{"id":62706,"date":"2025-04-12T13:20:28","date_gmt":"2025-04-12T13:20:28","guid":{"rendered":""},"modified":"2025-04-12T13:20:28","modified_gmt":"2025-04-12T13:20:28","slug":"%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706","status":"publish","type":"product","link":"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/","title":{"rendered":"\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633"},"content":{"rendered":"\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
\u0639\u0646\u0648\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u0628\u0647 \u0627\u0646\u06af\u0644\u06cc\u0633\u06cc <\/td>\nP\/D-Serve: Serving Disaggregated Large Language Model at Scale<\/td>\n<\/tr>\n
\u0639\u0646\u0648\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u0628\u0647 \u0641\u0627\u0631\u0633\u06cc <\/td>\n\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633<\/td>\n<\/tr>\n
\u0646\u0648\u06cc\u0633\u0646\u062f\u06af\u0627\u0646 <\/td>\nYibo Jin, Tao Wang, Huimin Lin, Mingyang Song, Peiyang Li, Yipeng Ma, Yicheng Shan, Zhengfan Yuan, Cailong Li, Yajing Sun, Tiandeng Wu, Xing Chu, Ruizhi Huan, Li Ma, Xiao You, Wenting Zhou, Yunpeng Ye, Wen Liu, Xiangkun Xu, Yongsheng Zhang, Tiantian Dong, Jiawei Zhu, Zhe Wang, Xijian Ju, Jianxun Song<\/td>\n<\/tr>\n
\u0641\u0631\u0645\u062a \u0645\u0642\u0627\u0644\u0647 \u0627\u0646\u06af\u0644\u06cc\u0633\u06cc <\/td>\nPDF<\/td>\n<\/tr>\n
\u0632\u0628\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u062a\u062d\u0648\u06cc\u0644\u06cc <\/td>\n\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc<\/td>\n<\/tr>\n
\u0641\u0631\u0645\u062a \u0645\u0642\u0627\u0644\u0647 \u062a\u0631\u062c\u0645\u0647 \u0634\u062f\u0647 <\/td>\n\u0628\u0647 \u0635\u0648\u0631\u062a \u0641\u0627\u06cc\u0644 \u0648\u0631\u062f<\/td>\n<\/tr>\n
\u0646\u062d\u0648\u0647 \u062a\u062d\u0648\u06cc\u0644 \u062a\u0631\u062c\u0645\u0647 <\/td>\n\u062f\u0648 \u062a\u0627 \u0633\u0647 \u0631\u0648\u0632 \u067e\u0633 \u0627\u0632 \u062b\u0628\u062a \u0633\u0641\u0627\u0631\u0634 (\u0628\u0647 \u0635\u0648\u0631\u062a \u0641\u0627\u06cc\u0644 \u062f\u0627\u0646\u0644\u0648\u062f\u06cc)<\/td>\n<\/tr>\n
\u062a\u0639\u062f\u0627\u062f \u0635\u0641\u062d\u0627\u062a<\/td>\n15<\/td>\n<\/tr>\n
\u0644\u06cc\u0646\u06a9 \u062f\u0627\u0646\u0644\u0648\u062f \u0631\u0627\u06cc\u06af\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u0627\u0646\u06af\u0644\u06cc\u0633\u06cc<\/td>\n\u062f\u0627\u0646\u0644\u0648\u062f \u0645\u0642\u0627\u0644\u0647<\/a><\/td>\n<\/tr>\n
\u062f\u0633\u062a\u0647 \u0628\u0646\u062f\u06cc \u0645\u0648\u0636\u0648\u0639\u0627\u062a <\/td>\nDistributed, Parallel, and Cluster Computing,Computation and Language,Machine Learning,\u0645\u062d\u0627\u0633\u0628\u0627\u062a , \u0645\u062d\u0627\u0633\u0628\u0627\u062a \u0648 \u0632\u0628\u0627\u0646 , \u0645\u062d\u0627\u0633\u0628\u0647 \u0648 \u062e\u0648\u0634\u0647 \u0647\u0627\u06cc \u062a\u0648\u0632\u06cc\u0639 \u0634\u062f\u0647 , \u0645\u0648\u0627\u0632\u06cc \u0648 \u062e\u0648\u0634\u0647 \u0627\u06cc ,<\/td>\n<\/tr>\n
\u062a\u0648\u0636\u06cc\u062d\u0627\u062a <\/td>\nSubmitted 15 August, 2024; originally announced August 2024.<\/td>\n<\/tr>\n
\u062a\u0648\u0636\u06cc\u062d\u0627\u062a \u0628\u0647 \u0641\u0627\u0631\u0633\u06cc <\/td>\n\u0627\u0631\u0633\u0627\u0644 \u0634\u062f\u0647 \u062f\u0631 15 \u0627\u0648\u062a 2024 \u061b\u062f\u0631 \u0627\u0628\u062a\u062f\u0627 \u0627\u0648\u062a 2024 \u0627\u0639\u0644\u0627\u0645 \u0634\u062f.<\/td>\n<\/tr>\n
\u0627\u0637\u0644\u0627\u0639\u0627\u062a \u0628\u06cc\u0634\u062a\u0631 \u0627\u0632 \u0627\u06cc\u0646 \u0645\u0642\u0627\u0644\u0647 \u062f\u0631 \u067e\u0627\u06cc\u06af\u0627\u0647 \u0647\u0627\u06cc \u0639\u0644\u0645\u06cc <\/td>\n\n INSPIRE HEP<\/a>
\n
\n NASA ADS<\/a>
\n
\n Google Scholar<\/a>
\n
\n Semantic Scholar<\/a>
\n
\n arXiv<\/a><\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\r\n\r\n \r\n \r\n \r\n \r\n
\u0641\u0631\u0645\u062a \u0627\u0631\u0627\u0626\u0647 \u062a\u0631\u062c\u0645\u0647 \u0645\u0642\u0627\u0644\u0647 <\/td>\r\n \u062a\u062d\u0648\u06cc\u0644 \u0628\u0647 \u0635\u0648\u0631\u062a \u0641\u0627\u06cc\u0644 \u0648\u0631\u062f<\/td>\r\n <\/tr>\r\n
\u0632\u0645\u0627\u0646 \u062a\u062d\u0648\u06cc\u0644 \u062a\u0631\u062c\u0645\u0647 \u0645\u0642\u0627\u0644\u0647 <\/td>\r\n \u0628\u06cc\u0646 2 \u062a\u0627 3 \u0631\u0648\u0632 \u067e\u0633 \u0627\u0632 \u062b\u0628\u062a \u0633\u0641\u0627\u0631\u0634<\/td>\r\n <\/tr>\r\n\t
\u06a9\u06cc\u0641\u06cc\u062a \u062a\u0631\u062c\u0645\u0647 <\/td>\r\n \u0628\u0633\u06cc\u0627\u0631 \u0628\u0627\u0644\u0627. \u0645\u0642\u0627\u0644\u0647 \u0641\u0642\u0637 \u062a\u0648\u0633\u0637 \u0645\u062a\u0631\u062c\u0645\u06cc\u0646 \u0628\u0627 \u0645\u062f\u0631\u06a9 \u062f\u0627\u0646\u0634\u06af\u0627\u0647\u06cc \u0645\u062a\u0631\u062c\u0645\u06cc \u062a\u0631\u062c\u0645\u0647 \u0645\u06cc\u200c\u0634\u0648\u062f.<\/td>\r\n <\/tr>\r\n\t\t
\u062c\u062f\u0627\u0648\u0644 \u0648 \u0641\u0631\u0645\u0648\u0644 \u0647\u0627 <\/td>\r\n \u06a9\u0644\u06cc\u0647 \u062c\u062f\u0627\u0648\u0644 \u0648 \u0641\u0631\u0645\u0648\u0644 \u0647\u0627 \u0646\u06cc\u0632 \u062f\u0631 \u0641\u0627\u06cc\u0644 \u062a\u062d\u0648\u06cc\u0644\u06cc \u0648\u0631\u062f \u062f\u0631\u062c \u0645\u06cc\u200c\u0634\u0648\u0646\u062f.<\/td>\r\n <\/tr>\r\n<\/table>\r\n\r\n\n

\u0686\u06a9\u06cc\u062f\u0647<\/h2>\n

Serving disaggregated large language models (LLMs) over tens of thousands of xPU devices (GPUs or NPUs) with reliable performance faces multiple challenges. 1) Ignoring the diversity (various prefixes and tidal requests), treating all the prompts in a mixed pool is inadequate. To facilitate the similarity per scenario and minimize the inner mismatch on P\/D (prefill and decoding) processing, fine-grained organization is required, dynamically adjusting P\/D ratios for better performance. 2) Due to inaccurate estimation on workload (queue status or maintained connections), the global scheduler easily incurs unnecessary timeouts in prefill. 3) Block-fixed device-to-device (D2D) KVCache transfer over cluster-level RDMA (remote direct memory access) fails to achieve desired D2D utilization as expected. To overcome previous problems, this paper proposes an end-to-end system P\/D-Serve, complying with the paradigm of MLOps (machine learning operations), which models end-to-end (E2E) P\/D performance and enables: 1) fine-grained P\/D organization, mapping the service with RoCE (RDMA over converged ethernet) as needed, to facilitate similar processing and dynamic adjustments on P\/D ratios; 2) on-demand forwarding upon rejections for idle prefill, decoupling the scheduler from regular inaccurate reports and local queues, to avoid timeouts in prefill; and 3) efficient KVCache transfer via optimized D2D access. P\/D-Serve is implemented upon Ascend and MindSpore, has been deployed over tens of thousands of NPUs for more than eight months in commercial use, and further achieves 60\\%, 42\\% and 46\\% improvements on E2E throughput, time-to-first-token (TTFT) SLO (service level objective) and D2D transfer time. As the E2E system with optimizations, P\/D-Serve achieves 6.7x increase on throughput, compared with aggregated LLMs.<\/p>\n

\u0686\u06a9\u06cc\u062f\u0647 \u0628\u0647 \u0641\u0627\u0631\u0633\u06cc (\u062a\u0631\u062c\u0645\u0647 \u0645\u0627\u0634\u06cc\u0646\u06cc)<\/h2>\n

\u062e\u062f\u0645\u062a \u0628\u0647 \u0645\u062f\u0644 \u0647\u0627\u06cc \u0628\u0632\u0631\u06af \u0632\u0628\u0627\u0646 (LLM) \u0628\u06cc\u0634 \u0627\u0632 \u062f\u0647 \u0647\u0627 \u0647\u0632\u0627\u0631 \u062f\u0633\u062a\u06af\u0627\u0647 XPU (GPU \u06cc\u0627 NPU) \u0628\u0627 \u0639\u0645\u0644\u06a9\u0631\u062f \u0642\u0627\u0628\u0644 \u0627\u0639\u062a\u0645\u0627\u062f \u0628\u0627 \u0686\u0627\u0644\u0634 \u0647\u0627\u06cc \u0645\u062e\u062a\u0644\u0641\u06cc \u0631\u0648\u0628\u0631\u0648 \u0627\u0633\u062a.1) \u0646\u0627\u062f\u06cc\u062f\u0647 \u06af\u0631\u0641\u062a\u0646 \u062a\u0646\u0648\u0639 (\u067e\u06cc\u0634\u0648\u0646\u062f\u0647\u0627\u06cc \u0645\u062e\u062a\u0644\u0641 \u0648 \u062f\u0631\u062e\u0648\u0627\u0633\u062a \u0647\u0627\u06cc \u062c\u0632\u0631 \u0648 \u0645\u062f\u06cc) \u060c \u062f\u0631\u0645\u0627\u0646 \u062a\u0645\u0627\u0645 \u0645\u0637\u0627\u0644\u0628 \u0645\u0648\u062c\u0648\u062f \u062f\u0631 \u0627\u0633\u062a\u062e\u0631 \u0645\u062e\u062a\u0644\u0637 \u0646\u0627\u06a9\u0627\u0641\u06cc \u0627\u0633\u062a.\u0628\u0631\u0627\u06cc \u062a\u0633\u0647\u06cc\u0644 \u0634\u0628\u0627\u0647\u062a \u062f\u0631 \u0647\u0631 \u0633\u0646\u0627\u0631\u06cc\u0648 \u0648 \u0628\u0647 \u062d\u062f\u0627\u0642\u0644 \u0631\u0633\u0627\u0646\u062f\u0646 \u0639\u062f\u0645 \u062a\u0637\u0627\u0628\u0642 \u062f\u0627\u062e\u0644\u06cc \u062f\u0631 \u067e\u0631\u062f\u0627\u0632\u0634 P\/D (\u0645\u0642\u062f\u0645\u0647 \u0648 \u0631\u0645\u0632\u06af\u0634\u0627\u06cc\u06cc) \u060c \u0633\u0627\u0632\u0645\u0627\u0646 \u0631\u06cc\u0632 \u062f\u0627\u0646\u0647 \u0645\u0648\u0631\u062f \u0646\u06cc\u0627\u0632 \u0627\u0633\u062a \u060c \u0628\u0647 \u0637\u0648\u0631 \u067e\u0648\u06cc\u0627 \u0646\u0633\u0628\u062a P\/D \u0631\u0627 \u0628\u0631\u0627\u06cc \u0639\u0645\u0644\u06a9\u0631\u062f \u0628\u0647\u062a\u0631 \u062a\u0646\u0638\u06cc\u0645 \u0645\u06cc \u06a9\u0646\u062f.2) \u0628\u0647 \u062f\u0644\u06cc\u0644 \u0628\u0631\u0622\u0648\u0631\u062f \u0646\u0627\u062f\u0631\u0633\u062a \u062f\u0631 \u0645\u0648\u0631\u062f \u0628\u0627\u0631 \u06a9\u0627\u0631 (\u0648\u0636\u0639\u06cc\u062a \u0635\u0641 \u06cc\u0627 \u0627\u062a\u0635\u0627\u0644\u0627\u062a \u062d\u0641\u0638 \u0634\u062f\u0647) \u060c \u0628\u0631\u0646\u0627\u0645\u0647 \u0631\u06cc\u0632 \u062c\u0647\u0627\u0646\u06cc \u0628\u0647 \u0631\u0627\u062d\u062a\u06cc \u0632\u0645\u0627\u0646 \u0647\u0627\u06cc \u063a\u06cc\u0631 \u0636\u0631\u0648\u0631\u06cc \u0631\u0627 \u062f\u0631 \u0645\u0642\u062f\u0645\u0647 \u0645\u062a\u062d\u0645\u0644 \u0645\u06cc \u0634\u0648\u062f.3) \u0627\u0646\u062a\u0642\u0627\u0644 KVCache \u062f\u0633\u062a\u06af\u0627\u0647 \u0628\u0647 \u062f\u0633\u062a\u06af\u0627\u0647 \u0628\u0647 \u062f\u0633\u062a\u06af\u0627\u0647 (D2D) \u0627\u0632 \u0637\u0631\u06cc\u0642 \u0633\u0637\u062d \u062e\u0648\u0634\u0647 RDMA (\u062f\u0633\u062a\u0631\u0633\u06cc \u0628\u0647 \u062d\u0627\u0641\u0638\u0647 \u0645\u0633\u062a\u0642\u06cc\u0645 \u0627\u0632 \u0631\u0627\u0647 \u062f\u0648\u0631) \u062f\u0631 \u062f\u0633\u062a\u06cc\u0627\u0628\u06cc \u0628\u0647 \u0627\u0633\u062a\u0641\u0627\u062f\u0647 \u0627\u0632 D2D \u0645\u0648\u0631\u062f \u0646\u0638\u0631 \u0647\u0645\u0627\u0646\u0637\u0648\u0631 \u06a9\u0647 \u0627\u0646\u062a\u0638\u0627\u0631 \u0645\u06cc \u0631\u0648\u062f \u060c \u0646\u062a\u0648\u0627\u0646\u0633\u062a\u0647 \u0627\u0633\u062a.\u0628\u0631\u0627\u06cc \u063a\u0644\u0628\u0647 \u0628\u0631 \u0645\u0634\u06a9\u0644\u0627\u062a \u0642\u0628\u0644\u06cc \u060c \u0627\u06cc\u0646 \u0645\u0642\u0627\u0644\u0647 \u06cc\u06a9 \u0633\u06cc\u0633\u062a\u0645 P\/D-End-D-End \u0631\u0627 \u0627\u0631\u0627\u0626\u0647 \u0645\u06cc \u062f\u0647\u062f \u060c \u0645\u0637\u0627\u0628\u0642 \u0628\u0627 \u0627\u0644\u06af\u0648\u06cc MLOPS (\u0639\u0645\u0644\u06cc\u0627\u062a \u06cc\u0627\u062f\u06af\u06cc\u0631\u06cc \u0645\u0627\u0634\u06cc\u0646) \u060c \u06a9\u0647 \u0645\u062f\u0644 \u0647\u0627\u06cc \u067e\u0627\u06cc\u0627\u0646 \u062a\u0627 \u0627\u0646\u062a\u0647\u0627 (E2E) P\/D \u0631\u0627 \u0641\u0639\u0627\u0644 \u0645\u06cc \u06a9\u0646\u062f \u0648 \u0627\u0645\u06a9\u0627\u0646 \u067e\u0630\u06cc\u0631 \u0627\u0633\u062a:1) \u0633\u0627\u0632\u0645\u0627\u0646 P\/D \u0631\u06cc\u0632 \u062f\u0627\u0646\u0647 \u060c \u0646\u0642\u0634\u0647 \u0628\u0631\u062f\u0627\u0631\u06cc \u0627\u0632 \u0633\u0631\u0648\u06cc\u0633 \u0628\u0627 ROCE (RDMA \u0628\u06cc\u0634 \u0627\u0632 \u0627\u062a\u0631\u0646\u062a \u0647\u0645\u06af\u0631\u0627) \u062f\u0631 \u0635\u0648\u0631\u062a \u0644\u0632\u0648\u0645 \u060c \u0628\u0631\u0627\u06cc \u062a\u0633\u0647\u06cc\u0644 \u067e\u0631\u062f\u0627\u0632\u0634 \u0645\u0634\u0627\u0628\u0647 \u0648 \u062a\u0646\u0638\u06cc\u0645\u0627\u062a \u067e\u0648\u06cc\u0627 \u062f\u0631 \u0646\u0633\u0628\u062a \u0647\u0627\u06cc P\/D.2) \u062d\u0645\u0644 \u0648 \u0646\u0642\u0644 \u062f\u0631 \u0635\u0648\u0631\u062a \u062a\u0642\u0627\u0636\u0627 \u067e\u0633 \u0627\u0632 \u0631\u062f \u0628\u0631\u0627\u06cc \u0645\u0642\u062f\u0645\u0647 \u0628\u06cc\u06a9\u0627\u0631 \u060c \u062c\u062f\u0627 \u06a9\u0631\u062f\u0646 \u0628\u0631\u0646\u0627\u0645\u0647 \u0631\u06cc\u0632 \u0627\u0632 \u06af\u0632\u0627\u0631\u0634 \u0647\u0627\u06cc \u0646\u0627\u062f\u0631\u0633\u062a \u0648 \u0635\u0641 \u0647\u0627\u06cc \u0645\u062d\u0644\u06cc \u060c \u0628\u0631\u0627\u06cc \u062c\u0644\u0648\u06af\u06cc\u0631\u06cc \u0627\u0632 \u0632\u0645\u0627\u0646 \u0628\u0646\u062f\u06cc \u062f\u0631 \u0645\u0642\u062f\u0645\u0647.\u0648 3) \u0627\u0646\u062a\u0642\u0627\u0644 KVCache \u0627\u0632 \u0637\u0631\u06cc\u0642 \u062f\u0633\u062a\u0631\u0633\u06cc \u0628\u0647\u06cc\u0646\u0647 \u0634\u062f\u0647 D2D.P\/D-Serve \u0628\u0627 Ascend \u0648 MindSpore \u0627\u062c\u0631\u0627 \u0645\u06cc \u0634\u0648\u062f \u060c \u0628\u06cc\u0634 \u0627\u0632 \u0647\u0634\u062a \u0645\u0627\u0647 \u062f\u0631 \u062f\u0647 \u0647\u0627 \u0647\u0632\u0627\u0631 NPU \u062f\u0631 \u0627\u0633\u062a\u0641\u0627\u062f\u0647 \u062a\u062c\u0627\u0631\u06cc \u0645\u0633\u062a\u0642\u0631 \u0634\u062f\u0647 \u0627\u0633\u062a \u0648 \u0628\u06cc\u0634\u062a\u0631 \u0628\u0647 \u067e\u06cc\u0634\u0631\u0641\u062a 60 \\ \u066a \u060c 42 \\ \u066a \u0648 46 \\ \u066a \u062f\u0631 \u0639\u0645\u0644\u06a9\u0631\u062f E2E \u060c \u0632\u0645\u0627\u0646 \u0645\u06cc \u0631\u0633\u062f.-to-first-token (ttft) SLO (\u0647\u062f\u0641 \u0633\u0637\u062d \u062e\u062f\u0645\u0627\u062a) \u0648 \u0632\u0645\u0627\u0646 \u0627\u0646\u062a\u0642\u0627\u0644 D2D.\u0628\u0647 \u0639\u0646\u0648\u0627\u0646 \u0633\u06cc\u0633\u062a\u0645 E2E \u0628\u0627 \u0628\u0647\u06cc\u0646\u0647 \u0633\u0627\u0632\u06cc \u060c P\/D- \u062e\u062f\u0645\u0627\u062a \u062f\u0631 \u0645\u0642\u0627\u06cc\u0633\u0647 \u0628\u0627 LLMS \u062c\u0645\u0639 \u0634\u062f\u0647 \u060c 6.7 \u0628\u0631\u0627\u0628\u0631 \u0627\u0641\u0632\u0627\u06cc\u0634 \u0645\u06cc \u06cc\u0627\u0628\u062f.<\/p>\n\r\n\r\n \r\n \r\n \r\n \r\n
\u0641\u0631\u0645\u062a \u0627\u0631\u0627\u0626\u0647 \u062a\u0631\u062c\u0645\u0647 \u0645\u0642\u0627\u0644\u0647 <\/td>\r\n \u062a\u062d\u0648\u06cc\u0644 \u0628\u0647 \u0635\u0648\u0631\u062a \u0641\u0627\u06cc\u0644 \u0648\u0631\u062f<\/td>\r\n <\/tr>\r\n
\u0632\u0645\u0627\u0646 \u062a\u062d\u0648\u06cc\u0644 \u062a\u0631\u062c\u0645\u0647 \u0645\u0642\u0627\u0644\u0647 <\/td>\r\n \u0628\u06cc\u0646 2 \u062a\u0627 3 \u0631\u0648\u0632 \u067e\u0633 \u0627\u0632 \u062b\u0628\u062a \u0633\u0641\u0627\u0631\u0634<\/td>\r\n <\/tr>\r\n\t
\u06a9\u06cc\u0641\u06cc\u062a \u062a\u0631\u062c\u0645\u0647 <\/td>\r\n \u0628\u0633\u06cc\u0627\u0631 \u0628\u0627\u0644\u0627. \u0645\u0642\u0627\u0644\u0647 \u0641\u0642\u0637 \u062a\u0648\u0633\u0637 \u0645\u062a\u0631\u062c\u0645\u06cc\u0646 \u0628\u0627 \u0645\u062f\u0631\u06a9 \u062f\u0627\u0646\u0634\u06af\u0627\u0647\u06cc \u0645\u062a\u0631\u062c\u0645\u06cc \u062a\u0631\u062c\u0645\u0647 \u0645\u06cc\u200c\u0634\u0648\u062f.<\/td>\r\n <\/tr>\r\n\t\t
\u062c\u062f\u0627\u0648\u0644 \u0648 \u0641\u0631\u0645\u0648\u0644 \u0647\u0627 <\/td>\r\n \u06a9\u0644\u06cc\u0647 \u062c\u062f\u0627\u0648\u0644 \u0648 \u0641\u0631\u0645\u0648\u0644 \u0647\u0627 \u0646\u06cc\u0632 \u062f\u0631 \u0641\u0627\u06cc\u0644 \u062a\u062d\u0648\u06cc\u0644\u06cc \u0648\u0631\u062f \u062f\u0631\u062c \u0645\u06cc\u200c\u0634\u0648\u0646\u062f.<\/td>\r\n <\/tr>\r\n<\/table>\r\n\r\n\n","protected":false},"excerpt":{"rendered":"

\u0639\u0646\u0648\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u0628\u0647 \u0627\u0646\u06af\u0644\u06cc\u0633\u06cc P\/D-Serve: Serving Disaggregated Large Language Model at Scale \u0639\u0646\u0648\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u0628\u0647 \u0641\u0627\u0631\u0633\u06cc \u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: […]<\/p>\n","protected":false},"featured_media":27,"comment_status":"open","ping_status":"open","template":"","meta":{"pmpro_default_level":"","site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-gradient":""}}},"product_cat":[21],"product_tag":[],"class_list":{"0":"post-62706","1":"product","2":"type-product","3":"status-publish","4":"has-post-thumbnail","6":"product_cat-21","7":"pmpro-has-access","8":"desktop-align-left","9":"tablet-align-left","10":"mobile-align-left","12":"first","13":"instock","14":"downloadable","15":"shipping-taxable","16":"purchasable","17":"product-type-simple"},"yoast_head":"\n\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633 - \u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/express24.ir\/d\/product\/\u062a\u0631\u062c\u0645\u0647-\u0641\u0627\u0631\u0633\u06cc-\u0645\u0642\u0627\u0644\u0647-62706\/\" \/>\n<meta property=\"og:locale\" content=\"fa_IR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633 - \u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633\" \/>\n<meta