From 6314de8bad6f28e235652585d63d75bd38d51dab Mon Sep 17 00:00:00 2001 From: zofia <110436990+zufangzhu@users.noreply.github.com> Date: Tue, 2 Jun 2026 22:26:20 +0800 Subject: [PATCH] [XPU] [Bug] remove xpuw4a16 output size check (#44168) Signed-off-by: Zhu, Zufang Co-authored-by: Kunshang Ji --- vllm/model_executor/kernels/linear/mixed_precision/xpu.py | 7 ------- 1 file changed, 7 deletions(-) diff --git a/vllm/model_executor/kernels/linear/mixed_precision/xpu.py b/vllm/model_executor/kernels/linear/mixed_precision/xpu.py index 68528bbd488..17900c75058 100644 --- a/vllm/model_executor/kernels/linear/mixed_precision/xpu.py +++ b/vllm/model_executor/kernels/linear/mixed_precision/xpu.py @@ -51,13 +51,6 @@ class XPUwNa16LinearKernel(MPLinearKernel): "XPUwNa16, supported sizes are multiples of 32", ) - if c.partition_weight_shape[1] % 32 != 0: - return ( - False, - f"Output size ({c.partition_weight_shape[1]}) not supported by " - "XPUWNA16, supported sizes are multiples of 32", - ) - return True, None def process_weights_after_loading(self, layer: torch.nn.Module):