@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 0780bb5783
Fork 0

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, 22:35 UTC

Removing amdproduction Tests (#22027)

Signed-off-by: Alexei V. Ivanov <alexei.ivanov@amd.com>

Alexei-V-Ivanov-AMD committed Jul 31, 2025 at 11:53 UTC 0780bb57835dcd9ee666aaf807c37086de67422b
1 file changed +23 -23
.buildkite/test-pipeline.yaml
+23 -23
@@ -82,7 +82,7 @@ steps:
82 - bash standalone_tests/python_only_compile.sh
83
84 - label: Basic Correctness Test # 30min
85 - mirror_hardwares: [amdexperimental, amdproduction]
85 + mirror_hardwares: [amdexperimental]
86 fast_check: true
87 torch_nightly: true
88 source_file_dependencies:
@@ -99,7 +99,7 @@ steps:
99 - VLLM_TEST_ENABLE_ARTIFICIAL_PREEMPT=1 pytest -v -s basic_correctness/test_preemption.py
100
101 - label: Chunked Prefill Test
102 - mirror_hardwares: [amdexperimental, amdproduction]
102 + mirror_hardwares: [amdexperimental]
103 source_file_dependencies:
104 - vllm/
105 - tests/basic_correctness/test_chunked_prefill
@@ -108,7 +108,7 @@ steps:
108 - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s basic_correctness/test_chunked_prefill.py
109
110 - label: Core Test # 10min
111 - mirror_hardwares: [amdexperimental, amdproduction]
111 + mirror_hardwares: [amdexperimental]
112 fast_check: true
113 source_file_dependencies:
114 - vllm/core
@@ -209,7 +209,7 @@ steps:
209 - pytest -v -s distributed/test_eplb_execute.py
210
211 - label: Metrics, Tracing Test # 10min
212 - mirror_hardwares: [amdexperimental, amdproduction]
212 + mirror_hardwares: [amdexperimental]
213 num_gpus: 2
214 source_file_dependencies:
215 - vllm/
@@ -228,7 +228,7 @@ steps:
228 ##### 1 GPU test #####
229
230 - label: Regression Test # 5min
231 - mirror_hardwares: [amdexperimental, amdproduction]
231 + mirror_hardwares: [amdexperimental]
232 source_file_dependencies:
233 - vllm/
234 - tests/test_regression
@@ -280,7 +280,7 @@ steps:
280 - pytest -v -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine
281
282 - label: Examples Test # 25min
283 - mirror_hardwares: [amdexperimental, amdproduction]
283 + mirror_hardwares: [amdexperimental]
284 working_dir: "/vllm-workspace/examples"
285 source_file_dependencies:
286 - vllm/entrypoints
@@ -305,7 +305,7 @@ steps:
305 - VLLM_USE_V1=0 python3 offline_inference/profiling.py --model facebook/opt-125m run_num_steps --num-steps 2
306
307 - label: Prefix Caching Test # 9min
308 - mirror_hardwares: [amdexperimental, amdproduction]
308 + mirror_hardwares: [amdexperimental]
309 source_file_dependencies:
310 - vllm/
311 - tests/prefix_caching
@@ -314,7 +314,7 @@ steps:
314
315
316 - label: Platform Tests (CUDA)
317 - mirror_hardwares: [amdexperimental, amdproduction]
317 + mirror_hardwares: [amdexperimental]
318 source_file_dependencies:
319 - vllm/
320 - tests/cuda
@@ -355,7 +355,7 @@ steps:
355 - pytest -v -s compile/test_async_tp.py
356
357 - label: PyTorch Fullgraph Smoke Test # 9min
358 - mirror_hardwares: [amdexperimental, amdproduction]
358 + mirror_hardwares: [amdexperimental]
359 torch_nightly: true
360 source_file_dependencies:
361 - vllm/
@@ -368,7 +368,7 @@ steps:
368 - pytest -v -s compile/piecewise/test_full_cudagraph.py
369
370 - label: PyTorch Fullgraph Test # 18min
371 - mirror_hardwares: [amdexperimental, amdproduction]
371 + mirror_hardwares: [amdexperimental]
372 torch_nightly: true
373 source_file_dependencies:
374 - vllm/
@@ -377,7 +377,7 @@ steps:
377 - pytest -v -s compile/test_full_graph.py
378
379 - label: Kernels Core Operation Test
380 - mirror_hardwares: [amdexperimental, amdproduction]
380 + mirror_hardwares: [amdexperimental]
