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 12, 17:15 UTC
[CI/Build Don't add FLASHINFER backend in test_cpu_offloading.py (#29229)
Signed-off-by: Randall Smith <ransmith@amd.com> Co-authored-by: Randall Smith <ransmith@amd.com>
rasmith committed
Nov 22, 2025 at 05:00 UTC
8e22da1d7fcd43efd8fec18c0c0bf6a8e7cf61a6
1 file changed
+5
-1
tests/v1/kv_offload/test_cpu_offloading.py
+5
-1
@@ -12,10 +12,14 @@ from tqdm import tqdm
12
from vllm import LLM, SamplingParams, TokensPrompt
13
from vllm.config import KVEventsConfig, KVTransferConfig
14
from vllm.distributed.kv_events import BlockStored, KVEventBatch
15
+from vllm.platforms import current_platform
16
from vllm.utils.system_utils import set_env_var
17
18
CPU_BLOCK_SIZES = [48]
18
-ATTN_BACKENDS = ["FLASH_ATTN", "FLASHINFER"]
19
+ATTN_BACKENDS = ["FLASH_ATTN"]
20
+
21
+if current_platform.is_cuda():
22
+ ATTN_BACKENDS.append("FLASHINFER")
23
24
25
class MockSubscriber: