+关注
凯瑞昂
暂无个人介绍
IP属地:贵州
19
关注
1
粉丝
0
主题
0
勋章
主贴
热门
凯瑞昂
03-07
$台积电(TSM)$
2月6号买的,拿着没动到今天。
凯瑞昂
2023-11-29
$拼多多(PDD)$
是不是该买put了?这个加速有点不可持续的感觉。另外就是,坦白讲,拼多多对于卖家来讲太伤了,买家是重要,但买家如果叛逃了呢?卖家会一直情愿被这么剥削吗?就一直愿意被白嫖吗?不好说。
凯瑞昂
2023-06-21
$特斯拉(TSLA)$
上300去
凯瑞昂
02-02
$特斯拉(TSLA)$
150恐怕是很难到了
凯瑞昂
2021-07-01
$每日优鲜(MF)$
看图就好
凯瑞昂
2021-06-23
$满帮(YMM)$
为什么这么晚不开盘呢。
凯瑞昂
2020-11-05
问题是就是不涨啊
特斯拉在全球范围内拥有超过20000个超级充电桩
凯瑞昂
2021-06-30
$叮咚(DDL)$
夜里两点开始交易,就那么几股,真是无语。
凯瑞昂
04-24
$特斯拉(TSLA)$
这是要拉爆空头了吗?
凯瑞昂
2023-04-05
好
谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作
凯瑞昂
2023-03-09
与Facebook无关,为什么要挂在这里呢?
美股回购规模创纪录,全靠这5家公司
凯瑞昂
2021-07-22
2021.07.22 复盘笔记 by 凯瑞昂
2021-07-22 成本 现价 浮动收益 老虎证券(TIGR) 16.98 20.15 +18.65% 买入时机要求再严格一些,终于等来了一个不错的位置。
2021.07.22 复盘笔记 by 凯瑞昂
凯瑞昂
2021-06-23
刚下载了APP去储值,参加活动。发现竟然还有几块钱的余额,你看看,感觉盈利多了几块钱。这活动好。
凯瑞昂
2021-06-23
申购后发现了活动页,然后发现可用资金不足了,下次要先看活动再出手。
凯瑞昂
2021-06-15
看起来手机成色不错的样子,爱回收的服务也很棒,接触过几次,工作人员都会引导使用优惠券。相信会大涨。
凯瑞昂
2021-04-28
怪不得
特斯拉已完全偿还上海超级工厂6.14亿美元贷款
凯瑞昂
2021-04-27
多打了个万吧?
Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY
凯瑞昂
2021-04-22
会不会是玩游戏玩的兴奋了一跺脚,踩到油门了。
女车主自称特斯拉失控自动倒车!地库被撞出一个大洞
凯瑞昂
2021-02-02
2021.02.02 复盘笔记 by 凯瑞昂
2021-02-02 成本 现价 浮动收益 声网(API) 41.42 74.50 +81.08% 有时候,炒股是需要一些运气的。大涨追涨买入了几天,赶上风头正热的clubhouse了。
2021.02.02 复盘笔记 by 凯瑞昂
去老虎APP查看更多动态
{"i18n":{"language":"zh_CN"},"userPageInfo":{"id":"3453412771192240","uuid":"3453412771192240","gmtCreate":1502961081943,"gmtModify":1609847130555,"name":"凯瑞昂","pinyin":"krakairuiang","introduction":"","introductionEn":"","signature":"","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","hat":null,"hatId":null,"hatName":null,"vip":1,"status":2,"fanSize":1,"headSize":19,"tweetSize":19,"questionSize":0,"limitLevel":900,"accountStatus":4,"level":{"id":2,"name":"无畏虎","nameTw":"無畏虎","represent":"初生牛犊","factor":"发布3条非转发主帖,1条获得他人回复或点赞","iconColor":"3C9E83","bgColor":"A2F1D9"},"themeCounts":0,"badgeCounts":0,"badges":[],"moderator":false,"superModerator":false,"manageSymbols":null,"badgeLevel":null,"boolIsFan":false,"boolIsHead":false,"favoriteSize":0,"symbols":null,"coverImage":null,"realNameVerified":null,"userBadges":[{"badgeId":"228c86a078844d74991fff2b7ab2428d-2","templateUuid":"228c86a078844d74991fff2b7ab2428d","name":"投资总监虎","description":"证券账户累计交易金额达到30万美元","bigImgUrl":"https://static.tigerbbs.com/9d20b23f1b6335407f882bc5c2ad12c0","smallImgUrl":"https://static.tigerbbs.com/ada3b4533518ace8404a3f6dd192bd29","grayImgUrl":"https://static.tigerbbs.com/177f283ba21d1c077054dac07f88f3bd","redirectLinkEnabled":0,"redirectLink":null,"hasAllocated":1,"isWearing":0,"stamp":null,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2023.07.14","exceedPercentage":"80.21%","individualDisplayEnabled":0,"backgroundColor":null,"fontColor":null,"individualDisplaySort":0,"categoryType":1101},{"badgeId":"e50ce593bb40487ebfb542ca54f6a561-4","templateUuid":"e50ce593bb40487ebfb542ca54f6a561","name":"明星虎友","description":"加入老虎社区2000天","bigImgUrl":"https://static.tigerbbs.com/dddf24b906c7011de2617d4fb3f76987","smallImgUrl":"https://static.tigerbbs.com/53d58ad32c97254c6f74db8b97e6ec49","grayImgUrl":"https://static.tigerbbs.com/6304700d92ad91c7a33e2e92ec32ecc1","redirectLinkEnabled":0,"redirectLink":null,"hasAllocated":1,"isWearing":0,"stamp":null,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2023.02.21","exceedPercentage":null,"individualDisplayEnabled":0,"backgroundColor":null,"fontColor":null,"individualDisplaySort":0,"categoryType":1001},{"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},{"badgeId":"35ec162348d5460f88c959321e554969-2","templateUuid":"35ec162348d5460f88c959321e554969","name":"宗师交易员","description":"证券或期货账户累计交易次数达到100次","bigImgUrl":"https://static.tigerbbs.com/ad22cfbe2d05aa393b18e9226e4b0307","smallImgUrl":"https://static.tigerbbs.com/36702e6ff3ffe46acafee66cc85273ca","grayImgUrl":"https://static.tigerbbs.com/d52eb88fa385cf5abe2616ed63781765","redirectLinkEnabled":0,"redirectLink":null,"hasAllocated":1,"isWearing":0,"stamp":null,"stampPosition":0,"hasStamp":0,"allocationCount":1,"allocatedDate":"2021.12.21","exceedPercentage":"80.16%","individualDisplayEnabled":0,"backgroundColor":null,"fontColor":null,"individualDisplaySort":0,"categoryType":1100}],"userBadgeCount":5,"currentWearingBadge":null,"individualDisplayBadges":null,"crmLevel":2,"crmLevelSwitch":0,"location":"贵州","starInvestorFollowerNum":0,"starInvestorFlag":false,"starInvestorOrderShareNum":0,"subscribeStarInvestorNum":2,"ror":null,"winRationPercentage":null,"showRor":false,"investmentPhilosophy":null,"starInvestorSubscribeFlag":false},"baikeInfo":{},"tab":"hot","tweets":[{"id":298754890772528,"gmtCreate":1713967478994,"gmtModify":1713967480493,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a><v-v data-views=\"1\"></v-v> 这是要拉爆空头了吗?","listText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a><v-v data-views=\"1\"></v-v> 这是要拉爆空头了吗?","text":"$特斯拉(TSLA)$ 这是要拉爆空头了吗?","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/298754890772528","isVote":1,"tweetType":1,"viewCount":513,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":281796405420344,"gmtCreate":1709826414550,"gmtModify":1709826544837,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/TSM\">$台积电(TSM)$ </a> 2月6号买的,拿着没动到今天。","listText":"<a href=\"https://laohu8.com/S/TSM\">$台积电(TSM)$ </a> 2月6号买的,拿着没动到今天。","text":"$台积电(TSM)$ 