社区
首页
集团介绍
社区
资讯
行情
学堂
TigerAI
登录
注册
007笑
IP属地：广东
+关注
帖子 · 184
帖子 · 184
关注 · 0
关注 · 0
粉丝 · 0
粉丝 · 0
007笑
007笑
·
03-19
$小米集团-W(01810)$
涨到80元就扔掉🫳[开心] [开心] [开心] [鬼脸] [鬼脸] [得意] [得意] [得意] [得意]
看
664
回复
评论
点赞
点赞
编组 21备份 2
分享
举报
007笑
007笑
·
03-03
好👍🏾
加密货币股集体狂飙，Bit Origin盘初一度升逾380%
美国将创建一项加密货币战略储备。
加密货币股集体狂飙，Bit Origin盘初一度升逾380%
看
557
回复
评论
点赞
点赞
编组 21备份 2
分享
举报
007笑
007笑
·
03-01
[开心] [开心] [贱笑]
DeepSeek王炸开源第六弹：全面揭秘V3/R1推理系统秘密，成本利润率高达545%！
DeepSeek V3/R1推理系统通过跨节点专家并行 、计算-通信重叠和精细的负载均衡策略，实现了惊人的性能和效率。本来以为DeepSeek开源周连续五天的开源项目已经结束了，万万没想到DeepSeek还有one more thing ，补了一个王炸开源项目第六弹：深度揭秘DeepSeek V3/R1 推理系统背后的秘密本号第一时间给大家划个重点V3/R1系统设计原则：效率至上！DeepSeek V3/R1 推理系统的核心目标非常明确：更高吞吐量，更低延迟！DeepSeek 团队为了解决 EP 带来的复杂性，可谓是下足了功夫！成本利润率高达 545%！
DeepSeek王炸开源第六弹：全面揭秘V3/R1推理系统秘密，成本利润率高达545%！
看
661
回复
评论
点赞
点赞
编组 21备份 2
分享
举报
007笑
007笑
·
02-26
$小米集团-W(01810)$
好👌！100%
看
807
回复
评论
点赞
点赞
编组 21备份 2
分享
举报
007笑
007笑
·
02-21
$小米集团-W(01810)$
80元，冲
看
803
回复
评论
点赞
点赞
编组 21备份 2
分享
举报
007笑
007笑
·
02-19
$NVDA 20250417 135.0 CALL$
153，冲
看
770
回复
评论
点赞
点赞
编组 21备份 2
分享
举报
007笑
007笑
·
02-14
$小米集团-W(01810)$
继续加油呀嘻嘻
看
1,060
回复
评论
点赞
1
编组 21备份 2
分享
举报
007笑
007笑
·
02-14
$NVDA 20250417 135.0 CALL$
好👌
看
846
回复
评论
点赞
3
编组 21备份 2
分享
举报
007笑
007笑
·
02-13
好，记2035年
美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%
2月12日，美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%，昆腾、Rigetti Computing涨超4%，Quantum Computing涨超3%。
美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%
看
413
回复
评论
点赞
点赞
编组 21备份 2
分享
举报
007笑
007笑
·
02-11
$腾讯音乐(TME)$
真实就好了 ……
看
1,261
回复
评论
点赞
点赞
编组 21备份 2
分享
举报
加载更多
热议股票
{"i18n":{"language":"zh_CN"},"isCurrentUser":false,"userPageInfo":{"id":"3547634688703059","uuid":"3547634688703059","gmtCreate":1584540666394,"gmtModify":1584542529924,"name":"007笑","pinyin":"007x007xiao","introduction":"","introductionEn":"","signature":"","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","hat":null,"hatId":null,"hatName":null,"vip":1,"status":2,"fanSize":17,"headSize":106,"tweetSize":184,"questionSize":0,"limitLevel":999,"accountStatus":4,"level":{"id":3,"name":"书生虎","nameTw":"書生虎","represent":"努力向上","factor":"发布10条非转发主帖,其中5条获得他人回复或点赞","iconColor":"3C9E83","bgColor":"A2F1D9"},"themeCounts":7,"badgeCounts":0,"badges":[],"moderator":false,"superModerator":false,"manageSymbols":null,"badgeLevel":null,"boolIsFan":false,"boolIsHead":false,"favoriteSize":1,"symbols":null,"coverImage":null,"realNameVerified":null,"userBadges":[{"badgeId":"e50ce593bb40487ebfb542ca54f6a561-3","templateUuid":"e50ce593bb40487ebfb542ca54f6a561","name":"偶像虎友","description":"加入老虎社区1500天","bigImgUrl":"https://static.tigerbbs.com/8b40ae7da5bf081a1c84df14bf9e6367","smallImgUrl":"https://static.tigerbbs.com/f160eceddd7c284a8e1136557615cfad","grayImgUrl":"https://static.tigerbbs.com/11792805c468334a9b31c39f95a41c6a","redirectLinkEnabled":0,"redirectLink":null,"hasAllocated":1,"isWearing":0,"stamp":null,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2024.04.27","exceedPercentage":null,"individualDisplayEnabled":0,"backgroundColor":null,"fontColor":null,"individualDisplaySort":0,"categoryType":1001},{"badgeId":"228c86a078844d74991fff2b7ab2428d-3","templateUuid":"228c86a078844d74991fff2b7ab2428d","name":"投资合伙人虎","description":"证券账户累计交易金额达到100万美元","bigImgUrl":"https://static.tigerbbs.com/fbeac6bb240db7da8b972e5183d050ba","smallImgUrl":"https://static.tigerbbs.com/436cdf80292b99f0a992e78750ac4e3a","grayImgUrl":"https://static.tigerbbs.com/506a259a7b456f037592c3b23c779599","redirectLinkEnabled":0,"redirectLink":null,"hasAllocated":1,"isWearing":0,"stamp":null,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2023.07.14","exceedPercentage":"93.83%","individualDisplayEnabled":0,"backgroundColor":null,"fontColor":null,"individualDisplaySort":0,"categoryType":1101},{"badgeId":"35ec162348d5460f88c959321e554969-3","templateUuid":"35ec162348d5460f88c959321e554969","name":"传说交易员","description":"证券或期货账户累计交易次数达到300次","bigImgUrl":"https://static.tigerbbs.com/656db16598a0b8f21429e10d6c1cb033","smallImgUrl":"https://static.tigerbbs.com/03f10910d4dd9234f9b5702a3342193a","grayImgUrl":"https://static.tigerbbs.com/0c767e35268feb729d50d3fa9a386c5a","redirectLinkEnabled":0,"redirectLink":null,"hasAllocated":1,"isWearing":0,"stamp":null,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2022.08.04","exceedPercentage":"93.93%","individualDisplayEnabled":0,"backgroundColor":null,"fontColor":null,"individualDisplaySort":0,"categoryType":1100},{"badgeId":"976c19eed35f4cd78f17501c2e99ef37-1","templateUuid":"976c19eed35f4cd78f17501c2e99ef37","name":"博闻投资者","description":"累计交易超过10只正股","bigImgUrl":"https://static.tigerbbs.com/e74cc24115c4fbae6154ec1b1041bf47","smallImgUrl":"https://static.tigerbbs.com/d48265cbfd97c57f9048db29f22227b0","grayImgUrl":"https://static.tigerbbs.com/76c6d6898b073c77e1c537ebe9ac1c57","redirectLinkEnabled":0,"redirectLink":null,"hasAllocated":1,"isWearing":0,"stamp":null,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2021.12.21","exceedPercentage":null,"individualDisplayEnabled":0,"backgroundColor":null,"fontColor":null,"individualDisplaySort":0,"categoryType":1102},{"badgeId":"518b5610c3e8410da5cfad115e4b0f5a-1","templateUuid":"518b5610c3e8410da5cfad115e4b0f5a","name":"实盘交易者","description":"完成一笔实盘交易","bigImgUrl":"https://static.tigerbbs.com/2e08a1cc2087a1de93402c2c290fa65b","smallImgUrl":"https://static.tigerbbs.com/4504a6397ce1137932d56e5f4ce27166","grayImgUrl":"https://static.tigerbbs.com/4b22c79415b4cd6e3d8ebc4a0fa32604","redirectLinkEnabled":0,"redirectLink":null,"hasAllocated":1,"isWearing":0,"stamp":null,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2021.12.21","exceedPercentage":null,"individualDisplayEnabled":0,"backgroundColor":null,"fontColor":null,"individualDisplaySort":0,"categoryType":1100}],"userBadgeCount":5,"currentWearingBadge":null,"individualDisplayBadges":null,"crmLevel":4,"crmLevelSwitch":0,"location":"广东","starInvestorFollowerNum":0,"starInvestorFlag":false,"starInvestorOrderShareNum":0,"subscribeStarInvestorNum":8,"ror":null,"winRationPercentage":null,"showRor":false,"investmentPhilosophy":null,"starInvestorSubscribeFlag":false},"page":1,"watchlist":null,"tweetList":[{"id":415047395639296,"gmtCreate":1742348716553,"gmtModify":1742349032742,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/01810\">$小米集团-W(01810)$ </a> 涨到80元就扔掉🫳[开心] [开心] [开心] [鬼脸] [鬼脸] [得意] [得意] [得意] [得意] ","listText":"<a href=\"https://laohu8.com/S/01810\">$小米集团-W(01810)$ </a> 涨到80元就扔掉🫳[开心] [开心] [开心] [鬼脸] [鬼脸] [得意] [得意] [得意] [得意] ","text":"$小米集团-W(01810)$ 涨到80元就扔掉🫳[开心] [开心] [开心] [鬼脸] [鬼脸] [得意] [得意] [得意] [得意]","images":[{"img":"https://static.tigerbbs.com/ef9f82f7d511324cdf783d8e897abaf0","width":"972","height":"1631"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/415047395639296","isVote":1,"tweetType":1,"viewCount":664,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":1,"langContent":"CN","totalScore":0},{"id":409612563947736,"gmtCreate":1741013102675,"gmtModify":1741013104904,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"好👍🏾","listText":"好👍🏾","text":"好👍🏾","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/409612563947736","repostId":"1191049478","repostType":2,"repost":{"id":"1191049478","kind":"news","weMediaInfo":{"introduction":"为用户提供金融资讯、行情、数据，旨在帮助投资者理解世界，做投资决策。","home_visible":1,"media_name":"老虎资讯综合","id":"102","head_image":"https://static.tigerbbs.com/8274c5b9d4c2852bfb1c4d6ce16c68ba"},"pubTimestamp":1741012864,"share":"https://www.laohu8.com/m/news/1191049478?lang=&edition=full","pubTime":"2025-03-03 22:41","market":"us","language":"zh","title":"加密货币股集体狂飙，Bit Origin盘初一度升逾380%","url":"https://stock-news.laohu8.com/highlight/detail?id=1191049478","media":"老虎资讯综合","summary":"美国将创建一项加密货币战略储备。","content":"<html><head></head><body><p>周一，Bit Origin盘初一度飙升逾380%，截至发稿涨幅超340%。Strategy涨逾15%，CleanSpark、MARA Holdings涨逾11%。消息面上，特朗普周日在社交媒体上发文称，美国将创建一项加密货币战略储备，其中将包括比特币、以太坊，以及瑞波币、Solana和艾达币。受此消息影响，加密货币在周日纷纷上涨。比特币一度涨至95000美元。</p><p></p><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/5c6daf7b244ab0456c08499a531f70af\" tg-width=\"560\" tg-height=\"240\"/></p><p></p></body></html>","collect":0,"html":"<!DOCTYPE html>\n<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\" />\n<meta name=\"viewport\" content=\"width=device-width,initial-scale=1.0,minimum-scale=1.0,maximum-scale=1.0,user-scalable=no\"/>\n<meta name=\"format-detection\" content=\"telephone=no,email=no,address=no\" />\n<title>加密货币股集体狂飙，Bit Origin盘初一度升逾380%</title>\n<style type=\"text/css\">\na,abbr,acronym,address,applet,article,aside,audio,b,big,blockquote,body,canvas,caption,center,cite,code,dd,del,details,dfn,div,dl,dt,\nem,embed,fieldset,figcaption,figure,footer,form,h1,h2,h3,h4,h5,h6,header,hgroup,html,i,iframe,img,ins,kbd,label,legend,li,mark,menu,nav,\nobject,ol,output,p,pre,q,ruby,s,samp,section,small,span,strike,strong,sub,summary,sup,table,tbody,td,tfoot,th,thead,time,tr,tt,u,ul,var,video{ font:inherit;margin:0;padding:0;vertical-align:baseline;border:0 }\nbody{ font-size:16px; line-height:1.5; color:#999; background:transparent; }\n.wrapper{ overflow:hidden;word-break:break-all;padding:10px; }\nh1,h2{ font-weight:normal; line-height:1.35; margin-bottom:.6em; }\nh3,h4,h5,h6{ line-height:1.35; margin-bottom:1em; }\nh1{ font-size:24px; }\nh2{ font-size:20px; }\nh3{ font-size:18px; }\nh4{ font-size:16px; }\nh5{ font-size:14px; }\nh6{ font-size:12px; }\np,ul,ol,blockquote,dl,table{ margin:1.2em 0; }\nul,ol{ margin-left:2em; }\nul{ list-style:disc; }\nol{ list-style:decimal; }\nli,li p{ margin:10px 0;}\nimg{ max-width:100%;display:block;margin:0 auto 1em; }\nblockquote{ color:#B5B2B1; border-left:3px solid #aaa; padding:1em; }\nstrong,b{font-weight:bold;}\nem,i{font-style:italic;}\ntable{ width:100%;border-collapse:collapse;border-spacing:1px;margin:1em 0;font-size:.9em; }\nth,td{ padding:5px;text-align:left;border:1px solid #aaa; }\nth{ font-weight:bold;background:#5d5d5d; }\n.symbol-link{font-weight:bold;}\n/* header{ border-bottom:1px solid #494756; } */\n.title{ margin:0 0 8px;line-height:1.3;color:#ddd; }\n.meta {color:#5e5c6d;font-size:13px;margin:0 0 .5em; }\na{text-decoration:none; color:#2a4b87;}\n.meta .head { display: inline-block; overflow: hidden}\n.head .h-thumb { width: 30px; height: 30px; margin: 0; padding: 0; border-radius: 50%; float: left;}\n.head .h-content { margin: 0; padding: 0 0 0 9px; float: left;}\n.head .h-name {font-size: 13px; color: #eee; margin: 0;}\n.head .h-time {font-size: 11px; color: #7E829C; margin: 0;line-height: 11px;}\n.small {font-size: 12.5px; display: inline-block; transform: scale(0.9); -webkit-transform: scale(0.9); transform-origin: left; -webkit-transform-origin: left;}\n.smaller {font-size: 12.5px; display: inline-block; transform: scale(0.8); -webkit-transform: scale(0.8); transform-origin: left; -webkit-transform-origin: left;}\n.bt-text {font-size: 12px;margin: 1.5em 0 0 0}\n.bt-text p {margin: 0}\n</style>\n</head>\n<body>\n<div class=\"wrapper\">\n<header>\n<h2 class=\"title\">\n加密货币股集体狂飙，Bit Origin盘初一度升逾380%\n</h2>\n\n<h4 class=\"meta\">\n\n\n<a class=\"head\" href=\"https://laohu8.com/wemedia/102\">\n\n\n<div class=\"h-thumb\" style=\"background-image:url(https://static.tigerbbs.com/8274c5b9d4c2852bfb1c4d6ce16c68ba);background-size:cover;\"></div>\n\n<div class=\"h-content\">\n<p class=\"h-name\">老虎资讯综合 </p>\n<p class=\"h-time\">2025-03-03 22:41</p>\n</div>\n\n</a>\n\n\n</h4>\n\n</header>\n<article>\n<html><head></head><body><p>周一，Bit Origin盘初一度飙升逾380%，截至发稿涨幅超340%。Strategy涨逾15%，CleanSpark、MARA Holdings涨逾11%。消息面上，特朗普周日在社交媒体上发文称，美国将创建一项加密货币战略储备，其中将包括比特币、以太坊，以及瑞波币、Solana和艾达币。受此消息影响，加密货币在周日纷纷上涨。比特币一度涨至95000美元。</p><p></p><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/5c6daf7b244ab0456c08499a531f70af\" tg-width=\"560\" tg-height=\"240\"/></p><p></p></body></html>\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/8a3bd0a69fef53e5d74e0f4d1ca42b6a","relate_stocks":{},"source_url":"","is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"1191049478","content_text":"周一，Bit Origin盘初一度飙升逾380%，截至发稿涨幅超340%。Strategy涨逾15%，CleanSpark、MARA Holdings涨逾11%。消息面上，特朗普周日在社交媒体上发文称，美国将创建一项加密货币战略储备，其中将包括比特币、以太坊，以及瑞波币、Solana和艾达币。受此消息影响，加密货币在周日纷纷上涨。比特币一度涨至95000美元。","news_type":1,"symbols_score_info":{"BTOG":1.1}},"isVote":1,"tweetType":1,"viewCount":557,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":408790378217904,"gmtCreate":1740820297067,"gmtModify":1740820298708,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"[开心] [开心] [贱笑] ","listText":"[开心] [开心] [贱笑] ","text":"[开心] [开心] [贱笑]","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/408790378217904","repostId":"1153894248","repostType":4,"repost":{"id":"1153894248","kind":"news","pubTimestamp":1740819306,"share":"https://www.laohu8.com/m/news/1153894248?lang=&edition=full","pubTime":"2025-03-01 16:55","market":"us","language":"zh","title":"DeepSeek王炸开源第六弹：全面揭秘V3/R1推理系统秘密，成本利润率高达545%！","url":"https://stock-news.laohu8.com/highlight/detail?id=1153894248","media":"AI寒武纪","summary":"DeepSeek V3/R1推理系统通过跨节点专家并行 、计算-通信重叠和精细的负载均衡策略，实现了惊人的性能和效率。本来以为DeepSeek开源周连续五天的开源项目已经结束了，万万没想到DeepSeek还有one more thing ，补了一个王炸开源项目第六弹：深度揭秘DeepSeek V3/R1 推理系统背后的秘密本号第一时间给大家划个重点V3/R1系统设计原则：效率至上！DeepSeek V3/R1 推理系统的核心目标非常明确：更高吞吐量，更低延迟！DeepSeek 团队为了解决 EP 带来的复杂性，可谓是下足了功夫！成本利润率高达 545%！","content":"<html><head></head><body><p>DeepSeek V3/R1推理系统通过跨节点专家并行 (EP)、计算-通信重叠和精细的负载均衡策略，实现了惊人的性能和效率。简单来说，EP就像是“多人协作”，把模型中的“专家”分散到多张 GPU 上进行计算，大幅提升Batch Size，榨干 GPU 算力，同时专家分散，降低内存压力，更快响应。