#48186)

Signed-off-by: khluu <khluu000@gmail.com> Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

Kevin H. Luu committed Jul 10, 2026 at 00:53 UTC 433f291195ded3ca8d278bc78da9280c5d4e5329
30 files changed +183 -183
.buildkite/test_areas/attention.yaml
+3 -3
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: V1 attention (H100-MI300)
6 key: v1-attention-h100-mi300
7 - timeout_in_minutes: 30
7 + timeout_in_minutes: 85
8 device: h100
9 source_file_dependencies:
10 - vllm/config/attention.py
@@ -16,7 +16,7 @@ steps:
16 mirror:
17 amd:
18 device: mi325_1
19 - timeout_in_minutes: 70
19 + timeout_in_minutes: 95
20 depends_on:
21 - image-build-amd
22 source_file_dependencies:
@@ -30,7 +30,7 @@ steps:
30
31 - label: V1 attention (B200)
32 key: v1-attention-b200
33 - timeout_in_minutes: 30
33 + timeout_in_minutes: 80
34 device: b200-k8s
35 source_file_dependencies:
36 - vllm/config/attention.py
.buildkite/test_areas/basic_correctness.yaml
+2 -2
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Basic Correctness
6 key: basic-correctness
7 - timeout_in_minutes: 30
7 + timeout_in_minutes: 45
8 device: h200_18gb
9 source_file_dependencies:
10 - vllm/
@@ -19,6 +19,6 @@ steps:
19 mirror:
20 amd:
21 device: mi325_1
22 - timeout_in_minutes: 50
22 + timeout_in_minutes: 70
23 depends_on:
24 - image-build-amd
.buildkite/test_areas/benchmarks.yaml
+2 -2
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Benchmarks CLI Test
6 key: benchmarks-cli-test
7 - timeout_in_minutes: 20
7 + timeout_in_minutes: 30
8 device: h200_18gb
9 source_file_dependencies:
10 - vllm/
@@ -23,7 +23,7 @@ steps:
23 num_gpus: 2
24 optional: true
25 working_dir: "/vllm-workspace/"
26 - timeout_in_minutes: 10
26 + timeout_in_minutes: 20
27 source_file_dependencies:
28 - benchmarks/attention_benchmarks/
29 - vllm/v1/attention/
.buildkite/test_areas/compile.yaml
+11 -11
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Sequence Parallel Correctness Tests (2 GPUs)
6 key: sequence-parallel-correctness-tests-2-gpus
7 - timeout_in_minutes: 50
7 + timeout_in_minutes: 80
8 working_dir: "/vllm-workspace/"
9 num_devices: 2
10 source_file_dependencies:
@@ -19,7 +19,7 @@ steps:
19
20 - label: Sequence Parallel Correctness Tests (2xH100)
21 key: sequence-parallel-correctness-tests-2xh100
22 - timeout_in_minutes: 50
22 + timeout_in_minutes: 75
23 working_dir: "/vllm-workspace/"
24 device: h100
25 optional: true
@@ -30,7 +30,7 @@ steps:
30
31 - label: AsyncTP Correctness Tests (2xH100)
32 key: asynctp-correctness-tests-2xh100
33 - timeout_in_minutes: 50
33 + timeout_in_minutes: 30
34 working_dir: "/vllm-workspace/"
35 device: h100
36 optional: true
@@ -41,7 +41,7 @@ steps:
41
42 - label: AsyncTP Correctness Tests (B200)
43 key: asynctp-correctness-tests-b200
44 - timeout_in_minutes: 50
44 + timeout_in_minutes: 30
45 working_dir: "/vllm-workspace/"
46 device: b200-k8s
47 optional: true
@@ -52,7 +52,7 @@ steps:
52
53 - label: Distributed Compile Unit Tests (2xH100)
54 key: distributed-compile-unit-tests-2xh100
55 - timeout_in_minutes: 20
55 + timeout_in_minutes: 45
56 working_dir: "/vllm-workspace/"
57 device: h100
58 num_devices: 2
@@ -66,7 +66,7 @@ steps:
66
67 - label: Fusion and Compile Unit Tests (2xB200)
68 key: fusion-and-compile-unit-tests-2xb200
69 - timeout_in_minutes: 20
69 + timeout_in_minutes: 30
70 working_dir: "/vllm-workspace/"
71 device: b200-k8s
72 source_file_dependencies:
@@ -96,7 +96,7 @@ steps:
96
97 - label: Fusion E2E Quick (H100)
98 key: fusion-e2e-quick-h100
99 - timeout_in_minutes: 15
99 + timeout_in_minutes: 25
100 working_dir: "/vllm-workspace/"
101 device: h100
102 num_devices: 1
@@ -115,7 +115,7 @@ steps:
115
116 - label: Fusion E2E Config Sweep (H100)
117 key: fusion-e2e-config-sweep-h100
118 - timeout_in_minutes: 30
118 + timeout_in_minutes: 25
119 working_dir: "/vllm-workspace/"
120 device: h100
121 num_devices: 1
@@ -149,7 +149,7 @@ steps:
149
150 - label: Fusion E2E TP2 Quick (H100)
151 key: fusion-e2e-tp2-quick-h100
152 - timeout_in_minutes: 20
152 + timeout_in_minutes: 35
153 working_dir: "/vllm-workspace/"
154 device: h100
155 num_devices: 2
@@ -167,7 +167,7 @@ steps:
167
168 - label: Fusion E2E TP2 AR-RMS Config Sweep (H100)
169 key: fusion-e2e-tp2-ar-rms-config-sweep-h100
170 - timeout_in_minutes: 40
170 + timeout_in_minutes: 30
171 working_dir: "/vllm-workspace/"
172 device: h100
173 num_devices: 2
@@ -207,7 +207,7 @@ steps:
207
208 - label: Fusion E2E TP2 (B200)
209 key: fusion-e2e-tp2-b200
210 - timeout_in_minutes: 20
210 + timeout_in_minutes: 45
211 working_dir: "/vllm-workspace/"
212 device: b200-k8s
213 num_devices: 2
.buildkite/test_areas/cuda.yaml
+2 -2
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Platform Tests
6 key: platform-tests
7 - timeout_in_minutes: 15
7 + timeout_in_minutes: 20
8 device: h200_18gb
9 source_file_dependencies:
10 - vllm/envs.py
@@ -19,7 +19,7 @@ steps:
19
20 - label: Cudagraph
21 key: cudagraph
22 - timeout_in_minutes: 20
22 + timeout_in_minutes: 30
23 source_file_dependencies:
24 - tests/v1/cudagraph
25 - vllm/v1/cudagraph_dispatcher.py
.buildkite/test_areas/disaggregated.yaml
+12 -12
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Distributed NixlConnector PD accuracy (4 GPUs)
6 key: distributed-nixlconnector-pd-accuracy-4-gpus
7 - timeout_in_minutes: 30
7 + timeout_in_minutes: 55
8 working_dir: "/vllm-workspace/tests"
9 num_devices: 4
10 source_file_dependencies:
@@ -16,7 +16,7 @@ steps:
16 mirror:
17 amd:
18 device: mi300_4
19 - timeout_in_minutes: 110
19 + timeout_in_minutes: 85
20 depends_on:
21 - image-build-amd
22 source_file_dependencies:
@@ -29,7 +29,7 @@ steps:
29
