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 3, 08:00 UTC
[ci] Route 28 gpu_1_queue tests to h200_35gb queue (#43030)
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Kevin H. Luu committed
May 18, 2026 at 21:58 UTC
6e889b582b6a0b11f22b3764be174266faa9ff5e
12 files changed
+28
.buildkite/test_areas/engine.yaml
+1
@@ -60,6 +60,7 @@ steps:
60
- image-build-amd
61
62
- label: e2e Core (1 GPU)
63
+ device: h200_35gb
64
key: e2e-core-1-gpu
65
timeout_in_minutes: 30
66
source_file_dependencies:
.buildkite/test_areas/entrypoints.yaml
+2
@@ -88,6 +88,7 @@ steps:
88
- image-build-amd
89
90
- label: Entrypoints Integration (API Server 2)
91
+ device: h200_35gb
92
key: entrypoints-integration-api-server-2
93
timeout_in_minutes: 130
94
working_dir: "/vllm-workspace/tests"
@@ -108,6 +109,7 @@ steps:
109
- image-build-amd
110
111
- label: Entrypoints Integration (Speech to Text)
112
+ device: h200_35gb
113
key: entrypoints-integration-speech_to_text
114
timeout_in_minutes: 50
115
working_dir: "/vllm-workspace/tests"
.buildkite/test_areas/lm_eval.yaml
+2
@@ -3,6 +3,7 @@ depends_on:
3
- image-build
4
steps:
5
- label: LM Eval Small Models
6
+ device: h200_35gb
7
key: lm-eval-small-models
8
timeout_in_minutes: 75
9
source_file_dependencies:
@@ -152,6 +153,7 @@ steps:
153
- pytest -s -v evals/gpt_oss/test_gpqa_correctness.py --config-list-file=configs/models-b200.txt
154
155
- label: MRCR Eval Small Models
156
+ device: h200_35gb
157
timeout_in_minutes: 30
158
source_file_dependencies:
159
- tests/evals/mrcr/
.buildkite/test_areas/lora.yaml
+1
@@ -3,6 +3,7 @@ depends_on:
3
- image-build
4
steps:
5
- label: LoRA %N
6
+ device: h200_35gb
7
key: lora
8
timeout_in_minutes: 30
9
source_file_dependencies:
.buildkite/test_areas/misc.yaml
+4
@@ -3,6 +3,7 @@ depends_on:
3
- image-build
4
steps:
5
- label: V1 Spec Decode
6
+ device: h200_35gb
7
key: v1-spec-decode
8
timeout_in_minutes: 30
9
source_file_dependencies:
@@ -165,6 +166,7 @@ steps:
166
working_dir: "/vllm-workspace/tests" # optional
167
168
- label: Examples
169
+ device: h200_35gb
170
key: examples
171
timeout_in_minutes: 45
172
working_dir: "/vllm-workspace/examples"
@@ -235,6 +237,7 @@ steps:
237
- bash standalone_tests/python_only_compile.sh
238
239
- label: Async Engine, Inputs, Utils, Worker
240
+ device: h200_35gb
241
key: async-engine-inputs-utils-worker
242
timeout_in_minutes: 50
243
source_file_dependencies:
@@ -366,6 +369,7 @@ steps:
369
- pytest -v -s v1/determinism/test_nvfp4_batch_invariant.py
370
371
- label: Acceptance Length Test (Large Models) # optional
372
+ device: h200_35gb
373
key: acceptance-length-test-large-models
374
timeout_in_minutes: 25
375
gpu: h100
.buildkite/test_areas/model_executor.yaml
+1
@@ -3,6 +3,7 @@ depends_on:
3
- image-build
4
steps:
5
- label: Model Executor
6
+ device: h200_35gb
7
key: model-executor
8
timeout_in_minutes: 35
9
source_file_dependencies:
.buildkite/test_areas/model_runner_v2.yaml
+3
@@ -3,6 +3,7 @@ depends_on:
3
- image-build
4
steps:
5
- label: Model Runner V2 Core Tests
6
+ device: h200_35gb
7
key: model-runner-v2-core-tests
8
timeout_in_minutes: 45
9
source_file_dependencies:
@@ -26,6 +27,7 @@ steps:
27
- pytest -v -s entrypoints/llm/test_struct_output_generate.py -k "xgrammar and not speculative_config6 and not speculative_config7 and not speculative_config8 and not speculative_config0"
28
29
- label: Model Runner V2 Examples
30
+ device: h200_35gb
31
key: model-runner-v2-examples
32
timeout_in_minutes: 45
33
working_dir: "/vllm-workspace/examples"
@@ -99,6 +101,7 @@ steps:
101
- pytest -v -s distributed/test_pp_cudagraph.py -k "not ray"
102
103
- label: Model Runner V2 Spec Decode
104
+ device: h200_35gb
105
key: model-runner-v2-spec-decode
106
