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 14, 05:30 UTC
[ci] Adjusting AMD test composition 2025-10-14 (#26852)
Signed-off-by: Alexei V. Ivanov <alexei.ivanov@amd.com>
Alexei-V-Ivanov-AMD committed
Oct 15, 2025 at 18:52 UTC
938c43ea7fa72cf882403b97b3ea3884308a7e0f
1 file changed
+6
-5
.buildkite/test-amd.yaml
+6
-5
@@ -63,7 +63,7 @@ steps:
63
64
- label: Async Engine, Inputs, Utils, Worker Test (CPU) # 4 mins
65
timeout_in_minutes: 10
66
- mirror_hardwares: [amdexperimental]
66
+ mirror_hardwares: [amdexperimental, amdproduction]
67
agent_pool: mi325_1
68
# grade: Blocking
69
source_file_dependencies:
@@ -353,7 +353,7 @@ steps:
353
- pytest -v -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
354
355
- label: V1 Test others (CPU) # 5 mins
356
- mirror_hardwares: [amdexperimental]
356
+ mirror_hardwares: [amdexperimental, amdproduction]
357
agent_pool: mi325_1
358
# grade: Blocking
359
source_file_dependencies:
@@ -459,6 +459,7 @@ steps:
459
- pytest -v -s compile/test_fusion_all_reduce.py
460
- pytest -v -s compile/test_decorator.py
461
- pytest -v -s compile/test_noop_elimination.py
462
+ - pytest -v -s compile/test_aot_compile.py
463
464
- label: PyTorch Fullgraph Smoke Test # 15min
465
timeout_in_minutes: 30
@@ -487,14 +488,14 @@ steps:
488
489
- label: Kernels Core Operation Test # 48min
490
timeout_in_minutes: 75
490
- mirror_hardwares: [amdexperimental]
491
+ mirror_hardwares: [amdexperimental, amdproduction]
492
agent_pool: mi325_1
493
# grade: Blocking
494
source_file_dependencies:
495
- csrc/
496
- tests/kernels/core
497
commands:
497
- - pytest -v -s kernels/core
498
+ - pytest -v -s kernels/core kernels/test_top_k_per_row.py
499
500
- label: Kernels Attention Test %N # 23min
501
timeout_in_minutes: 35
@@ -632,7 +633,7 @@ steps:
633
634
- label: OpenAI-Compatible Tool Use # 23 min
635
timeout_in_minutes: 35
635
- mirror_hardwares: [amdexperimental]
636
+ mirror_hardwares: [amdexperimental, amdproduction]
637
agent_pool: mi325_1
638
# grade: Blocking
639
fast_check: false