30 - label: Distributed FlashInfer NixlConnector PD accuracy (4 GPUs)
31 key: distributed-flashinfer-nixlconnector-pd-accuracy-4-gpus
32 - timeout_in_minutes: 30
32 + timeout_in_minutes: 55
33 working_dir: "/vllm-workspace/tests"
34 num_devices: 4
35 source_file_dependencies:
@@ -66,7 +66,7 @@ steps:
66 mirror:
67 amd:
68 device: mi300_4
69 - timeout_in_minutes: 50
69 + timeout_in_minutes: 60
70 depends_on:
71 - image-build-amd
72 source_file_dependencies:
@@ -79,7 +79,7 @@ steps:
79
80 - label: CrossLayer KV layout Distributed NixlConnector PD accuracy tests (4 GPUs)
81 key: crosslayer-kv-layout-distributed-nixlconnector-pd-accuracy-tests-4-gpus
82 - timeout_in_minutes: 30
82 + timeout_in_minutes: 55
83 working_dir: "/vllm-workspace/tests"
84 num_devices: 4
85 source_file_dependencies:
@@ -91,7 +91,7 @@ steps:
91 mirror:
92 amd:
93 device: mi300_4
94 - timeout_in_minutes: 110
94 + timeout_in_minutes: 85
95 depends_on:
96 - image-build-amd
97 source_file_dependencies:
@@ -104,7 +104,7 @@ steps:
104
105 - label: Hybrid SSM NixlConnector PD accuracy tests (4 GPUs)
106 key: hybrid-ssm-nixlconnector-pd-accuracy-tests-4-gpus
107 - timeout_in_minutes: 25
107 + timeout_in_minutes: 60
108 working_dir: "/vllm-workspace/tests"
109 num_devices: 4
110 source_file_dependencies:
@@ -116,7 +116,7 @@ steps:
116 mirror:
117 amd:
118 device: mi300_4
119 - timeout_in_minutes: 60
119 + timeout_in_minutes: 80
120 depends_on:
121 - image-build-amd
122 source_file_dependencies:
@@ -143,7 +143,7 @@ steps:
143
144 - label: MultiConnector (Nixl+Offloading) PD accuracy (2 GPUs)
145 key: multiconnector-nixl-offloading-pd-accuracy-2-gpus
146 - timeout_in_minutes: 30
146 + timeout_in_minutes: 40
147 working_dir: "/vllm-workspace/tests"
148 num_devices: 2
149 source_file_dependencies:
@@ -158,7 +158,7 @@ steps:
158
159 - label: NixlConnector PD + Spec Decode acceptance (2 GPUs)
160 key: nixlconnector-pd-spec-decode-acceptance-2-gpus
161 - timeout_in_minutes: 30
161 + timeout_in_minutes: 45
162 device: a100
163 working_dir: "/vllm-workspace/tests"
164 num_devices: 2
@@ -172,7 +172,7 @@ steps:
172 mirror:
173 amd:
174 device: mi300_2
175 - timeout_in_minutes: 60
175 + timeout_in_minutes: 70
176 depends_on:
177 - image-build-amd
178 source_file_dependencies:
@@ -186,7 +186,7 @@ steps:
186
187 - label: MultiConnector (Nixl+Offloading) PD edge cases (2 GPUs)
188 key: multiconnector-nixl-offloading-pd-edge-cases-2-gpus
189 - timeout_in_minutes: 30
189 + timeout_in_minutes: 25
190 working_dir: "/vllm-workspace/tests"
191 num_devices: 2
192 source_file_dependencies:
.buildkite/test_areas/distributed.yaml
+10 -10
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Distributed Comm Ops
6 key: distributed-comm-ops
7 - timeout_in_minutes: 20
7 + timeout_in_minutes: 25
8 working_dir: "/vllm-workspace/tests"
9 num_devices: 2
10 source_file_dependencies:
@@ -18,7 +18,7 @@ steps:
18
19 - label: Distributed DP Tests (2 GPUs)
20 key: distributed-dp-tests-2-gpus
21 - timeout_in_minutes: 20
21 + timeout_in_minutes: 35
22 working_dir: "/vllm-workspace/tests"
23 num_devices: 2
24 source_file_dependencies:
@@ -55,7 +55,7 @@ steps:
55
56 - label: Distributed Compile + RPC Tests (2 GPUs)
57 key: distributed-compile-rpc-tests-2-gpus
58 - timeout_in_minutes: 20
58 + timeout_in_minutes: 65
59 working_dir: "/vllm-workspace/tests"
60 num_devices: 2
61 source_file_dependencies:
@@ -78,7 +78,7 @@ steps:
78
79 - label: Distributed Torchrun + Shutdown Tests (2 GPUs)
80 key: distributed-torchrun-shutdown-tests-2-gpus
81 - timeout_in_minutes: 20
81 + timeout_in_minutes: 30
82 working_dir: "/vllm-workspace/tests"
83 num_devices: 2
84 source_file_dependencies:
@@ -133,7 +133,7 @@ steps:
133
134 - label: Distributed DP Tests (4 GPUs)
135 key: distributed-dp-tests-4-gpus
136 - timeout_in_minutes: 30
136 + timeout_in_minutes: 45
137 working_dir: "/vllm-workspace/tests"
138 num_devices: 4
139 source_file_dependencies:
@@ -154,7 +154,7 @@ steps:
154
155 - label: Distributed Compile + Comm (4 GPUs)
156 key: distributed-compile-comm-4-gpus
157 - timeout_in_minutes: 30
157 + timeout_in_minutes: 70
158 working_dir: "/vllm-workspace/tests"
159 num_devices: 4
160 source_file_dependencies:
@@ -176,7 +176,7 @@ steps:
176
177 - label: Distributed Tests (8xH100)
178 key: distributed-tests-8xh100
179 - timeout_in_minutes: 10
179 + timeout_in_minutes: 20
180 device: h100
181 num_devices: 8
182 working_dir: "/vllm-workspace/tests"
@@ -212,7 +212,7 @@ steps:
212
213 - label: Distributed Tests (2xH100-2xMI300)
214 key: distributed-tests-2xh100-2xmi300
215 - timeout_in_minutes: 15
215 + timeout_in_minutes: 30
216 device: h100
217 optional: true
218 working_dir: "/vllm-workspace/"
@@ -259,7 +259,7 @@ steps:
259
260 - label: Pipeline + Context Parallelism (4 GPUs)
261 key: pipeline-context-parallelism-4-gpus
262 - timeout_in_minutes: 60
262 + timeout_in_minutes: 55
263 working_dir: "/vllm-workspace/tests"
264 num_devices: 4
265 source_file_dependencies:
@@ -274,7 +274,7 @@ steps:
274
275 - label: RayExecutorV2 (4 GPUs)
276 key: rayexecutorv2-4-gpus
277 - timeout_in_minutes: 60
277 + timeout_in_minutes: 45
278 working_dir: "/vllm-workspace/tests"
279 num_devices: 4
280 source_file_dependencies:
.buildkite/test_areas/docker.yaml
+1 -1
@@ -3,7 +3,7 @@ depends_on:
3 - image-build-cpu
4 steps:
5 - label: Docker Build Metadata
6 - timeout_in_minutes: 10
6 + timeout_in_minutes: 20
7 device: cpu-small
8 source_file_dependencies:
9 - .buildkite/release-pipeline.yaml
.buildkite/test_areas/e2e_integration.yaml
+5 -5
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: DeepSeek V2-Lite Sync EPLB Accuracy (4xH100)
6 key: deepseek-v2-lite-sync-eplb-accuracy-4xh100
7 - timeout_in_minutes: 60
7 + timeout_in_minutes: 25
8 device: h100
9 optional: true
10 num_devices: 4
@@ -14,7 +14,7 @@ steps:
14
15 - label: Qwen3-30B-A3B-FP8-block Sync EPLB Accuracy (4xH100)
16 key: qwen3-30b-a3b-fp8-block-sync-eplb-accuracy-4xh100
17 - timeout_in_minutes: 60
17 + timeout_in_minutes: 25
18 device: h100
19 optional: true
20 num_devices: 4
@@ -24,7 +24,7 @@ steps:
24
25 - label: Qwen3-30B-A3B-FP8-block Sync EPLB Accuracy (2xB200)
26 key: qwen3-30b-a3b-fp8-block-sync-eplb-accuracy-2xb200
27 - timeout_in_minutes: 60
27 + timeout_in_minutes: 20
28 device: b200-k8s
29 optional: true
30 num_devices: 2
@@ -34,7 +34,7 @@ steps:
34
35 - label: Qwen3-30B-A3B-FP8 DP4 Async EPLB Accuracy
36 key: qwen3-30b-a3b-fp8-dp4-async-eplb-accuracy
37 - timeout_in_minutes: 60
37 + timeout_in_minutes: 25
38 device: h100
39 optional: true
40 num_devices: 4
@@ -44,7 +44,7 @@ steps:
44
45 - label: DeepSeek V2-Lite Prefetch Offload Accuracy (H100)
46 key: deepseek-v2-lite-prefetch-offload-accuracy-h100
47 - timeout_in_minutes: 60
47 + timeout_in_minutes: 20
48 device: h100
49 optional: true
50 num_devices: 1
.buildkite/test_areas/engine.yaml
+11 -11
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Engine
6 key: engine
7 - timeout_in_minutes: 15
7 + timeout_in_minutes: 30
8 device: h200_18gb
9 source_file_dependencies:
10 - vllm/compilation/
@@ -29,13 +29,13 @@ steps:
29 mirror:
30 amd:
31 device: mi325_1
32 - timeout_in_minutes: 60
32 + timeout_in_minutes: 50
33 depends_on:
34 - image-build-amd
35
36 - label: Engine (1 GPU)
37 key: engine-1-gpu
38 - timeout_in_minutes: 30
38 + timeout_in_minutes: 45
39 source_file_dependencies:
40 - vllm/v1/engine/
41 - tests/v1/engine/
@@ -45,13 +45,13 @@ steps:
45 mirror:
46 amd:
47 device: mi325_1
48 - timeout_in_minutes: 40
48 + timeout_in_minutes: 55
49 depends_on:
50 - image-build-amd
51
52 - label: e2e Scheduling (1 GPU)
53 key: e2e-scheduling-1-gpu
54 - timeout_in_minutes: 30
54 + timeout_in_minutes: 35
55 device: h200_18gb
56 source_file_dependencies:
57 - vllm/v1/
@@ -61,14 +61,14 @@ steps:
61 mirror:
62 amd:
63 device: mi325_1
64 - timeout_in_minutes: 60
64 + timeout_in_minutes: 70
65 depends_on:
66 - image-build-amd
67
68 - label: e2e Core (1 GPU)
69 device: h200_35gb
70 key: e2e-core-1-gpu
71 - timeout_in_minutes: 30
71 + timeout_in_minutes: 40
72 source_file_dependencies:
73 - vllm/v1/
74 - tests/v1/e2e/general/
@@ -77,7 +77,7 @@ steps:
77 mirror:
78 amd:
79 device: mi325_1
80 - timeout_in_minutes: 35
80 + timeout_in_minutes: 60
81 depends_on:
82 - image-build-amd
83 source_file_dependencies:
@@ -87,7 +87,7 @@ steps:
87
88 - label: V1 e2e (2 GPUs)
89 key: v1-e2e-2-gpus
90 - timeout_in_minutes: 60 # TODO: Fix timeout after we have more confidence in the test stability
90 + timeout_in_minutes: 25 # TODO: Fix timeout after we have more confidence in the test stability
91 optional: true
92 num_devices: 2
93 source_file_dependencies:
@@ -120,7 +120,7 @@ steps:
120
121 - label: V1 e2e (4 GPUs)
122 key: v1-e2e-4-gpus
123 - timeout_in_minutes: 60 # TODO: Fix timeout after we have more confidence in the test stability
123 + timeout_in_minutes: 20 # TODO: Fix timeout after we have more confidence in the test stability
124 optional: true
125 num_devices: 4
126 source_file_dependencies:
@@ -148,7 +148,7 @@ steps:
148
149 - label: V1 e2e (4xH100)
150 key: v1-e2e-4xh100
151 - timeout_in_minutes: 60
151 + timeout_in_minutes: 35
152 device: h100
153 num_devices: 4
154 optional: true
.buildkite/test_areas/entrypoints.yaml
+10 -10
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Entrypoints Unit Tests
6 key: entrypoints-unit-tests
7 - timeout_in_minutes: 10
7 + timeout_in_minutes: 25
8 working_dir: "/vllm-workspace/tests"
9 source_file_dependencies:
10 - vllm/entrypoints
@@ -16,7 +16,7 @@ steps:
16
17 - label: Entrypoints Integration (LLM)
18 key: entrypoints-integration-llm
19 - timeout_in_minutes: 40
19 + timeout_in_minutes: 60
20 working_dir: "/vllm-workspace/tests"
21 source_file_dependencies:
22 - vllm/
@@ -37,7 +37,7 @@ steps:
37 - label: Entrypoints Integration (API Server)
38 key: entrypoints-integration-api-server
39 device: h200_35gb
40 - timeout_in_minutes: 130
40 + timeout_in_minutes: 50
41 working_dir: "/vllm-workspace/tests"
42 source_file_dependencies:
43 - vllm/
@@ -56,7 +56,7 @@ steps:
56
57 - label: Entrypoints Integration (API Server OpenAI - Part 1)
58 key: entrypoints-integration-api-server-openai-part-1
59 - timeout_in_minutes: 50
59 + timeout_in_minutes: 45
60 working_dir: "/vllm-workspace/tests"
61 source_file_dependencies:
62 - vllm/
@@ -68,13 +68,13 @@ steps:
68 mirror:
69 amd:
70 device: mi325_1
71 - timeout_in_minutes: 80
71 + timeout_in_minutes: 65
72 depends_on:
73 - image-build-amd
74
75 - label: Entrypoints Integration (API Server OpenAI - Part 2)
76 key: entrypoints-integration-api-server-openai-part-2
77 - timeout_in_minutes: 50
77 + timeout_in_minutes: 45
78 working_dir: "/vllm-workspace/tests"
79 source_file_dependencies:
80 - vllm/
@@ -109,7 +109,7 @@ steps:
109 mirror:
110 amd:
111 device: mi325_1
112 - timeout_in_minutes: 60
112 + timeout_in_minutes: 65
113 depends_on:
114 - image-build-amd
115
@@ -126,7 +126,7 @@ steps:
126 - label: Entrypoints Integration (Speech to Text)
127 device: h200_35gb
128 key: entrypoints-integration-speech_to_text
129 - timeout_in_minutes: 50
129 + timeout_in_minutes: 45
130 working_dir: "/vllm-workspace/tests"
131 source_file_dependencies:
132 - vllm/
@@ -138,7 +138,7 @@ steps:
138 - label: Entrypoints Integration (Multimodal)
139 device: h200_35gb
140 key: entrypoints-integration-multimodal
141 - timeout_in_minutes: 50