property=\"og:description\" content=\"\u0639\u0646\u0648\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u0628\u0647 \u0627\u0646\u06af\u0644\u06cc\u0633\u06cc P\/D-Serve: Serving Disaggregated Large Language Model at Scale \u0639\u0646\u0648\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u0628\u0647 \u0641\u0627\u0631\u0633\u06cc \u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: […]\" \/>\n<meta property=\"og:url\" content=\"https:\/\/express24.ir\/d\/product\/\u062a\u0631\u062c\u0645\u0647-\u0641\u0627\u0631\u0633\u06cc-\u0645\u0642\u0627\u0644\u0647-62706\/\" \/>\n<meta property=\"og:site_name\" content=\"\u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633\" \/>\n<meta property=\"og:image\" content=\"https:\/\/express24.ir\/d\/wp-content\/uploads\/2024\/02\/Elsevier_logo_2019.svg_.png\" \/>\n\t<meta property=\"og:image:width\" content=\"440\" \/>\n\t<meta property=\"og:image:height\" content=\"486\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/\",\"url\":\"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/\",\"name\":\"\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633 - \u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633\",\"isPartOf\":{\"@id\":\"https:\/\/express24.ir\/d\/#website\"},\"datePublished\":\"2025-04-12T13:20:28+00:00\",\"dateModified\":\"2025-04-12T13:20:28+00:00\",\"breadcrumb\":{\"@id\":\"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/#breadcrumb\"},\"inLanguage\":\"fa-IR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u062e\u0627\u0646\u0647\",\"item\":\"https:\/\/express24.ir\/d\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u0641\u0631\u0648\u0634\u06af\u0627\u0647\",\"item\":\"https:\/\/express24.ir\/d\/\"},{\"@type\":\"ListItem\",\"position\":3,\"name\":\"\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/express24.ir\/d\/#website\",\"url\":\"https:\/\/express24.ir\/d\/\",\"name\":\"\u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633\",\"description\":\"\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/express24.ir\/d\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"fa-IR\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633 - \u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/express24.ir\/d\/product\/\u062a\u0631\u062c\u0645\u0647-\u0641\u0627\u0631\u0633\u06cc-\u0645\u0642\u0627\u0644\u0647-62706\/","og_locale":"fa_IR","og_type":"article","og_title":"\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633 - \u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633","og_description":"\u0639\u0646\u0648\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u0628\u0647 \u0627\u0646\u06af\u0644\u06cc\u0633\u06cc P\/D-Serve: Serving Disaggregated Large Language Model at Scale \u0639\u0646\u0648\u0627\u0646 \u0645\u0642\u0627\u0644\u0647 \u0628\u0647 \u0641\u0627\u0631\u0633\u06cc \u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: […]","og_url":"https:\/\/express24.ir\/d\/product\/\u062a\u0631\u062c\u0645\u0647-\u0641\u0627\u0631\u0633\u06cc-\u0645\u0642\u0627\u0644\u0647-62706\/","og_site_name":"\u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633","og_image":[{"width":440,"height":486,"url":"https:\/\/express24.ir\/d\/wp-content\/uploads\/2024\/02\/Elsevier_logo_2019.svg_.png","type":"image\/png"}],"twitter_card":"summary_large_image","schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/","url":"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/","name":"\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633 - \u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633","isPartOf":{"@id":"https:\/\/express24.ir\/d\/#website"},"datePublished":"2025-04-12T13:20:28+00:00","dateModified":"2025-04-12T13:20:28+00:00","breadcrumb":{"@id":"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/#breadcrumb"},"inLanguage":"fa-IR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/express24.ir\/d\/product\/%d8%aa%d8%b1%d8%ac%d9%85%d9%87-%d9%81%d8%a7%d8%b1%d8%b3%db%8c-%d9%85%d9%82%d8%a7%d9%84%d9%87-62706\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u062e\u0627\u0646\u0647","item":"https:\/\/express24.ir\/d\/"},{"@type":"ListItem","position":2,"name":"\u0641\u0631\u0648\u0634\u06af\u0627\u0647","item":"https:\/\/express24.ir\/d\/"},{"@type":"ListItem","position":3,"name":"\u062a\u0631\u062c\u0645\u0647 \u0641\u0627\u0631\u0633\u06cc \u0645\u0642\u0627\u0644\u0647 P\/D-Serve: \u0627\u0631\u0627\u0626\u0647 \u0645\u062f\u0644 \u0632\u0628\u0627\u0646 \u0628\u0632\u0631\u06af \u062a\u0641\u06a9\u06cc\u06a9 \u0634\u062f\u0647 \u062f\u0631 \u0645\u0642\u06cc\u0627\u0633"}]},{"@type":"WebSite","@id":"https:\/\/express24.ir\/d\/#website","url":"https:\/\/express24.ir\/d\/","name":"\u0641\u0631\u0648\u0634\u06af\u0627\u0647 \u0627\u06a9\u0633\u067e\u0631\u0633","description":"","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/express24.ir\/d\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"fa-IR"}]}},"_links":{"self":[{"href":"https:\/\/express24.ir\/d\/wp-json\/wp\/v2\/product\/62706","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/express24.ir\/d\/wp-json\/wp\/v2\/product"}],"about":[{"href":"https:\/\/express24.ir\/d\/wp-json\/wp\/v2\/types\/product"}],"replies":[{"embeddable":true,"href":"https:\/\/express24.ir\/d\/wp-json\/wp\/v2\/comments?post=62706"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/express24.ir\/d\/wp-json\/wp\/v2\/media\/27"}],"wp:attachment":[{"href":"https:\/\/express24.ir\/d\/wp-json\/wp\/v2\/media?parent=62706"}],"wp:term":[{"taxonomy":"product_cat","embeddable":true,"href":"https:\/\/express24.ir\/d\/wp-json\/wp\/v2\/product_cat?post=62706"},{"taxonomy":"product_tag","embeddable":true,"href":"https:\/\/express24.ir\/d\/wp-json\/wp\/v2\/product_tag?post=62706"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}