</p><p style=\"text-align: justify;\">本来以为DeepSeek开源周连续五天的开源项目已经结束了，万万没想到DeepSeek还有one more thing ，补了一个王炸开源项目第六弹：深度揭秘DeepSeek V3/R1 推理系统背后的秘密</p><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/c4966df229fb52178fbd42809e12dfac\" tg-width=\"1364\" tg-height=\"1004\"/></p><p style=\"text-align: justify;\">本号第一时间给大家划个重点</p><h2 id=\"id_4159871898\">V3/R1系统设计原则：效率至上！</h2><p style=\"text-align: justify;\">DeepSeek V3/R1 推理系统的核心目标非常明确：更高吞吐量，更低延迟！ 为了实现这两个目标，DeepSeek 团队祭出了一个大招 —— 跨节点专家并行 (Expert Parallelism, EP)！</p><p style=\"text-align: justify;\">专家并行 (EP) 是什么神仙操作？</p><p style=\"text-align: justify;\">简单来说，EP就像是“多人协作”，把模型中的“专家”分散到多张 GPU 上进行计算。这样做有两大好处：</p><ul style=\"\"><li><p>• 大幅提升 Batch Size，榨干 GPU 算力！ ???? 更大的 Batch Size 意味着 GPU 矩阵运算效率更高，推理吞吐量自然水涨船高！</p></li><li><p>• 专家分散，降低内存压力，更快响应！ ⚡️ 每个 GPU 只需处理一小部分专家，减少了内存访问需求，延迟也就降下来啦！</p></li></ul><p style=\"text-align: justify;\">当然，EP 也不是完美无瑕的，它也带来了新的挑战：</p><ol start=\"1\" style=\"\"><li><p>1. 跨节点通信！ ???? 专家分散在不同节点，节点间的通信就成了性能瓶颈。DeepSeek 团队必须精心设计计算流程，让通信和计算“无缝衔接”，最大化效率！</p></li><li><p>2. 多节点数据并行 (DP) + 负载均衡！ ⚖️ EP 本身就涉及多节点，再加上数据并行，负载均衡就显得尤为重要！必须保证所有 GPU 都“吃饱喝足”，避免出现“木桶效应”。</p></li></ol><h2 id=\"id_784007131\">硬核技术揭秘！如何应对 EP 带来的挑战？</h2><p style=\"text-align: justify;\">DeepSeek 团队为了解决 EP 带来的复杂性，可谓是下足了功夫！他们主要从以下几个方面入手：</p><blockquote><p style=\"text-align: justify;\">1.规模化跨节点专家并行 (Large-scale Cross-node Expert Parallelism (EP))</p></blockquote><p style=\"text-align: justify;\">DeepSeek V3/R1 模型参数量巨大，专家数量也相当惊人 (256个专家中只有8个被激活！)。这种高稀疏性决定了模型需要 超大的 Batch Size 才能充分发挥性能。 因此，大规模跨节点 EP 是必然选择！</p><p style=\"text-align: justify;\">为了适应预填充 (prefill) 和解码 (decode) 阶段的不同特点，DeepSeek 采用了不同程度的并行策略：</p><ul style=\"\"><li><p>• 预填充阶段 [Routed Expert EP32, MLA/Shared Expert DP32]： 每个部署单元跨越 4 个节点，拥有 32 个冗余路由专家。每张 GPU 管理 9 个路由专家和 1 个共享专家。</p></li><li><p>• 解码阶段 [Routed Expert EP144, MLA/Shared Expert DP144]： 每个部署单元扩展到 18 个节点，依然是 32 个冗余路由专家。每张 GPU 管理 2 个路由专家和 1 个共享专家。</p></li><li><p>• 计算-通信重叠 (Computation-Communication Overlapping) </p></li></ul><p style=\"text-align: justify;\">大规模跨节点 EP 引入了巨大的通信开销。为了解决这个问题，DeepSeek 采用了 双批次重叠策略 (dual-batch overlap strategy)。 简单来说，就是把一个大的请求 Batch 分成两个 Micro-Batch，交替执行。这样，一个 Micro-Batch 的通信开销就可以巧妙地隐藏在另一个 Micro-Batch 的计算过程中！</p><p style=\"text-align: justify;\">以下是预填充阶段的计算-通信重叠示意图：</p><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/853213d98b58e869d80bbc872e2aa296\" tg-width=\"1346\" tg-height=\"318\"/></p><p style=\"text-align: justify;\">解码阶段也采用了类似的策略，但更加精细，将 Attention 层进一步细分为两步，使用了 五阶段流水线 (5-stage pipeline)，实现更流畅的通信-计算重叠</p><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/fb27cc58ba62d232e7c5024d5b477d3c\" tg-width=\"1336\" tg-height=\"350\"/></p><p style=\"text-align: justify;\">想了解更多计算-通信重叠的细节？ 猛戳这里：https://github.com/deepseek-ai/profile-data (DeepSeek 官方性能分析数据仓库)</p><blockquote><p style=\"text-align: justify;\">2.最优负载均衡 (Optimal Load Balancing) ⚖️</p></blockquote><p style=\"text-align: justify;\">大规模并行 (DP + EP) 带来的另一个挑战就是 负载均衡。 一旦某个 GPU 成为瓶颈，整个系统的性能都会被拖累。为了最大化资源利用率，DeepSeek 团队在负载均衡方面也做了很多优化，主要包括以下三个方面：</p><p style=\"text-align: justify;\">a. 预填充负载均衡器 (Prefill Load Balancer)</p><ul style=\"\"><li><p>• 关键问题： 不同 DP 实例的请求数量和序列长度不同，导致核心注意力计算和分发发送负载不均衡。</p></li><li><p>• 优化目标：<br/>* 平衡 GPU 之间的核心注意力计算负载 (核心注意力计算负载均衡)。<br/>* 均衡每个 GPU 的输入 Token 数量 (分发发送负载均衡)，防止特定 GPU 成为性能瓶颈。</p></li></ul><p style=\"text-align: justify;\">b. 解码负载均衡器 (Decode Load Balancer)</p><ul style=\"\"><li><p>• 关键问题： 不同 DP 实例的请求数量和序列长度不均，导致核心注意力计算 (与 KVCache 使用量相关) 和分发发送负载差异。</p></li><li><p>• 优化目标：<br/>* 平衡 GPU 之间的 KVCache 使用量 (核心注意力计算负载均衡)<br/>* 均衡每个 GPU 的请求数量 (分发发送负载均衡)</p></li></ul><p style=\"text-align: justify;\">c. 专家并行负载均衡器 (Expert-Parallel Load Balancer)</p><ul style=\"\"><li><p>• 关键问题： 对于 MoE 模型，存在一些天生高负载的专家，导致不同 GPU 上的专家计算负载不均衡。</p></li><li><p>• 优化目标：<br/>* 平衡每个 GPU 上的专家计算负载 (即，最小化所有 GPU 的最大分发接收负载)。</p></li></ul><h2 id=\"id_2867931366\">系统架构一览!DeepSeek 在线推理系统概览</h2><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/ffb15a20e3ec96a0590ed6029f1334c1\" tg-width=\"1350\" tg-height=\"630\"/></p><p style=\"text-align: justify;\">整个 DeepSeek 在线推理系统架构清晰明了，各个组件协同工作，保证了高性能和稳定性。</p><h2 id=\"id_460851006\">硬核数据说话!) DeepSeek 在线服务性能统计 </h2><p style=\"text-align: justify;\">DeepSeek V3/R1 推理服务全部部署在 H800 GPU 上，并采用了与训练一致的精度策略：</p><ul style=\"\"><li><p>• 矩阵乘法和分发传输** 使用 FP8 格式</p></li><li><p>• 核心 MLA 计算和组合传输** 使用 BF16格式</p></li></ul><p style=\"text-align: justify;\">保证了最佳服务性能！</p><p style=\"text-align: justify;\">根据统计数据 (UTC+8 02/27/2025 12:00 PM to 02/28/2025 12:00 PM)：</p><p style=\"text-align: justify;\">单 H800 节点平均吞吐量： 预填充阶段约 73.7k tokens/s (输入，包含缓存命中)，解码阶段约14.8k tokens/s (输出)！</p><ul style=\"\"><li><p>• 成本利润率高达 545%！ 这简直逆天！