141 + timeout_in_minutes: 45
142 working_dir: "/vllm-workspace/tests"
143 source_file_dependencies:
144 - vllm/
@@ -160,7 +160,7 @@ steps:
160
161 - label: OpenAI API Correctness
162 key: openai-api-correctness
163 - timeout_in_minutes: 30
163 + timeout_in_minutes: 20
164 device: h200_18gb
165 source_file_dependencies:
166 - csrc/
.buildkite/test_areas/expert_parallelism.yaml
+3 -3
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: EPLB Algorithm
6 key: eplb-algorithm
7 - timeout_in_minutes: 15
7 + timeout_in_minutes: 20
8 device: h200_18gb
9 working_dir: "/vllm-workspace/tests"
10 source_file_dependencies:
@@ -27,7 +27,7 @@ steps:
27
28 - label: EPLB Execution # 17min
29 key: eplb-execution
30 - timeout_in_minutes: 27
30 + timeout_in_minutes: 25
31 working_dir: "/vllm-workspace/tests"
32 num_devices: 4
33 source_file_dependencies:
@@ -39,7 +39,7 @@ steps:
39
40 - label: Elastic EP Scaling Test
41 key: elastic-ep-scaling-test
42 - timeout_in_minutes: 20
42 + timeout_in_minutes: 30
43 device: h100
44 working_dir: "/vllm-workspace/tests"
45 num_devices: 4
.buildkite/test_areas/kernels.yaml
+19 -19
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: vLLM IR Tests
6 key: vllm-ir-tests
7 - timeout_in_minutes: 10
7 + timeout_in_minutes: 35
8 device: h200_18gb
9 working_dir: "/vllm-workspace/"
10 source_file_dependencies:
@@ -16,7 +16,7 @@ steps:
16
17 - label: Kernels Core Operation Test
18 key: kernels-core-operation-test
19 - timeout_in_minutes: 75
19 + timeout_in_minutes: 120
20 source_file_dependencies:
21 - csrc/
22 - tests/kernels/core
@@ -27,7 +27,7 @@ steps:
27
28 - label: Kernels MiniMax Reduce RMS Test (2 GPUs)
29 key: kernels-minimax-reduce-rms-test-2-gpus
30 - timeout_in_minutes: 15
30 + timeout_in_minutes: 20
31 num_devices: 2
32 device: h100
33 source_file_dependencies:
@@ -41,7 +41,7 @@ steps:
41
42 - label: Deepseek V4 Kernel Test (H100)
43 key: deepseek-v4-kernel-test-h100
44 - timeout_in_minutes: 15
44 + timeout_in_minutes: 30
45 device: h100
46 source_file_dependencies:
47 - csrc/fused_deepseek_v4_qnorm_rope_kv_insert_kernel.cu
@@ -54,7 +54,7 @@ steps:
54
55 - label: Deepseek V4 Kernel Test (B200)
56 key: deepseek-v4-kernel-test-b200
57 - timeout_in_minutes: 15
57 + timeout_in_minutes: 20
58 device: b200-k8s
59 source_file_dependencies:
60 - csrc/fused_deepseek_v4_qnorm_rope_kv_insert_kernel.cu
@@ -65,7 +65,7 @@ steps:
65
66 - label: Kernels Attention Test %N
67 key: kernels-attention-test
68 - timeout_in_minutes: 35
68 + timeout_in_minutes: 65
69 source_file_dependencies:
70 - csrc/attention/
71 - vllm/v1/attention
@@ -79,7 +79,7 @@ steps:
79 mirror:
80 amd:
81 device: mi325_1
82 - timeout_in_minutes: 55
82 + timeout_in_minutes: 90
83 depends_on:
84 - image-build-amd
85 source_file_dependencies:
@@ -106,7 +106,7 @@ steps:
106
107 - label: Kernels Quantization Test %N
108 key: kernels-quantization-test
109 - timeout_in_minutes: 90
109 + timeout_in_minutes: 60
110 source_file_dependencies:
111 - csrc/quantization/
112 - vllm/model_executor/layers/quantization
@@ -131,7 +131,7 @@ steps:
131
132 - label: Kernels MoE Test %N
133 key: kernels-moe-test
134 - timeout_in_minutes: 25
134 + timeout_in_minutes: 50
135 source_file_dependencies:
136 - csrc/quantization/cutlass_w8a8/moe/
137 - csrc/moe/
@@ -147,7 +147,7 @@ steps:
147 mirror:
148 amd:
149 device: mi325_1
150 - timeout_in_minutes: 50
150 + timeout_in_minutes: 65
151 source_file_dependencies:
152 - csrc/quantization/cutlass_w8a8/moe/
153 - csrc/moe/
@@ -163,7 +163,7 @@ steps:
163
164 - label: Kernels Mamba Test
165 key: kernels-mamba-test
166 - timeout_in_minutes: 45
166 + timeout_in_minutes: 40
167 source_file_dependencies:
168 - csrc/mamba/
169 - tests/kernels/mamba
@@ -172,7 +172,7 @@ steps:
172 - pytest -v -s kernels/mamba
173
174 - label: Kernels KDA Test
175 - timeout_in_minutes: 20
175 + timeout_in_minutes: 25
176 device: h200_18gb
177 source_file_dependencies:
178 - vllm/model_executor/layers/fla/ops/kda.py
@@ -184,7 +184,7 @@ steps:
184
185 - label: Kernels DeepGEMM Test (H100)
186 key: kernels-deepgemm-test-h100
187 - timeout_in_minutes: 45
187 + timeout_in_minutes: 35
188 device: h100
189 num_devices: 1
190 source_file_dependencies:
@@ -211,7 +211,7 @@ steps:
211
212 - label: Kernels (B200)
213 key: kernels-b200
214 - timeout_in_minutes: 30
214 + timeout_in_minutes: 80
215 working_dir: "/vllm-workspace/"
216 device: b200-k8s
217 # optional: true
@@ -264,7 +264,7 @@ steps:
264
265 - label: Kernels Helion Test
266 key: kernels-helion-test
267 - timeout_in_minutes: 30
267 + timeout_in_minutes: 115
268 device: h100
269 source_file_dependencies:
270 - vllm/utils/import_utils.py
@@ -276,7 +276,7 @@ steps:
276
277 - label: Kernels FP8 MoE Test (1xH100)
278 key: kernels-fp8-moe-test-1xh100
279 - timeout_in_minutes: 90
279 + timeout_in_minutes: 40
280 device: h100
281 num_devices: 1
282 optional: true
@@ -293,7 +293,7 @@ steps:
293
294 - label: Kernels FP8 MoE Test (2xH100)
295 key: kernels-fp8-moe-test-2xh100
296 - timeout_in_minutes: 90
296 + timeout_in_minutes: 45
297 device: h100
298 num_devices: 2
299 optional: true
@@ -303,7 +303,7 @@ steps:
303
304 - label: Kernels Fp4 MoE Test (B200)
305 key: kernels-fp4-moe-test-b200
306 - timeout_in_minutes: 60
306 + timeout_in_minutes: 25
307 device: b200-k8s
308 num_devices: 1
309 optional: true
@@ -316,7 +316,7 @@ steps:
316
317 - label: Kernels FusedMoE Layer Test (2 H100s)
318 key: kernels-fusedmoe-layer-test-2-h100s
319 - timeout_in_minutes: 90
319 + timeout_in_minutes: 30
320 device: h100
321 num_devices: 2
322 source_file_dependencies:
.buildkite/test_areas/lm_eval.yaml
+19 -19
@@ -5,7 +5,7 @@ steps:
5 - label: LM Eval Small Models
6 device: h200_35gb
7 key: lm-eval-small-models
8 - timeout_in_minutes: 75
8 + timeout_in_minutes: 45
9 source_file_dependencies:
10 - csrc/
11 - vllm/model_executor/layers/quantization
@@ -56,7 +56,7 @@ steps:
56
57 - label: LM Eval Small Models (1xB200)
58 key: lm-eval-small-models-1xb200
59 - timeout_in_minutes: 120
59 + timeout_in_minutes: 50
60 device: b200-k8s
61 optional: true
62 source_file_dependencies:
@@ -80,7 +80,7 @@ steps:
80
81 - label: LM Eval Large Models EP (2xB200)
82 key: lm-eval-large-models-ep-2xb200
83 - timeout_in_minutes: 120
83 + timeout_in_minutes: 60
84 device: b200-k8s
85 optional: true
86 num_devices: 2
@@ -92,7 +92,7 @@ steps:
92
93 - label: LM Eval Qwen3.5 Models (2xB200)
94 key: lm-eval-qwen3-5-models-2xb200
95 - timeout_in_minutes: 120
95 + timeout_in_minutes: 45
96 device: b200-k8s
97 optional: true
98 num_devices: 2
@@ -109,7 +109,7 @@ steps:
109
110 - label: LM Eval Large Models (8xH200)
111 key: lm-eval-large-models-8xh200
112 - timeout_in_minutes: 60
112 + timeout_in_minutes: 50
113 device: h200
114 optional: true
115 num_devices: 8
@@ -118,7 +118,7 @@ steps:
118 mirror:
119 amd:
120 device: mi300_8
121 - timeout_in_minutes: 180
121 + timeout_in_minutes: 60
122 depends_on:
123 - image-build-amd
124 commands:
@@ -152,7 +152,7 @@ steps:
152
153 - label: LM Eval Humming f16 (A100 - TEMPORARY)
154 key: lm-eval-humming-f16-a100
155 - timeout_in_minutes: 120
155 + timeout_in_minutes: 75
156 device: a100
157 optional: true
158 num_devices: 1
@@ -167,7 +167,7 @@ steps:
167
168 - label: LM Eval Humming Act int8 (A100 - TEMPORARY)
169 key: lm-eval-humming-act-a100
170 - timeout_in_minutes: 120
170 + timeout_in_minutes: 45
171 device: a100
172 optional: true
173 num_devices: 1
@@ -182,7 +182,7 @@ steps:
182
183 - label: LM Eval Humming f16 (H100 - TEMPORARY)
184 key: lm-eval-humming-f16-h100
185 - timeout_in_minutes: 120
185 + timeout_in_minutes: 70
186 device: h100
187 optional: true
188 num_devices: 1
@@ -197,7 +197,7 @@ steps:
197
198 - label: LM Eval Humming Act fp8/int8 (H100 - TEMPORARY)
199 key: lm-eval-humming-act-h100
200 - timeout_in_minutes: 120
200 + timeout_in_minutes: 70
201 device: h100
202 optional: true
203 num_devices: 1
@@ -213,7 +213,7 @@ steps:
213
214 - label: LM Eval Humming f16 (B200 - TEMPORARY)
215 key: lm-eval-humming-f16-b200
216 - timeout_in_minutes: 120
216 + timeout_in_minutes: 50
217 device: b200-k8s
218 optional: true
219 num_devices: 1
@@ -228,7 +228,7 @@ steps:
228
229 - label: LM Eval Humming Act fp8/int8 (B200 - TEMPORARY)
230 key: lm-eval-humming-act-b200
231 - timeout_in_minutes: 120
231 + timeout_in_minutes: 50
232 device: b200-k8s
233 optional: true
234 num_devices: 1
@@ -244,7 +244,7 @@ steps:
244
245 - label: LM Eval TurboQuant KV Cache
246 key: lm-eval-turboquant-kv-cache
247 - timeout_in_minutes: 75
247 + timeout_in_minutes: 55
248 device: h200_18gb
249 source_file_dependencies:
250 - vllm/model_executor/layers/quantization/turboquant/
@@ -256,7 +256,7 @@ steps:
256
257 - label: GPQA Eval (GPT-OSS) (2xH100)
258 key: gpqa-eval-gpt-oss-2xh100
259 - timeout_in_minutes: 120
259 + timeout_in_minutes: 35
260 device: h100
261 optional: true
262 num_devices: 2
@@ -270,7 +270,7 @@ steps:
270
271 - label: GPQA Eval (GPT-OSS) (2xB200)
272 key: gpqa-eval-gpt-oss-2xb200
273 - timeout_in_minutes: 120
273 + timeout_in_minutes: 30
274 device: b200-k8s
275 optional: true
276 num_devices: 2
@@ -284,7 +284,7 @@ steps:
284
285 - label: GPQA Eval (GPT-OSS) (DGX Spark)
286 key: gpqa-eval-gpt-oss-spark
287 - timeout_in_minutes: 120
287 + timeout_in_minutes: 35
288 device: dgx-spark
289 optional: true
290 num_devices: 1
@@ -313,7 +313,7 @@ steps:
313
314 - label: LM Eval KV-Offload (2xH100)
315 key: kv-offload-medium
316 - timeout_in_minutes: 60
316 + timeout_in_minutes: 30
317 device: h100
318 num_devices: 2
319 source_file_dependencies:
@@ -327,7 +327,7 @@ steps:
327
328 - label: LM Eval KV-Offload (4xH100)
329 key: kv-offload-large
330 - timeout_in_minutes: 60
330 + timeout_in_minutes: 40
331 device: h100
332 num_devices: 4
333 source_file_dependencies:
@@ -341,7 +341,7 @@ steps:
341
342 - label: MRCR Eval Small Models
343 device: h200_35gb
344 - timeout_in_minutes: 30
344 + timeout_in_minutes: 25
345 source_file_dependencies:
346 - tests/evals/mrcr/
347 commands:
.buildkite/test_areas/lora.yaml
+3 -3
@@ -5,7 +5,7 @@ steps:
5 - label: LoRA %N
6 device: h200_35gb
7 key: lora
8 - timeout_in_minutes: 30
8 + timeout_in_minutes: 40
9 source_file_dependencies:
10 - vllm/lora
11 - tests/lora
@@ -16,7 +16,7 @@ steps:
16 amd:
17 device: mi325_1
18 working_dir: "/vllm-workspace/tests"
19 - timeout_in_minutes: 60
19 + timeout_in_minutes: 65
20 source_file_dependencies:
21 - vllm/lora
22 - tests/lora
@@ -27,7 +27,7 @@ steps:
27
28 - label: LoRA TP (Distributed)
29 key: lora-tp-distributed
30 - timeout_in_minutes: 30
30 + timeout_in_minutes: 60
31 num_devices: 4
32 source_file_dependencies:
33 - vllm/lora
.buildkite/test_areas/misc.yaml
+15 -15
@@ -5,7 +5,7 @@ steps:
5 - label: V1 Spec Decode
6 device: h200_35gb
7 key: v1-spec-decode
8 - timeout_in_minutes: 30
8 + timeout_in_minutes: 40
9 source_file_dependencies:
10 - vllm/config/
11 - vllm/distributed/
@@ -24,13 +24,13 @@ steps:
24 mirror:
25 amd:
26 device: mi300_1
27 - timeout_in_minutes: 65
27 + timeout_in_minutes: 75
28 depends_on:
29 - image-build-amd
30
31 - label: V1 Sample + Logits
32 key: v1-sample-logits
33 - timeout_in_minutes: 30
33 + timeout_in_minutes: 45
34 device: h200_18gb
35 source_file_dependencies:
36 - vllm/config/
@@ -64,7 +64,7 @@ steps:
64
65 - label: V1 Core + KV + Metrics
66 key: v1-core-kv-metrics
67 - timeout_in_minutes: 30
67 + timeout_in_minutes: 60
68 source_file_dependencies:
69 - vllm/config/
70 - vllm/distributed/
@@ -108,7 +108,7 @@ steps:
108 mirror:
109 amd:
110 device: mi325_1
111 - timeout_in_minutes: 60
111 + timeout_in_minutes: 75
112 depends_on:
113 - image-build-amd
114
@@ -172,7 +172,7 @@ steps:
172
173 - label: Regression
174 key: regression
175 - timeout_in_minutes: 20
175 + timeout_in_minutes: 30
176 device: h200_18gb
177 source_file_dependencies:
178 - vllm/config/
@@ -195,7 +195,7 @@ steps:
195 - label: Examples
196 device: h200_35gb
197 key: examples
198 - timeout_in_minutes: 45
198 + timeout_in_minutes: 40
199 working_dir: "/vllm-workspace/examples"
200 source_file_dependencies:
201 - vllm/entrypoints
@@ -237,7 +237,7 @@ steps:
237
238 - label: Metrics, Tracing (2 GPUs)
239 key: metrics-tracing-2-gpus
240 - timeout_in_minutes: 20
240 + timeout_in_minutes: 25
241 num_devices: 2
242 source_file_dependencies:
243 - vllm/config/
@@ -281,7 +281,7 @@ steps:
281 mirror:
282 amd:
283 device: mi325_1
284 - timeout_in_minutes: 20
284 + timeout_in_minutes: 45
285 depends_on:
286 - image-build-amd
287 source_file_dependencies:
@@ -292,7 +292,7 @@ steps:
292 - label: Async Engine, Inputs, Utils, Worker
293 device: h200_35gb
294 key: async-engine-inputs-utils-worker
295 - timeout_in_minutes: 50
295 + timeout_in_minutes: 25
296 source_file_dependencies:
297 - vllm/assets/
298 - vllm/config/
@@ -319,7 +319,7 @@ steps:
319 key: async-engine-inputs-utils-worker-config-cpu
320 depends_on:
321 - image-build-cpu
322 - timeout_in_minutes: 30
322 + timeout_in_minutes: 65
323 source_file_dependencies:
324 - vllm/assets/
325 - vllm/config/
@@ -381,7 +381,7 @@ steps:
381
382 - label: Batch Invariance (A100)
383 key: batch-invariance-a100
384 - timeout_in_minutes: 30
384 + timeout_in_minutes: 40
385 device: a100
386 source_file_dependencies:
387 - vllm/v1/attention
@@ -395,7 +395,7 @@ steps:
395
396 - label: Batch Invariance (H100)
397 key: batch-invariance-h100
398 - timeout_in_minutes: 30
398 + timeout_in_minutes: 40
399 device: h100
400 source_file_dependencies:
401 - vllm/v1/attention
@@ -411,7 +411,7 @@ steps:
411
412 - label: Batch Invariance (B200)
413 key: batch-invariance-b200
414 - timeout_in_minutes: 30
414 + timeout_in_minutes: 35
415 device: b200-k8s
416 source_file_dependencies:
417 - vllm/v1/attention
@@ -430,7 +430,7 @@ steps:
430 - label: Acceptance Length Test (Large Models) # optional
431 device: h200_35gb
432 key: acceptance-length-test-large-models
433 - timeout_in_minutes: 25
433 + timeout_in_minutes: 20
434 gpu: h100
435 optional: true
436 num_gpus: 1
.buildkite/test_areas/model_executor.yaml
+1 -1
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Model Executor
6 key: model-executor
7 - timeout_in_minutes: 35
7 + timeout_in_minutes: 45
8 source_file_dependencies:
9 - vllm/engine/arg_utils.py
10 - vllm/config/model.py
.buildkite/test_areas/model_runner_v2.yaml
+4 -4
@@ -5,7 +5,7 @@ steps:
5 - label: Model Runner V2 Core Tests
6 device: h200_35gb
7 key: model-runner-v2-core-tests
8 - timeout_in_minutes: 45
8 + timeout_in_minutes: 35
9 source_file_dependencies:
10 - vllm/v1/worker/gpu/
11 - vllm/v1/worker/gpu_worker.py
@@ -27,7 +27,7 @@ steps:
27 - label: Model Runner V2 Examples
28 device: h200_35gb
29 key: model-runner-v2-examples
30 - timeout_in_minutes: 45
30 + timeout_in_minutes: 35
31 working_dir: "/vllm-workspace/examples"
32 source_file_dependencies:
33 - vllm/v1/worker/gpu/
@@ -63,7 +63,7 @@ steps:
63
64 - label: Model Runner V2 Distributed (2 GPUs)
65 key: model-runner-v2-distributed-2-gpus
66 - timeout_in_minutes: 45
66 + timeout_in_minutes: 30
67 working_dir: "/vllm-workspace/tests"
68 num_devices: 2
69 source_file_dependencies:
@@ -84,7 +84,7 @@ steps:
84
85 - label: Model Runner V2 Pipeline Parallelism (4 GPUs)
86 key: model-runner-v2-pipeline-parallelism-4-gpus
87 - timeout_in_minutes: 60
87 + timeout_in_minutes: 50
88 working_dir: "/vllm-workspace/tests"
89 num_devices: 4
90 source_file_dependencies:
.buildkite/test_areas/models_basic.yaml
+4 -4
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Basic Models Tests (Initialization)
6 key: basic-models-tests-initialization
7 - timeout_in_minutes: 45
7 + timeout_in_minutes: 25
8 device: h200_18gb
9 source_file_dependencies:
10 - vllm/
@@ -17,7 +17,7 @@ steps:
17 - label: Basic Models Tests (Extra Initialization) %N
18 device: h200_35gb
19 key: basic-models-tests-extra-initialization
20 - timeout_in_minutes: 45
20 + timeout_in_minutes: 100
21 source_file_dependencies:
22 - vllm/model_executor/models/
23 - tests/models/test_initialization.py
@@ -32,7 +32,7 @@ steps:
32 - label: Basic Models Tests (Other)
33 device: h200_35gb
34 key: basic-models-tests-other
35 - timeout_in_minutes: 45
35 + timeout_in_minutes: 35
36 source_file_dependencies:
37 - vllm/
38 - tests/models/test_terratorch.py
@@ -50,7 +50,7 @@ steps:
50 key: basic-models-test-other-cpu
51 depends_on:
52 - image-build-cpu
53 - timeout_in_minutes: 10
53 + timeout_in_minutes: 20
54 source_file_dependencies:
55 - vllm/
56 - tests/models/test_utils.py
.buildkite/test_areas/models_distributed.yaml
+1 -1
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Distributed Model Tests (2 GPUs)
6 key: distributed-model-tests-2-gpus
7 - timeout_in_minutes: 50
7 + timeout_in_minutes: 60
8 working_dir: "/vllm-workspace/tests"
9 num_devices: 2
10 source_file_dependencies:
.buildkite/test_areas/models_language.yaml
+8 -8
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Language Models Tests (Standard)
6 key: language-models-tests-standard
7 - timeout_in_minutes: 25
7 + timeout_in_minutes: 30
