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, 19:00 UTC
[CI] Migrate 6 verified jobs from gpu_1_queue to h200_18gb MIG (#42446)
Signed-off-by: khluu <khluu000@gmail.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Kevin H. Luu committed
May 12, 2026 at 11:52 UTC
379f0ec369d300d6fb45983882f1a4d548065d73
5 files changed
+5
.buildkite/test_areas/kernels.yaml
+1
@@ -91,6 +91,7 @@ steps:
91
92
- label: Kernels KDA Test
93
timeout_in_minutes: 20
94
+ device: h200_18gb
95
source_file_dependencies:
96
- vllm/model_executor/layers/fla/ops/kda.py
97
- vllm/model_executor/layers/fla/ops/chunk_delta_h.py
.buildkite/test_areas/lm_eval.yaml
+1
@@ -114,6 +114,7 @@ steps:
114
- label: LM Eval TurboQuant KV Cache
115
key: lm-eval-turboquant-kv-cache
116
timeout_in_minutes: 75
117
+ device: h200_18gb
118
source_file_dependencies:
119
- vllm/model_executor/layers/quantization/turboquant/
120
- vllm/v1/attention/backends/turboquant_attn.py
.buildkite/test_areas/misc.yaml
+1
@@ -212,6 +212,7 @@ steps:
212
depends_on: ~
213
optional: true
214
timeout_in_minutes: 20
215
+ device: h200_18gb
216
source_file_dependencies:
217
- tests/standalone_tests/python_only_compile.sh
218
- setup.py
.buildkite/test_areas/models_basic.yaml
+1
@@ -5,6 +5,7 @@ steps:
5
- label: Basic Models Tests (Initialization)
6
key: basic-models-tests-initialization
7
timeout_in_minutes: 45
8
+ device: h200_18gb
9
torch_nightly: true
10
source_file_dependencies:
11
- vllm/
.buildkite/test_areas/models_language.yaml
+1
@@ -5,6 +5,7 @@ steps:
5
- label: Language Models Tests (Standard)
6
key: language-models-tests-standard
7
timeout_in_minutes: 25
8
+ device: h200_18gb
9
source_file_dependencies:
10
- vllm/
11
- tests/models/language