[Bugfix] Add replacement of _compute_slot_mapping_kernel on CPU (#37987)

Signed-off-by: jiang1.li <jiang1.li@intel.com>
2026-03-24 22:00:20 +08:00
parent 1c0aabdeb0
commit 352b90c4a4
7 changed files with 105 additions and 9 deletions
--- a/vllm/model_executor/layers/quantization/awq_marlin.py
+++ b/vllm/model_executor/layers/quantization/awq_marlin.py
@@ -309,7 +309,7 @@ class AWQMarlinConfig(QuantizationConfig):
        group_size = quant_config.get("group_size")
        zero_point = quant_config.get("zero_point")

-        if not (current_platform.is_cuda_alike() or current_platform.is_cpu()):
+        if not current_platform.is_cuda_alike():
            return False

        if quant_method != "awq":