</p></li><li><p></p><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/8eadd7daca6e10675a5053556adaafb4\" tg-width=\"1326\" tg-height=\"474\"/></p></li></ul><p style=\"text-align: justify;\">其他关键数据：</p><ul style=\"\"><li><p>• 总输入 Tokens：608B，其中 342B (56.3%) 命中 On-disk KV 缓存</p></li><li><p>• 总输出 Tokens：168B。</p></li><li><p>• 平均输出速度：20-22 tokens/秒。</p></li><li><p>• 平均每个输出 Token 的 KVCache 长度：4,989 tokens</p></li></ul><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/4e8b21a5c5119e946b8950bf7c994714\" tg-width=\"1350\" tg-height=\"528\"/></p><p style=\"text-align: justify;\">DeepSeek 还根据白天和晚上的服务负载动态调整推理节点数量，实现资源的最优利用！</p><h2 id=\"id_602653476\">写在最后：</h2><p style=\"text-align: justify;\">DeepSeek V3/R1 推理系统通过 跨节点专家并行 (EP)、计算-通信重叠 和 精细的负载均衡策略，实现了惊人的性能和效率！ 这不仅展现了 DeepSeek 强大的技术实力，也为大模型推理系统的优化提供了宝贵的经验</p><p style=\"text-align: justify;\">只能说deepseek太牛了，再让我震惊一会！</p></body></html>","source":"lsy1713840592850","collect":0,"html":"<!DOCTYPE html>\n<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\" />\n<meta name=\"viewport\" content=\"width=device-width,initial-scale=1.0,minimum-scale=1.0,maximum-scale=1.0,user-scalable=no\"/>\n<meta name=\"format-detection\" content=\"telephone=no,email=no,address=no\" />\n<title>DeepSeek王炸开源第六弹：全面揭秘V3/R1推理系统秘密，成本利润率高达545%！</title>\n<style type=\"text/css\">\na,abbr,acronym,address,applet,article,aside,audio,b,big,blockquote,body,canvas,caption,center,cite,code,dd,del,details,dfn,div,dl,dt,\nem,embed,fieldset,figcaption,figure,footer,form,h1,h2,h3,h4,h5,h6,header,hgroup,html,i,iframe,img,ins,kbd,label,legend,li,mark,menu,nav,\nobject,ol,output,p,pre,q,ruby,s,samp,section,small,span,strike,strong,sub,summary,sup,table,tbody,td,tfoot,th,thead,time,tr,tt,u,ul,var,video{ font:inherit;margin:0;padding:0;vertical-align:baseline;border:0 }\nbody{ font-size:16px; line-height:1.5; color:#999; background:transparent; }\n.wrapper{ overflow:hidden;word-break:break-all;padding:10px; }\nh1,h2{ font-weight:normal; line-height:1.35; margin-bottom:.6em; }\nh3,h4,h5,h6{ line-height:1.35; margin-bottom:1em; }\nh1{ font-size:24px; }\nh2{ font-size:20px; }\nh3{ font-size:18px; }\nh4{ font-size:16px; }\nh5{ font-size:14px; }\nh6{ font-size:12px; }\np,ul,ol,blockquote,dl,table{ margin:1.2em 0; }\nul,ol{ margin-left:2em; }\nul{ list-style:disc; }\nol{ list-style:decimal; }\nli,li p{ margin:10px 0;}\nimg{ max-width:100%;display:block;margin:0 auto 1em; }\nblockquote{ color:#B5B2B1; border-left:3px solid #aaa; padding:1em; }\nstrong,b{font-weight:bold;}\nem,i{font-style:italic;}\ntable{ width:100%;border-collapse:collapse;border-spacing:1px;margin:1em 0;font-size:.9em; }\nth,td{ padding:5px;text-align:left;border:1px solid #aaa; }\nth{ font-weight:bold;background:#5d5d5d; }\n.symbol-link{font-weight:bold;}\n/* header{ border-bottom:1px solid #494756; } */\n.title{ margin:0 0 8px;line-height:1.3;color:#ddd; }\n.meta {color:#5e5c6d;font-size:13px;margin:0 0 .5em; }\na{text-decoration:none; color:#2a4b87;}\n.meta .head { display: inline-block; overflow: hidden}\n.head .h-thumb { width: 30px; height: 30px; margin: 0; padding: 0; border-radius: 50%; float: left;}\n.head .h-content { margin: 0; padding: 0 0 0 9px; float: left;}\n.head .h-name {font-size: 13px; color: #eee; margin: 0;}\n.head .h-time {font-size: 11px; color: #7E829C; margin: 0;line-height: 11px;}\n.small {font-size: 12.5px; display: inline-block; transform: scale(0.9); -webkit-transform: scale(0.9); transform-origin: left; -webkit-transform-origin: left;}\n.smaller {font-size: 12.5px; display: inline-block; transform: scale(0.8); -webkit-transform: scale(0.8); transform-origin: left; -webkit-transform-origin: left;}\n.bt-text {font-size: 12px;margin: 1.5em 0 0 0}\n.bt-text p {margin: 0}\n</style>\n</head>\n<body>\n<div class=\"wrapper\">\n<header>\n<h2 class=\"title\">\nDeepSeek王炸开源第六弹：全面揭秘V3/R1推理系统秘密，成本利润率高达545%！\n</h2>\n\n<h4 class=\"meta\">\n\n\n2025-03-01 16:55 北京时间 <a href=https://mp.weixin.qq.com/s?__biz=Mzg3MTkxMjYzOA==&mid=2247502603&idx=1&sn=ca83bb0148d854b6a2266021cf26606d&chksm=cfc556cbed868bd48140d0090fda548537400bbcdbf5a3e42d6d3bb3f7d8a2338c5f2c56469d&mpshare=1&scene=23&srcid=0301o7sl2uvTeZvwyDZg6NU0&sharer_shareinfo=af5706b0d11227263611c9616a4ce25c&sharer_shareinfo_first=af5706b0d11227263611c9616a4ce25c#rd><strong>AI寒武纪</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>DeepSeek V3/R1推理系统通过跨节点专家并行 (EP)、计算-通信重叠和精细的负载均衡策略，实现了惊人的性能和效率。简单来说，EP就像是“多人协作”，把模型中的“专家”分散到多张 GPU 上进行计算，大幅提升Batch Size，榨干 GPU 算力，同时专家分散，降低内存压力，更快响应。本来以为DeepSeek开源周连续五天的开源项目已经结束了，万万没想到DeepSeek还有one ...</p>\n\n<a href=\"https://mp.weixin.qq.com/s?