8 device: h200_18gb
9 source_file_dependencies:
10 - vllm/
@@ -21,7 +21,7 @@ steps:
21
22 - label: Language Models Tests (Extra Standard) %N
23 key: language-models-tests-extra-standard
24 - timeout_in_minutes: 45
24 + timeout_in_minutes: 40
25 source_file_dependencies:
26 - vllm/model_executor/models/
27 - tests/models/language/pooling/test_embedding.py
@@ -52,7 +52,7 @@ steps:
52
53 - label: Language Models Tests (Hybrid) %N
54 key: language-models-tests-hybrid
55 - timeout_in_minutes: 75
55 + timeout_in_minutes: 65
56 source_file_dependencies:
57 - vllm/
58 - tests/models/language/generation
@@ -67,7 +67,7 @@ steps:
67 mirror:
68 amd:
69 device: mi325_1
70 - timeout_in_minutes: 90
70 + timeout_in_minutes: 70
71 depends_on:
72 - image-build-amd
73 commands:
@@ -78,7 +78,7 @@ steps:
78 - label: Language Models Test (Extended Generation) # 80min
79 device: h200_35gb
80 key: language-models-test-extended-generation
81 - timeout_in_minutes: 110
81 + timeout_in_minutes: 65
82 optional: true
83 source_file_dependencies:
84 - vllm/
@@ -92,7 +92,7 @@ steps:
92
93 - label: Language Models Test (PPL)
94 key: language-models-test-ppl
95 - timeout_in_minutes: 110
95 + timeout_in_minutes: 30
96 device: h200_18gb
97 optional: true
98 source_file_dependencies:
@@ -104,7 +104,7 @@ steps:
104 - label: Language Models Test (Extended Pooling) # 36min
105 device: h200_35gb
106 key: language-models-test-extended-pooling
107 - timeout_in_minutes: 50
107 + timeout_in_minutes: 70
108 optional: true
109 source_file_dependencies:
110 - vllm/
@@ -120,7 +120,7 @@ steps:
120
121 - label: Language Models Test (MTEB)
122 key: language-models-test-mteb
123 - timeout_in_minutes: 110
123 + timeout_in_minutes: 45
124 device: h200_18gb
125 optional: true
126 source_file_dependencies:
.buildkite/test_areas/models_multimodal.yaml
+7 -7
@@ -20,7 +20,7 @@ steps:
20
21 - label: "Multi-Modal Models (Standard) 2: qwen3 + gemma"
22 key: multi-modal-models-standard-2-qwen3-gemma
23 - timeout_in_minutes: 45
23 + timeout_in_minutes: 50
24 device: h200_18gb
25 source_file_dependencies:
26 - vllm/
@@ -38,7 +38,7 @@ steps:
38 - label: "Multi-Modal Models (Standard) 3: llava + qwen2_vl"
39 device: h200_35gb
40 key: multi-modal-models-standard-3-llava-qwen2-vl
41 - timeout_in_minutes: 45
41 + timeout_in_minutes: 40
42 source_file_dependencies:
43 - vllm/
44 - tests/models/multimodal
@@ -54,7 +54,7 @@ steps:
54 - label: "Multi-Modal Models (Standard) 4: other + whisper"
55 device: h200_35gb
56 key: multi-modal-models-standard-4-other-whisper
57 - timeout_in_minutes: 45
57 + timeout_in_minutes: 50
58 source_file_dependencies:
59 - vllm/
60 - tests/models/multimodal
@@ -73,7 +73,7 @@ steps:
73 key: multi-modal-processor-cpu
74 depends_on:
75 - image-build-cpu
76 - timeout_in_minutes: 60
76 + timeout_in_minutes: 125
77 source_file_dependencies:
78 - vllm/
79 - tests/models/multimodal
@@ -84,7 +84,7 @@ steps:
84
85 - label: Multi-Modal Processor # 44min
86 key: multi-modal-processor
87 - timeout_in_minutes: 60
87 + timeout_in_minutes: 65
88 device: h200_18gb
89 source_file_dependencies:
90 - vllm/
@@ -96,7 +96,7 @@ steps:
96 - label: Multi-Modal Accuracy Eval (Small Models) # 50min
97 device: h200_35gb
98 key: multi-modal-accuracy-eval-small-models
99 - timeout_in_minutes: 70
99 + timeout_in_minutes: 30
100 working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
101 source_file_dependencies:
102 - vllm/multimodal/
@@ -164,7 +164,7 @@ steps:
164 mirror:
165 amd:
166 device: mi325_1
167 - timeout_in_minutes: 60
167 + timeout_in_minutes: 75
168 depends_on:
169 - image-build-amd
170 source_file_dependencies:
.buildkite/test_areas/plugins.yaml
+1 -1
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Plugin Tests (2 GPUs)
6 key: plugin-tests-2-gpus
7 - timeout_in_minutes: 60
7 + timeout_in_minutes: 35
8 working_dir: "/vllm-workspace/tests"
9 num_devices: 2
10 source_file_dependencies:
.buildkite/test_areas/pytorch.yaml
+5 -5
@@ -5,7 +5,7 @@ steps:
5 - label: PyTorch Compilation Unit Tests
6 device: h200_35gb
7 key: pytorch-compilation-unit-tests
8 - timeout_in_minutes: 10
8 + timeout_in_minutes: 90
9 source_file_dependencies:
10 - vllm/__init__.py
11 - vllm/_aiter_ops.py
@@ -78,7 +78,7 @@ steps:
78
79 - label: PyTorch Compilation Passes Unit Tests
80 key: pytorch-compilation-passes-unit-tests
81 - timeout_in_minutes: 20
81 + timeout_in_minutes: 45
82 source_file_dependencies:
83 - vllm/__init__.py
84 - vllm/_aiter_ops.py
@@ -110,13 +110,13 @@ steps:
110 mirror:
111 amd:
112 device: mi300_1
113 - timeout_in_minutes: 180
113 + timeout_in_minutes: 65
114 depends_on:
115 - image-build-amd
116
117 - label: PyTorch Fullgraph Smoke Test
118 key: pytorch-fullgraph-smoke-test
119 - timeout_in_minutes: 35
119 + timeout_in_minutes: 60
120 source_file_dependencies:
121 - vllm/__init__.py
122 - vllm/_aiter_ops.py
@@ -152,7 +152,7 @@ steps:
152
153 - label: PyTorch Fullgraph
154 key: pytorch-fullgraph
155 - timeout_in_minutes: 30
155 + timeout_in_minutes: 40
156 device: h200_18gb
157 source_file_dependencies:
158 - vllm/__init__.py
.buildkite/test_areas/quantization.yaml
+4 -4
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Quantization
6 key: quantization
7 - timeout_in_minutes: 90
7 + timeout_in_minutes: 60
8 source_file_dependencies:
9 - csrc/
10 - vllm/model_executor/layers/quantization
@@ -23,7 +23,7 @@ steps:
23
24 - label: Quantized Fusions
25 key: quantized-fusions
26 - timeout_in_minutes: 30
26 + timeout_in_minutes: 20
27 source_file_dependencies:
28 - tests/fusion
29 - vllm/model_executor/layers/fusion
@@ -35,7 +35,7 @@ steps:
35
36 - label: Quantized MoE Test (B200)
37 key: quantized-moe-test-b200
38 - timeout_in_minutes: 60
38 + timeout_in_minutes: 120
39 working_dir: "/vllm-workspace/"
40 device: b200-k8s
41 source_file_dependencies:
@@ -53,7 +53,7 @@ steps:
53
54 - label: Quantized Models Test
55 key: quantized-models-test
56 - timeout_in_minutes: 60
56 + timeout_in_minutes: 50
57 source_file_dependencies:
58 - vllm/model_executor/layers/quantization
59 - tests/models/quantization
.buildkite/test_areas/rust_frontend.yaml
+5 -5
@@ -3,7 +3,7 @@ depends_on:
3 - image-build
4 steps:
5 - label: Rust Frontend OpenAI Coverage
6 - timeout_in_minutes: 90
6 + timeout_in_minutes: 30
7 device: h200_18gb
8 working_dir: "/vllm-workspace/tests"
9 source_file_dependencies:
@@ -38,7 +38,7 @@ steps:
38 - pytest -v -s v1/sample/test_logprobs_e2e.py -k "test_prompt_logprobs_e2e_server"
39
40 - label: Rust Frontend Serve/Admin Coverage
41 - timeout_in_minutes: 60
41 + timeout_in_minutes: 25
42 device: h200_18gb
43 working_dir: "/vllm-workspace/tests"
44 source_file_dependencies:
@@ -67,7 +67,7 @@ steps:
67 - pytest -v -s entrypoints/serve/tokenize/test_tokenization.py -k "not tokenizer_info"
68
69 - label: Rust Frontend Core Correctness
70 - timeout_in_minutes: 30
70 + timeout_in_minutes: 20
71 device: h200_18gb
72 working_dir: "/vllm-workspace/tests"
73 source_file_dependencies:
@@ -81,7 +81,7 @@ steps:
81 - pytest -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
82
83 - label: Rust Frontend Tool Use
84 - timeout_in_minutes: 60
84 + timeout_in_minutes: 25
85 working_dir: "/vllm-workspace/tests"
86 source_file_dependencies:
87 - rust/
@@ -95,7 +95,7 @@ steps:
95 - pytest -v -s tool_use --ignore=tool_use/mistral --models llama3.2 -k "not test_response_format_with_tool_choice_required and not test_parallel_tool_calls_false and not test_tool_call_and_choice"
96
97 - label: Rust Frontend Distributed
98 - timeout_in_minutes: 30
98 + timeout_in_minutes: 25
99 num_devices: 4
100 working_dir: "/vllm-workspace/tests"
101 source_file_dependencies:
.buildkite/test_areas/rust_frontend_cargo.yaml
+2 -2
@@ -4,7 +4,7 @@ steps:
4 - label: Rust Frontend Cargo Style + Clippy
5 key: rust-frontend-cargo-style-clippy
6 depends_on: []
7 - timeout_in_minutes: 30
7 + timeout_in_minutes: 20
8 device: cpu-medium
9 no_plugin: true
10 source_file_dependencies:
@@ -18,7 +18,7 @@ steps:
18 - label: Rust Frontend Cargo Tests
19 key: rust-frontend-cargo-tests
20 depends_on: []
21 - timeout_in_minutes: 30
21 + timeout_in_minutes: 20
22 device: cpu-medium
23 no_plugin: true
24 source_file_dependencies:
.buildkite/test_areas/samplers.yaml
+1 -1
@@ -5,7 +5,7 @@ steps:
5 - label: Samplers Test
6 device: h200_35gb
7 key: samplers-test
8 - timeout_in_minutes: 75
8 + timeout_in_minutes: 40
9 source_file_dependencies:
10 - vllm/model_executor/layers
11 - vllm/sampling_metadata.py
.buildkite/test_areas/spec_decode.yaml
+11 -11
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Spec Decode Eagle
6 key: spec-decode-eagle
7 - timeout_in_minutes: 30
7 + timeout_in_minutes: 25
8 device: h200_18gb
9 source_file_dependencies:
10 - vllm/v1/spec_decode/
@@ -15,7 +15,7 @@ steps:
15 mirror:
16 amd:
17 device: mi325_1
18 - timeout_in_minutes: 45
18 + timeout_in_minutes: 60
19 depends_on:
20 - image-build-amd
21 source_file_dependencies:
@@ -29,7 +29,7 @@ steps:
29
30 - label: Spec Decode Eagle Nightly B200
31 key: spec-decode-eagle-nightly-b200
32 - timeout_in_minutes: 30
32 + timeout_in_minutes: 25
33 device: b200-k8s
34 optional: true
35 source_file_dependencies:
@@ -41,7 +41,7 @@ steps:
41
42 - label: Spec Decode Speculators + MTP
43 key: spec-decode-speculators-mtp
44 - timeout_in_minutes: 30
44 + timeout_in_minutes: 20
45 device: h200_18gb
46 source_file_dependencies:
47 - vllm/v1/spec_decode/
@@ -82,7 +82,7 @@ steps:
82
83 - label: Spec Decode Ngram + Suffix
84 key: spec-decode-ngram-suffix
85 - timeout_in_minutes: 30
85 + timeout_in_minutes: 20
86 device: h200_18gb
87 source_file_dependencies:
88 - vllm/v1/spec_decode/
@@ -93,7 +93,7 @@ steps:
93 mirror:
94 amd:
95 device: mi325_1
96 - timeout_in_minutes: 65
96 + timeout_in_minutes: 55
97 # TODO(akaratza): Test after Torch >= 2.12 bump
98 soft_fail: true
99 depends_on:
@@ -109,7 +109,7 @@ steps:
109
110 - label: Spec Decode Draft Model
111 key: spec-decode-draft-model
112 - timeout_in_minutes: 30
112 + timeout_in_minutes: 45
113 device: h200_18gb
114 source_file_dependencies:
115 - vllm/v1/spec_decode/
@@ -120,7 +120,7 @@ steps:
120 mirror:
121 amd:
122 device: mi325_1
123 - timeout_in_minutes: 50
123 + timeout_in_minutes: 55
124 depends_on:
125 - image-build-amd
126 source_file_dependencies:
@@ -134,7 +134,7 @@ steps:
134
135 - label: Spec Decode Draft Model Nightly B200
136 key: spec-decode-draft-model-nightly-b200
137 - timeout_in_minutes: 30
137 + timeout_in_minutes: 40
138 device: b200-k8s
139 optional: true
140 source_file_dependencies:
@@ -146,7 +146,7 @@ steps:
146
147 - label: Speculators Correctness
148 key: speculators-correctness
149 - timeout_in_minutes: 60
149 + timeout_in_minutes: 30
150 device: h100
151 optional: true
152 num_devices: 1
@@ -159,7 +159,7 @@ steps:
159 - pytest -v -s v1/spec_decode/test_speculators_correctness.py -m slow_test
160
161 - label: Spec Decode MTP hybrid (B200)
162 - timeout_in_minutes: 30
162 + timeout_in_minutes: 20
163 device: b200-k8s
164 optional: true
165 source_file_dependencies:
.buildkite/test_areas/weight_loading.yaml
+1 -1
@@ -4,7 +4,7 @@ depends_on:
4 steps:
5 - label: Weight Loading Multiple GPU # 33min
6 key: weight-loading-multiple-gpu
7 - timeout_in_minutes: 45
7 + timeout_in_minutes: 50
8 working_dir: "/vllm-workspace/tests"
9 num_devices: 2
10 optional: true