381 source_file_dependencies:
382 - csrc/
383 - tests/kernels/core
@@ -416,7 +416,7 @@ steps:
416 parallelism: 2
417
418 - label: Kernels Mamba Test
419 - mirror_hardwares: [amdexperimental, amdproduction]
419 + mirror_hardwares: [amdexperimental]
420 source_file_dependencies:
421 - csrc/mamba/
422 - tests/kernels/mamba
@@ -424,7 +424,7 @@ steps:
424 - pytest -v -s kernels/mamba
425
426 - label: Tensorizer Test # 11min
427 - mirror_hardwares: [amdexperimental, amdproduction]
427 + mirror_hardwares: [amdexperimental]
428 soft_fail: true
429 source_file_dependencies:
430 - vllm/model_executor/model_loader
@@ -437,7 +437,7 @@ steps:
437 - pytest -v -s entrypoints/openai/test_tensorizer_entrypoint.py
438
439 - label: Model Executor Test
440 - mirror_hardwares: [amdexperimental, amdproduction]
440 + mirror_hardwares: [amdexperimental]
441 source_file_dependencies:
442 - vllm/model_executor
443 - tests/model_executor
@@ -447,7 +447,7 @@ steps:
447 - pytest -v -s model_executor
448
449 - label: Benchmarks # 9min
450 - mirror_hardwares: [amdexperimental, amdproduction]
450 + mirror_hardwares: [amdexperimental]
451 working_dir: "/vllm-workspace/.buildkite"
452 source_file_dependencies:
453 - benchmarks/
@@ -455,7 +455,7 @@ steps:
455 - bash scripts/run-benchmarks.sh
456
457 - label: Benchmarks CLI Test # 10min
458 - mirror_hardwares: [amdexperimental, amdproduction]
458 + mirror_hardwares: [amdexperimental]
459 source_file_dependencies:
460 - vllm/
461 - tests/benchmarks/
@@ -494,7 +494,7 @@ steps:
494 - pytest -s entrypoints/openai/correctness/
495
496 - label: Encoder Decoder tests # 5min
497 - mirror_hardwares: [amdexperimental, amdproduction]
497 + mirror_hardwares: [amdexperimental]
498 source_file_dependencies:
499 - vllm/
500 - tests/encoder_decoder
@@ -502,7 +502,7 @@ steps:
502 - pytest -v -s encoder_decoder
503
504 - label: OpenAI-Compatible Tool Use # 20 min
505 - mirror_hardwares: [amdexperimental, amdproduction]
505 + mirror_hardwares: [amdexperimental]
506 fast_check: false
507 source_file_dependencies:
508 - vllm/
@@ -623,7 +623,7 @@ steps:
623
624 # This test is used only in PR development phase to test individual models and should never run on main
625 - label: Custom Models Test
626 - mirror_hardwares: [amdexperimental, amdproduction]
626 + mirror_hardwares: [amdexperimental]
627 optional: true
628 commands:
629 - echo 'Testing custom models...'
@@ -658,7 +658,7 @@ steps:
658 ##### multi gpus test #####
659
660 - label: Distributed Comm Ops Test # 7min
661 - mirror_hardwares: [amdexperimental, amdproduction]
661 + mirror_hardwares: [amdexperimental]
662 working_dir: "/vllm-workspace/tests"
663 num_gpus: 2
664 source_file_dependencies:
@@ -755,7 +755,7 @@ steps:
755 - pytest -v -s plugins/lora_resolvers # unit tests for in-tree lora resolver plugins
756
757 - label: Multi-step Tests (4 GPUs) # 36min
758 - mirror_hardwares: [amdexperimental, amdproduction]
758 + mirror_hardwares: [amdexperimental]
759 working_dir: "/vllm-workspace/tests"
760 num_gpus: 4
761 source_file_dependencies:
@@ -776,7 +776,7 @@ steps:
776 - pytest -v -s multi_step/test_correctness_llm.py
777
778 - label: Pipeline Parallelism Test # 45min
779 - mirror_hardwares: [amdexperimental, amdproduction]
779 + mirror_hardwares: [amdexperimental]
780 working_dir: "/vllm-workspace/tests"
781 num_gpus: 4
782 source_file_dependencies:
@@ -790,7 +790,7 @@ steps:
790 - pytest -v -s distributed/test_pipeline_parallel.py
791
792 - label: LoRA TP Test (Distributed)
793 - mirror_hardwares: [amdexperimental, amdproduction]
793 + mirror_hardwares: [amdexperimental]
794 num_gpus: 4
795 source_file_dependencies:
796 - vllm/lora