__biz=Mzg3MTkxMjYzOA==&mid=2247502603&idx=1&sn=ca83bb0148d854b6a2266021cf26606d&chksm=cfc556cbed868bd48140d0090fda548537400bbcdbf5a3e42d6d3bb3f7d8a2338c5f2c56469d&mpshare=1&scene=23&srcid=0301o7sl2uvTeZvwyDZg6NU0&sharer_shareinfo=af5706b0d11227263611c9616a4ce25c&sharer_shareinfo_first=af5706b0d11227263611c9616a4ce25c#rd\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/6912f480d0a5aead184bbc80e00929ae","relate_stocks":{},"source_url":"https://mp.weixin.qq.com/s?__biz=Mzg3MTkxMjYzOA==&mid=2247502603&idx=1&sn=ca83bb0148d854b6a2266021cf26606d&chksm=cfc556cbed868bd48140d0090fda548537400bbcdbf5a3e42d6d3bb3f7d8a2338c5f2c56469d&mpshare=1&scene=23&srcid=0301o7sl2uvTeZvwyDZg6NU0&sharer_shareinfo=af5706b0d11227263611c9616a4ce25c&sharer_shareinfo_first=af5706b0d11227263611c9616a4ce25c#rd","is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"1153894248","content_text":"DeepSeek V3/R1推理系统通过跨节点专家并行 (EP)、计算-通信重叠和精细的负载均衡策略，实现了惊人的性能和效率。简单来说，EP就像是“多人协作”，把模型中的“专家”分散到多张 GPU 上进行计算，大幅提升Batch Size，榨干 GPU 算力，同时专家分散，降低内存压力，更快响应。本来以为DeepSeek开源周连续五天的开源项目已经结束了，万万没想到DeepSeek还有one more thing ，补了一个王炸开源项目第六弹：深度揭秘DeepSeek V3/R1 推理系统背后的秘密本号第一时间给大家划个重点V3/R1系统设计原则：效率至上！DeepSeek V3/R1 推理系统的核心目标非常明确：更高吞吐量，更低延迟！ 为了实现这两个目标，DeepSeek 团队祭出了一个大招 —— 跨节点专家并行 (Expert Parallelism, EP)！专家并行 (EP) 是什么神仙操作？简单来说，EP就像是“多人协作”，把模型中的“专家”分散到多张 GPU 上进行计算。这样做有两大好处：• 大幅提升 Batch Size，榨干 GPU 算力！ ???? 更大的 Batch Size 意味着 GPU 矩阵运算效率更高，推理吞吐量自然水涨船高！• 专家分散，降低内存压力，更快响应！ ⚡️ 每个 GPU 只需处理一小部分专家，减少了内存访问需求，延迟也就降下来啦！当然，EP 也不是完美无瑕的，它也带来了新的挑战：1. 跨节点通信！ ???? 专家分散在不同节点，节点间的通信就成了性能瓶颈。DeepSeek 团队必须精心设计计算流程，让通信和计算“无缝衔接”，最大化效率！2. 多节点数据并行 (DP) + 负载均衡！ ⚖️ EP 本身就涉及多节点，再加上数据并行，负载均衡就显得尤为重要！必须保证所有 GPU 都“吃饱喝足”，避免出现“木桶效应”。硬核技术揭秘！如何应对 EP 带来的挑战？DeepSeek 团队为了解决 EP 带来的复杂性，可谓是下足了功夫！他们主要从以下几个方面入手：1.规模化跨节点专家并行 (Large-scale Cross-node Expert Parallelism (EP))DeepSeek V3/R1 模型参数量巨大，专家数量也相当惊人 (256个专家中只有8个被激活！)。这种高稀疏性决定了模型需要 超大的 Batch Size 才能充分发挥性能。 因此，大规模跨节点 EP 是必然选择！为了适应预填充 (prefill) 和解码 (decode) 阶段的不同特点，DeepSeek 采用了不同程度的并行策略：• 预填充阶段 [Routed Expert EP32, MLA/Shared Expert DP32]： 每个部署单元跨越 4 个节点，拥有 32 个冗余路由专家。每张 GPU 管理 9 个路由专家和 1 个共享专家。• 解码阶段 [Routed Expert EP144, MLA/Shared Expert DP144]： 每个部署单元扩展到 18 个节点，依然是 32 个冗余路由专家。每张 GPU 管理 2 个路由专家和 1 个共享专家。• 计算-通信重叠 (Computation-Communication Overlapping) 大规模跨节点 EP 引入了巨大的通信开销。为了解决这个问题，DeepSeek 采用了 双批次重叠策略 (dual-batch overlap strategy)。 简单来说，就是把一个大的请求 Batch 分成两个 Micro-Batch，交替执行。这样，一个 Micro-Batch 的通信开销就可以巧妙地隐藏在另一个 Micro-Batch 的计算过程中！以下是预填充阶段的计算-通信重叠示意图：解码阶段也采用了类似的策略，但更加精细，将 Attention 层进一步细分为两步，使用了 五阶段流水线 (5-stage pipeline)，实现更流畅的通信-计算重叠想了解更多计算-通信重叠的细节？ 猛戳这里：https://github.com/deepseek-ai/profile-data (DeepSeek 官方性能分析数据仓库)2.最优负载均衡 (Optimal Load Balancing) ⚖️大规模并行 (DP + EP) 带来的另一个挑战就是 负载均衡。 一旦某个 GPU 成为瓶颈，整个系统的性能都会被拖累。为了最大化资源利用率，DeepSeek 团队在负载均衡方面也做了很多优化，主要包括以下三个方面：a. 预填充负载均衡器 (Prefill Load Balancer)• 关键问题： 不同 DP 实例的请求数量和序列长度不同，导致核心注意力计算和分发发送负载不均衡。• 优化目标：* 平衡 GPU 之间的核心注意力计算负载 (核心注意力计算负载均衡)。* 均衡每个 GPU 的输入 Token 数量 (分发发送负载均衡)，防止特定 GPU 成为性能瓶颈。b. 解码负载均衡器 (Decode Load Balancer)• 关键问题： 不同 DP 实例的请求数量和序列长度不均，导致核心注意力计算 (与 KVCache 使用量相关) 和分发发送负载差异。• 优化目标：* 平衡 GPU 之间的 KVCache 使用量 (核心注意力计算负载均衡)* 均衡每个 GPU 的请求数量 (分发发送负载均衡)c. 专家并行负载均衡器 (Expert-Parallel Load Balancer)• 关键问题： 对于 MoE 模型，存在一些天生高负载的专家，导致不同 GPU 上的专家计算负载不均衡。• 优化目标：* 平衡每个 GPU 上的专家计算负载 (即，最小化所有 GPU 的最大分发接收负载)。系统架构一览!DeepSeek 在线推理系统概览整个 DeepSeek 在线推理系统架构清晰明了，各个组件协同工作，保证了高性能和稳定性。硬核数据说话!) DeepSeek 在线服务性能统计 DeepSeek V3/R1 推理服务全部部署在 H800 GPU 上，并采用了与训练一致的精度策略：• 矩阵乘法和分发传输** 使用 FP8 格式• 核心 MLA 计算和组合传输** 使用 BF16格式保证了最佳服务性能！根据统计数据 (UTC+8 02/27/2025 12:00 PM to 02/28/2025 12:00 PM)：单 H800 节点平均吞吐量： 预填充阶段约 73.7k tokens/s (输入，包含缓存命中)，解码阶段约14.8k tokens/s (输出)！• 成本利润率高达 545%！ 这简直逆天！其他关键数据：• 总输入 Tokens：608B，其中 342B (56.3%) 命中 On-disk KV 缓存• 总输出 Tokens：168B。• 平均输出速度：20-22 tokens/秒。• 平均每个输出 Token 的 KVCache 长度：4,989 tokensDeepSeek 还根据白天和晚上的服务负载动态调整推理节点数量，实现资源的最优利用！写在最后：DeepSeek V3/R1 推理系统通过 跨节点专家并行 (EP)、计算-通信重叠 和 精细的负载均衡策略，实现了惊人的性能和效率！ 这不仅展现了 DeepSeek 强大的技术实力，也为大模型推理系统的优化提供了宝贵的经验只能说deepseek太牛了，再让我震惊一会！","news_type":1,"symbols_score_info":{}},"isVote":1,"tweetType":1,"viewCount":661,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"EN","totalScore":0},{"id":407620479033472,"gmtCreate":1740535650702,"gmtModify":1740540163671,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/01810\">$小米集团-W(01810)$ </a> 好👌！100%","listText":"<a href=\"https://laohu8.com/S/01810\">$小米集团-W(01810)$ </a> 好👌！100%","text":"$小米集团-W(01810)$ 好👌！