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 4, 03:45 UTC
[CI][XPU]Ignore some lora tests from LoRA Intel CI pipeline (#42010)
Signed-off-by: chaojun-zhang <chaojun.zhang@intel.com>
Chaojun Zhang committed
May 8, 2026 at 17:34 UTC
19df11f5d1610e51915d89034e169acb3dfeb7ea
1 file changed
+8
-7
.buildkite/intel_jobs/lora_intel.yaml
+8
-7
@@ -18,6 +18,7 @@ steps:
18
- >-
19
bash .buildkite/scripts/hardware_ci/run-intel-test.sh
20
'cd tests &&
21
+ export VLLM_WORKER_MULTIPROC_METHOD=spawn &&
22
pytest -v -s lora/test_layers.py &&
23
pytest -v -s lora/test_lora_checkpoints.py &&
24
pytest -v -s lora/test_lora_functions.py &&
@@ -46,6 +47,7 @@ steps:
47
- >-
48
bash .buildkite/scripts/hardware_ci/run-intel-test.sh
49
'cd tests &&
50
+ export VLLM_WORKER_MULTIPROC_METHOD=spawn &&
51
pytest -v -s lora/test_fused_moe_lora_kernel.py &&
52
pytest -v -s lora/test_moe_lora_align_sum.py'
53
@@ -65,6 +67,7 @@ steps:
67
- >-
68
bash .buildkite/scripts/hardware_ci/run-intel-test.sh
69
'cd tests &&
70
+ export VLLM_WORKER_MULTIPROC_METHOD=spawn &&
71
set -o pipefail &&
72
pytest -v -s lora/test_punica_ops.py --deselect="tests/lora/test_punica_ops.py::test_kernels_hidden_size[expand-0-xpu:0-dtype0-3-43264-32-4-4]" --deselect="tests/lora/test_punica_ops.py::test_kernels[shrink-0-xpu:0-dtype1-1-2049-64-128-16]" --deselect="tests/lora/test_punica_ops.py::test_kernels[shrink-0-xpu:0-dtype0-1-2049-128-1-32]" --deselect="tests/lora/test_punica_ops.py::test_kernels[shrink-0-xpu:0-dtype0-1-2049-256-1-4]" --deselect="tests/lora/test_punica_ops.py::test_kernels[shrink-0-xpu:0-dtype0-1-2049-256-8-4]" --deselect="tests/lora/test_punica_ops.py::test_kernels[expand-0-xpu:0-dtype0-3-2049-128-8-16]" --deselect="tests/lora/test_punica_ops.py::test_kernels[shrink-0-xpu:0-dtype0-1-2049-128-8-32]" --deselect="tests/lora/test_punica_ops.py::test_kernels[expand-0-xpu:0-dtype1-1-2049-256-128-32]" --deselect="tests/lora/test_punica_ops.py::test_kernels_hidden_size[shrink-0-xpu:0-dtype0-3-64256-32-4-4]" --deselect="tests/lora/test_punica_ops.py::test_kernels_hidden_size[shrink-0-xpu:0-dtype1-2-29696-32-4-4]" --deselect="tests/lora/test_punica_ops.py::test_kernels_hidden_size[shrink-0-xpu:0-dtype1-3-49408-32-4-4]" --deselect="tests/lora/test_punica_ops.py::test_kernels_hidden_size[shrink-0-xpu:0-dtype0-2-16384-32-4-4]" --deselect="tests/lora/test_punica_ops.py::test_kernels_hidden_size[expand-0-xpu:0-dtype0-2-51328-32-4-4]"'
73
@@ -84,6 +87,7 @@ steps:
87
- >-
88
bash .buildkite/scripts/hardware_ci/run-intel-test.sh
89
'cd tests &&
90
+ export VLLM_WORKER_MULTIPROC_METHOD=spawn &&
91
pytest -v -s lora/test_punica_ops_fp8.py &&
92
pytest -v -s lora/test_punica_xpu_ops.py'
93
@@ -103,14 +107,12 @@ steps:
107
- >-
108
bash .buildkite/scripts/hardware_ci/run-intel-test.sh
109
'cd tests &&
110
+ export VLLM_WORKER_MULTIPROC_METHOD=spawn &&
111
(pytest -v -s lora/test_mixtral.py --deselect="tests/lora/test_mixtral.py::test_mixtral_lora[4]" || true) &&
112
pytest -v -s lora/test_quant_model.py --deselect="tests/lora/test_quant_model.py::test_quant_model_lora[model0]" --deselect="tests/lora/test_quant_model.py::test_quant_model_lora[model1]" --deselect="tests/lora/test_quant_model.py::test_quant_model_tp_equality[model0]" &&
108
- (pytest -v -s lora/test_qwen35_densemodel_lora.py || true) &&
113
pytest -v -s lora/test_transformers_model.py &&
114
pytest -v -s lora/test_chatglm3_tp.py &&
111
- pytest -s -v lora/test_llama_tp.py &&
112
- pytest -s -v lora/test_minicpmv_tp.py &&
113
- pytest -s -v lora/test_olmoe_tp.py'
115
+ pytest -s -v lora/test_minicpmv_tp.py'
116
117
- label: LoRA Multimodal
118
timeout_in_minutes: 45
@@ -128,7 +130,6 @@ steps:
130
- >-
131
bash .buildkite/scripts/hardware_ci/run-intel-test.sh
132
'cd tests &&
133
+ export VLLM_WORKER_MULTIPROC_METHOD=spawn &&
134
pytest -v -s lora/test_default_mm_loras.py &&
132
- (pytest -v -s lora/test_qwen3_unembed.py || true) &&
133
- pytest -v -s lora/test_whisper.py &&
134
- (pytest -v -s lora/test_llm_with_multi_loras.py || true)'
135
+ pytest -v -s lora/test_whisper.py'