[XPU][CI] Fix ExampleConnector KV cache device selection (#51806)
Signed-off-by: zhenwei-intel <zhenwei.liu@intel.com> Co-authored-by: Kunshang Ji <kunshang.ji@intel.com>
liuzhenwei committed
Aug 11, 2026 at 23:40 UTC
d648236199da3146af2abd236a6e32ce17cbb178
3 files changed
+3
-3
.buildkite/intel_jobs/test-intel.yaml
+1
-1
@@ -105,7 +105,7 @@ steps:
105
pytest -v -s v1/test_serial_utils.py &&
106
VLLM_DISABLE_COMPILE_CACHE=1 pytest -v -s v1/e2e/general/test_correctness_sliding_window.py &&
107
pytest -v -s v1/spec_decode --ignore=v1/spec_decode/test_max_len.py --ignore=v1/spec_decode/test_speculators_eagle3.py --ignore=v1/spec_decode/test_acceptance_length.py --ignore=v1/spec_decode/test_speculators_correctness.py &&
108
- pytest -v -s v1/kv_connector/unit --ignore=v1/kv_connector/unit/test_example_connector.py --ignore=v1/kv_connector/unit/test_lmcache_integration.py --ignore=v1/kv_connector/unit/test_hf3fs_client.py --ignore=v1/kv_connector/unit/test_hf3fs_connector.py --ignore=v1/kv_connector/unit/test_hf3fs_metadata_server.py --ignore=v1/kv_connector/unit/test_offloading_connector.py'
108
+ pytest -v -s v1/kv_connector/unit --ignore=v1/kv_connector/unit/test_lmcache_integration.py --ignore=v1/kv_connector/unit/test_hf3fs_client.py --ignore=v1/kv_connector/unit/test_hf3fs_connector.py --ignore=v1/kv_connector/unit/test_hf3fs_metadata_server.py --ignore=v1/kv_connector/unit/test_offloading_connector.py'
109
- label: "XPU server test"
110
depends_on:
111
- image-build-xpu
tests/v1/kv_connector/unit/test_example_connector.py
+1
-1
@@ -115,7 +115,7 @@ def process_prompt(processor, llm: LLM, question: str, image_urls: list[Image]):
115
["FLASH_ATTN", "TRITON_ATTN"]
116
if current_platform.is_cuda()
117
else ["TRITON_ATTN"]
118
- if current_platform.is_rocm()
118
+ if current_platform.is_rocm() or current_platform.is_xpu()
119
else []
120
),
121
)
vllm/distributed/kv_transfer/kv_connector/v1/example_connector.py
+1
-1
@@ -180,7 +180,7 @@ class ExampleConnector(KVConnectorBase_V1):
180
layer_name, request.token_ids, request.mm_hashes
181
)
182
kv_cache_cpu = safetensors.torch.load_file(filename)["kv_cache"]
183
- kv_cache = kv_cache_cpu.to("cuda", non_blocking=True)
183
+ kv_cache = kv_cache_cpu.to(kv_cache_layer.device, non_blocking=True)
184
if isinstance(attn_metadata, dict):
185
inject_kv_into_layer(
186
kv_cache_layer,