100%","images":[{"img":"https://static.tigerbbs.com/8878d2994a1a5b19c6fb2e9a16f39137","width":"972","height":"1631"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/407620479033472","isVote":1,"tweetType":1,"viewCount":807,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":1,"langContent":"CN","totalScore":0},{"id":405860679189048,"gmtCreate":1740113821155,"gmtModify":1740114888620,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/01810\">$小米集团-W(01810)$ </a> 80元，冲","listText":"<a href=\"https://laohu8.com/S/01810\">$小米集团-W(01810)$ </a> 80元，冲","text":"$小米集团-W(01810)$ 80元，冲","images":[{"img":"https://static.tigerbbs.com/b6d46501aa78e562bab60ea19e567cab","width":"972","height":"1631"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/405860679189048","isVote":1,"tweetType":1,"viewCount":803,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":1,"langContent":"CN","totalScore":0},{"id":404977006686424,"gmtCreate":1739895515544,"gmtModify":1739896005010,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/OPT/NVDA 20250417 135.0 CALL\">$NVDA 20250417 135.0 CALL$</a> 153，冲","listText":"<a href=\"https://laohu8.com/OPT/NVDA 20250417 135.0 CALL\">$NVDA 20250417 135.0 CALL$</a> 153，冲","text":"$NVDA 20250417 135.0 CALL$ 153，冲","images":[{"img":"https://static.tigerbbs.com/d40f8e1e3c77d8d8cc39eb516f2f347a","width":"972","height":"1631"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/404977006686424","isVote":1,"tweetType":1,"viewCount":770,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":1,"langContent":"CN","totalScore":0},{"id":403545089356424,"gmtCreate":1739546783814,"gmtModify":1739546862899,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/01810\">$小米集团-W(01810)$ </a><v-v data-views=\"1\"></v-v> 继续加油呀嘻嘻","listText":"<a href=\"https://laohu8.com/S/01810\">$小米集团-W(01810)$ </a><v-v data-views=\"1\"></v-v> 继续加油呀嘻嘻","text":"$小米集团-W(01810)$ 继续加油呀嘻嘻","images":[{"img":"https://static.tigerbbs.com/61cc548a5c016c7cdb2124ad29bd3554","width":"972","height":"1631"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":1,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/403545089356424","isVote":1,"tweetType":1,"viewCount":1060,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":1,"langContent":"CN","totalScore":0},{"id":403544938947208,"gmtCreate":1739546747093,"gmtModify":1739546859915,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/OPT/NVDA 20250417 135.0 CALL\">$NVDA 20250417 135.0 CALL$</a> 好👌","listText":"<a href=\"https://laohu8.com/OPT/NVDA 20250417 135.0 CALL\">$NVDA 20250417 135.0 CALL$</a> 好👌","text":"$NVDA 20250417 135.0 CALL$ 好👌","images":[{"img":"https://static.tigerbbs.com/07be8fc497bc74579bc8850b7434d6f3","width":"972","height":"1631"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":3,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/403544938947208","isVote":1,"tweetType":1,"viewCount":846,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":1,"langContent":"CN","totalScore":0},{"id":402885734613400,"gmtCreate":1739379460747,"gmtModify":1739379462453,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"好，记2035年","listText":"好，记2035年","text":"好，记2035年","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/402885734613400","repostId":"1142674620","repostType":4,"repost":{"id":"1142674620","kind":"news","weMediaInfo":{"introduction":"为用户提供金融资讯、行情、数据，旨在帮助投资者理解世界，做投资决策。","home_visible":1,"media_name":"老虎资讯综合","id":"102","head_image":"https://static.tigerbbs.com/8274c5b9d4c2852bfb1c4d6ce16c68ba"},"pubTimestamp":1739374704,"share":"https://www.laohu8.com/m/news/1142674620?lang=&edition=full","pubTime":"2025-02-12 23:38","market":"us","language":"zh","title":"美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%","url":"https://stock-news.laohu8.com/highlight/detail?id=1142674620","media":"老虎资讯综合","summary":"2月12日，美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%，昆腾、Rigetti Computing涨超4%，Quantum Computing涨超3%。","content":"<html><head></head><body><p>2月12日，美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%，昆腾、Rigetti Computing涨超4%，Quantum Computing涨超3%。