Skip to content

GitLab

Unverified Commit 1b82fb0a authored by Kunshang Ji's avatar Kunshang Ji Committed by GitHub
Browse files

[XPU] work around for sp, avoid custom op import error (#28822) 


Signed-off-by: default avatarKunshang Ji <kunshang.ji@intel.com>
parent d4acf518
Hide whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
vllm/compilation/pass_manager.py
View file @ 1b82fb0a
...@@ -18,6 +18,7 @@ if current_platform.is_cuda_alike(): ...@@ -18,6 +18,7 @@ if current_platform.is_cuda_alike():
from .fusion import RMSNormQuantFusionPass from .fusion import RMSNormQuantFusionPass
from .fusion_attn import AttnFusionPass from .fusion_attn import AttnFusionPass
from .qk_norm_rope_fusion import QKNormRoPEFusionPass from .qk_norm_rope_fusion import QKNormRoPEFusionPass
from .sequence_parallelism import SequenceParallelismPass
if current_platform.is_cuda(): if current_platform.is_cuda():
from .collective_fusion import AllReduceFusionPass, AsyncTPPass from .collective_fusion import AllReduceFusionPass, AsyncTPPass
...@@ -25,7 +26,6 @@ if current_platform.is_cuda(): ...@@ -25,7 +26,6 @@ if current_platform.is_cuda():
from .fix_functionalization import FixFunctionalizationPass from .fix_functionalization import FixFunctionalizationPass
from .inductor_pass import CustomGraphPass, InductorPass, get_pass_context from .inductor_pass import CustomGraphPass, InductorPass, get_pass_context
from .noop_elimination import NoOpEliminationPass from .noop_elimination import NoOpEliminationPass
from .sequence_parallelism import SequenceParallelismPass
logger = init_logger(__name__) logger = init_logger(__name__)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment