[Kernel][MoE] fix computation order of MoE weight multiplication and improve flow#31962

mgoin merged 4 commits intovllm-project:mainvllm-project/vllm:mainfrom

xuebwang-amd:xuebin_triton_fused_moe_mul_routed_weight_issuexuebwang-amd/vllm:xuebin_triton_fused_moe_mul_routed_weight_issueCopy head branch name to clipboard