timeout_in_minutes: 30
107
working_dir: "/vllm-workspace/tests"
.buildkite/test_areas/models_basic.yaml
+4
@@ -18,6 +18,7 @@ steps:
18
torch_nightly: {}
19
20
- label: Basic Models Tests (Extra Initialization) %N
21
+ device: h200_35gb
22
key: basic-models-tests-extra-initialization
23
timeout_in_minutes: 45
24
source_file_dependencies:
@@ -34,6 +35,7 @@ steps:
35
torch_nightly: {}
36
37
- label: Basic Models Tests (Other)
38
+ device: h200_35gb
39
key: basic-models-tests-other
40
timeout_in_minutes: 45
41
source_file_dependencies:
@@ -58,6 +60,7 @@ steps:
60
- pytest -v -s models/test_utils.py models/test_vision.py
61
62
- label: Transformers Nightly Models
63
+ device: h200_35gb
64
key: transformers-nightly-models
65
working_dir: "/vllm-workspace/"
66
optional: true
@@ -74,6 +77,7 @@ steps:
77
- VLLM_WORKER_MULTIPROC_METHOD=spawn python3 examples/generate/multimodal/audio_language_offline.py --model-type whisper
78
79
- label: Transformers Backward Compatibility Models Test
80
+ device: h200_35gb
81
key: transformers-backward-compatibility-models-test
82
working_dir: "/vllm-workspace/"
83
optional: true
.buildkite/test_areas/models_language.yaml
+3
@@ -34,6 +34,7 @@ steps:
34
torch_nightly: {}
35
36
- label: Language Models Tests (Hybrid) %N
37
+ device: h200_35gb
38
key: language-models-tests-hybrid
39
timeout_in_minutes: 75
40
source_file_dependencies:
@@ -59,6 +60,7 @@ steps:
60
- pytest -v -s models/language/generation -m hybrid_model --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --shard-id=$$BUILDKITE_PARALLEL_JOB
61
62
- label: Language Models Test (Extended Generation) # 80min
63
+ device: h200_35gb
64
key: language-models-test-extended-generation
65
timeout_in_minutes: 110
66
optional: true
@@ -84,6 +86,7 @@ steps:
86
- pytest -v -s models/language/generation_ppl_test
87
88
- label: Language Models Test (Extended Pooling) # 36min
89
+ device: h200_35gb
90
key: language-models-test-extended-pooling
91
timeout_in_minutes: 50
92
optional: true
.buildkite/test_areas/models_multimodal.yaml
+5
@@ -38,6 +38,7 @@ steps:
38
- image-build-amd
39
40
- label: "Multi-Modal Models (Standard) 3: llava + qwen2_vl"
41
+ device: h200_35gb
42
key: multi-modal-models-standard-3-llava-qwen2-vl
43
timeout_in_minutes: 45
44
source_file_dependencies:
@@ -54,6 +55,7 @@ steps:
55
- image-build-amd
56
57
- label: "Multi-Modal Models (Standard) 4: other + whisper"
58
+ device: h200_35gb
59
key: multi-modal-models-standard-4-other-whisper
60
timeout_in_minutes: 45
61
source_file_dependencies:
@@ -92,6 +94,7 @@ steps:
94
- pytest -v -s models/multimodal/processing/test_tensor_schema.py
95
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
100
working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
@@ -120,6 +123,7 @@ steps:
123
- image-build-amd
124
125
- label: Multi-Modal Models (Extended Generation 2)
126
+ device: h200_35gb
127
key: multi-modal-models-extended-generation-2
128
optional: true
129
source_file_dependencies:
@@ -130,6 +134,7 @@ steps:
134
- pytest -v -s models/multimodal/generation/test_common.py -m 'split(group=0) and not core_model'
135
136
- label: Multi-Modal Models (Extended Generation 3)
137
+ device: h200_35gb
138
key: multi-modal-models-extended-generation-3
139
optional: true
140
source_file_dependencies:
.buildkite/test_areas/pytorch.yaml
+1
@@ -3,6 +3,7 @@ depends_on:
3
- image-build
4
steps:
5
- label: PyTorch Compilation Unit Tests
6
+ device: h200_35gb
7
key: pytorch-compilation-unit-tests
8
timeout_in_minutes: 10
9
source_file_dependencies:
.buildkite/test_areas/samplers.yaml
+1
@@ -3,6 +3,7 @@ depends_on:
3
- image-build
4
steps:
5
- label: Samplers Test
6
+ device: h200_35gb
7
key: samplers-test
8
timeout_in_minutes: 75
9
source_file_dependencies: