[Perf] Reduce memory usage by splitting tokens in fused_experts and avoiding unused tensor#833

ApsarasX wants to merge 2 commits intovllm-project:mainvllm-project/vllm-ascend:mainfrom

ApsarasX:wengang/memory-optimizationApsarasX/vllm-ascend:wengang/memory-optimizationCopy head branch name to clipboard