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
Mirroring the test definitions (2025-10-22) (#27362)
Signed-off-by: Alexei V. Ivanov <alexei.ivanov@amd.com>
Alexei-V-Ivanov-AMD committed
Oct 23, 2025 at 11:02 UTC
295c7f0267010832132404760837e8dc4e0e664a
1 file changed
+9
-2
.buildkite/test-amd.yaml
+9
-2
@@ -395,7 +395,9 @@ steps:
395
- python3 offline_inference/basic/embed.py
396
- python3 offline_inference/basic/score.py
397
- python3 offline_inference/spec_decode.py --test --method eagle --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 2048
398
- - python3 offline_inference/spec_decode.py --test --method eagle3 --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 2048
398
+ # https://github.com/vllm-project/vllm/pull/26682 uses slightly more memory in PyTorch 2.9+ causing this test to OOM in 1xL4 GPU
399
+ - python3 offline_inference/spec_decode.py --test --method eagle3 --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 1536
400
+ #- python3 offline_inference/spec_decode.py --test --method eagle3 --num_spec_tokens 3 --dataset-name hf --dataset-path philschmid/mt-bench --num-prompts 80 --temp 0 --top-p 1.0 --top-k -1 --tp 1 --enable-chunked-prefill --max-model-len 2048
401
402
- label: Platform Tests (CUDA) # 4min
403
timeout_in_minutes: 15
@@ -436,7 +438,11 @@ steps:
438
--num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT \
439
--ignore=lora/test_chatglm3_tp.py \
440
--ignore=lora/test_llama_tp.py \
439
- --ignore=lora/test_llm_with_multi_loras.py
441
+ --ignore=lora/test_llm_with_multi_loras.py \
442
+ --ignore=lora/test_olmoe_tp.py \
443
+ --ignore=lora/test_deepseekv2_tp.py \
444
+ --ignore=lora/test_gptoss.py \
445
+ --ignore=lora/test_qwen3moe_tp.py
446
parallelism: 4
447
448
- label: PyTorch Compilation Unit Tests # 15min
@@ -1208,6 +1214,7 @@ steps:
1214
- pytest -v -s -x lora/test_chatglm3_tp.py
1215
- pytest -v -s -x lora/test_llama_tp.py
1216
- pytest -v -s -x lora/test_llm_with_multi_loras.py
1217
+ - pytest -v -s -x lora/test_olmoe_tp.py
1218
1219
- label: Weight Loading Multiple GPU Test # 33min
1220
timeout_in_minutes: 45