</p><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/6495c6a7711d083d65eb23830b3724e2\" tg-width=\"408\" tg-height=\"589\"/></p></body></html>","collect":0,"html":"<!DOCTYPE html>\n<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset=utf-8\" />\n<meta name=\"viewport\" content=\"width=device-width,initial-scale=1.0,minimum-scale=1.0,maximum-scale=1.0,user-scalable=no\"/>\n<meta name=\"format-detection\" content=\"telephone=no,email=no,address=no\" />\n<title>美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%</title>\n<style type=\"text/css\">\na,abbr,acronym,address,applet,article,aside,audio,b,big,blockquote,body,canvas,caption,center,cite,code,dd,del,details,dfn,div,dl,dt,\nem,embed,fieldset,figcaption,figure,footer,form,h1,h2,h3,h4,h5,h6,header,hgroup,html,i,iframe,img,ins,kbd,label,legend,li,mark,menu,nav,\nobject,ol,output,p,pre,q,ruby,s,samp,section,small,span,strike,strong,sub,summary,sup,table,tbody,td,tfoot,th,thead,time,tr,tt,u,ul,var,video{ font:inherit;margin:0;padding:0;vertical-align:baseline;border:0 }\nbody{ font-size:16px; line-height:1.5; color:#999; background:transparent; }\n.wrapper{ overflow:hidden;word-break:break-all;padding:10px; }\nh1,h2{ font-weight:normal; line-height:1.35; margin-bottom:.6em; }\nh3,h4,h5,h6{ line-height:1.35; margin-bottom:1em; }\nh1{ font-size:24px; }\nh2{ font-size:20px; }\nh3{ font-size:18px; }\nh4{ font-size:16px; }\nh5{ font-size:14px; }\nh6{ font-size:12px; }\np,ul,ol,blockquote,dl,table{ margin:1.2em 0; }\nul,ol{ margin-left:2em; }\nul{ list-style:disc; }\nol{ list-style:decimal; }\nli,li p{ margin:10px 0;}\nimg{ max-width:100%;display:block;margin:0 auto 1em; }\nblockquote{ color:#B5B2B1; border-left:3px solid #aaa; padding:1em; }\nstrong,b{font-weight:bold;}\nem,i{font-style:italic;}\ntable{ width:100%;border-collapse:collapse;border-spacing:1px;margin:1em 0;font-size:.9em; }\nth,td{ padding:5px;text-align:left;border:1px solid #aaa; }\nth{ font-weight:bold;background:#5d5d5d; }\n.symbol-link{font-weight:bold;}\n/* header{ border-bottom:1px solid #494756; } */\n.title{ margin:0 0 8px;line-height:1.3;color:#ddd; }\n.meta {color:#5e5c6d;font-size:13px;margin:0 0 .5em; }\na{text-decoration:none; color:#2a4b87;}\n.meta .head { display: inline-block; overflow: hidden}\n.head .h-thumb { width: 30px; height: 30px; margin: 0; padding: 0; border-radius: 50%; float: left;}\n.head .h-content { margin: 0; padding: 0 0 0 9px; float: left;}\n.head .h-name {font-size: 13px; color: #eee; margin: 0;}\n.head .h-time {font-size: 11px; color: #7E829C; margin: 0;line-height: 11px;}\n.small {font-size: 12.5px; display: inline-block; transform: scale(0.9); -webkit-transform: scale(0.9); transform-origin: left; -webkit-transform-origin: left;}\n.smaller {font-size: 12.5px; display: inline-block; transform: scale(0.8); -webkit-transform: scale(0.8); transform-origin: left; -webkit-transform-origin: left;}\n.bt-text {font-size: 12px;margin: 1.5em 0 0 0}\n.bt-text p {margin: 0}\n</style>\n</head>\n<body>\n<div class=\"wrapper\">\n<header>\n<h2 class=\"title\">\n美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%\n</h2>\n\n<h4 class=\"meta\">\n\n\n<a class=\"head\" href=\"https://laohu8.com/wemedia/102\">\n\n\n<div class=\"h-thumb\" style=\"background-image:url(https://static.tigerbbs.com/8274c5b9d4c2852bfb1c4d6ce16c68ba);background-size:cover;\"></div>\n\n<div class=\"h-content\">\n<p class=\"h-name\">老虎资讯综合 </p>\n<p class=\"h-time\">2025-02-12 23:38</p>\n</div>\n\n</a>\n\n\n</h4>\n\n</header>\n<article>\n<html><head></head><body><p>2月12日，美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%，昆腾、Rigetti Computing涨超4%，Quantum Computing涨超3%。</p><p class=\"t-img-caption\"><img src=\"https://static.tigerbbs.com/6495c6a7711d083d65eb23830b3724e2\" tg-width=\"408\" tg-height=\"589\"/></p></body></html>\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/5d8fb95e65f042f352c6313989391357","relate_stocks":{},"source_url":"","is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"1142674620","content_text":"2月12日，美股量子计算概念股逆势集体冲高，D-Wave Quantum涨超10%，昆腾、Rigetti Computing涨超4%，Quantum Computing涨超3%。","news_type":1,"symbols_score_info":{"QBTS":1.1,"QMCO":1.1,"RGTI":1.1}},"isVote":1,"tweetType":1,"viewCount":413,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":402308102566520,"gmtCreate":1739238344764,"gmtModify":1739238638210,"author":{"id":"3547634688703059","authorId":"3547634688703059","name":"007笑","avatar":"https://static.tigerbbs.com/2b7dc0a5634de60afc7c246c20c642ff","crmLevel":4,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3547634688703059","authorIdStr":"3547634688703059"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/TME\">$腾讯音乐(TME)$ </a> 真实就好了 ……","listText":"<a href=\"https://laohu8.com/S/TME\">$腾讯音乐(TME)$ </a> 真实就好了 ……","text":"$腾讯音乐(TME)$ 真实就好了 ……","images":[{"img":"https://static.tigerbbs.com/f5ccb99010411f92346874a012b7b044","width":"972","height":"1631"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/402308102566520","isVote":1,"tweetType":1,"viewCount":1261,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":1,"langContent":"CN","totalScore":0}],"defaultTab":"posts","isTTM":false}