[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"tag-推論":3},{"tag":4,"articles":10},{"id":5,"name":6,"slug":6,"article_count":7,"description_zh":8,"description_en":9},"65f138e4-6319-4593-a264-431ca37eb1bc","推論",3,"推論指的是模型在部署後進行即時或批次預測的階段，重點不只在 GPU 算力，也在軟體堆疊、記憶體效率與延遲控制。像 MLPerf 成績、TensorRT-LLM、Dynamo 與伺服器級推論架構，都是這個主題的核心。","Inference is the deployment phase where models generate predictions in real time or in batches. For AI systems, performance depends not only on GPU throughput but also on software stacks, memory efficiency, latency, and serving architecture, from MLPerf results to TensorRT-LLM and server-side optimization.",[11,20,28],{"id":12,"slug":13,"title":14,"summary":15,"category":16,"image_url":17,"cover_image":17,"language":18,"created_at":19},"0b5979a7-dbb3-438f-b8a1-68de0f838df0","nvidia-mlperf-software-inference-benchmarks-zh","Nvidia MLPerf 成績證明軟體還很重要","Nvidia 在 MLPerf v6.0 交出最高 2.77x 推論提升。GB300 NVL72 的成績顯示，Dynamo、TensorRT-LLM 這類軟體優化，已經和 GPU 硬體同樣重要。","research","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775185790112-2r4u.png","zh","2026-04-03T03:09:34.300263+00:00",{"id":21,"slug":22,"title":23,"summary":24,"category":25,"image_url":26,"cover_image":26,"language":18,"created_at":27},"d9fda242-d695-4ea4-a0e0-c6c64ad72965","nvidia-sets-new-mlperf-inference-records-zh","NVIDIA 再刷 MLPerf 推論紀錄","NVIDIA 在 MLPerf Inference v6.0 再交出新成績，GB300 NVL72 對 DeepSeek-R1 伺服器推論提升 2.7x，Llama 3.1 405B 也提升 1.5x。","industry","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775122496881-vxz0.png","2026-04-02T08:48:38.43437+00:00",{"id":29,"slug":30,"title":31,"summary":32,"category":25,"image_url":33,"cover_image":33,"language":18,"created_at":34},"ea6be18b-c903-4e54-97b7-5f7447a612e0","nvidia-gtc-2026-big-ai-announcements-zh","NVIDIA GTC 2026 重點拆解","NVIDIA 在 GTC 2026 一口氣端出 1,000 場 session、2,000 位講者，還把 AI 工廠、推論基礎設施、Agent 平台與實體 AI 全部綁成一套銷售方案。這場大會重點不是單一 GPU，而是從晶片到軟體的整包系統。","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1774516049779-pr7v.png","2026-03-26T07:14:26.62638+00:00"]