2月6号买的,拿着没动到今天。","images":[{"img":"https://static.tigerbbs.com/877a3dac84f84a961ebece92ae6525bf","width":"927","height":"1599"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":1,"commentSize":4,"repostSize":0,"link":"https://laohu8.com/post/281796405420344","isVote":1,"tweetType":1,"viewCount":2148,"authorTweetTopStatus":1,"verified":2,"comments":[{"author":{"id":"3544073188249485","authorId":"3544073188249485","name":"jius","avatar":"https://static.laohu8.com/default-avatar.jpg","crmLevel":1,"crmLevelSwitch":0,"authorIdStr":"3544073188249485","idStr":"3544073188249485"},"content":"这是开了几倍杠杆?","text":"这是开了几倍杠杆?","html":"这是开了几倍杠杆?"}],"imageCount":1,"langContent":"CN","totalScore":0},{"id":269716813479976,"gmtCreate":1706856803875,"gmtModify":1706856804925,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a> 150恐怕是很难到了","listText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a> 150恐怕是很难到了","text":"$特斯拉(TSLA)$ 150恐怕是很难到了","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":1,"commentSize":2,"repostSize":0,"link":"https://laohu8.com/post/269716813479976","isVote":1,"tweetType":1,"viewCount":2729,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":246780873236504,"gmtCreate":1701270483487,"gmtModify":1701270484823,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/PDD\">$拼多多(PDD)$ </a><v-v data-views=\"0\"></v-v>是不是该买put了?这个加速有点不可持续的感觉。另外就是,坦白讲,拼多多对于卖家来讲太伤了,买家是重要,但买家如果叛逃了呢?卖家会一直情愿被这么剥削吗?就一直愿意被白嫖吗?不好说。","listText":"<a href=\"https://laohu8.com/S/PDD\">$拼多多(PDD)$ </a><v-v data-views=\"0\"></v-v>是不是该买put了?这个加速有点不可持续的感觉。另外就是,坦白讲,拼多多对于卖家来讲太伤了,买家是重要,但买家如果叛逃了呢?卖家会一直情愿被这么剥削吗?就一直愿意被白嫖吗?不好说。","text":"$拼多多(PDD)$ 是不是该买put了?这个加速有点不可持续的感觉。另外就是,坦白讲,拼多多对于卖家来讲太伤了,买家是重要,但买家如果叛逃了呢?卖家会一直情愿被这么剥削吗?就一直愿意被白嫖吗?不好说。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":5,"commentSize":3,"repostSize":0,"link":"https://laohu8.com/post/246780873236504","isVote":1,"tweetType":1,"viewCount":2649,"authorTweetTopStatus":1,"verified":2,"comments":[{"author":{"id":"3537567136208150","authorId":"3537567136208150","name":"阿Joe哥哥","avatar":"https://static.tigerbbs.com/4b837ef0611c78a61e6a699699e530e5","crmLevel":1,"crmLevelSwitch":0,"authorIdStr":"3537567136208150","idStr":"3537567136208150"},"content":"我就受不了一直给拼多多各种规则剥削血汗钱还有一直给白嫖,我已经退店了。在拼多多做商家根本就像奴隶一样!","text":"我就受不了一直给拼多多各种规则剥削血汗钱还有一直给白嫖,我已经退店了。在拼多多做商家根本就像奴隶一样!","html":"我就受不了一直给拼多多各种规则剥削血汗钱还有一直给白嫖,我已经退店了。在拼多多做商家根本就像奴隶一样!"}],"imageCount":0,"langContent":"CN","totalScore":0},{"id":189449376608256,"gmtCreate":1687277811767,"gmtModify":1687277812723,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a>上300去","listText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a>上300去","text":"$特斯拉(TSLA)$ 上300去","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":2,"commentSize":2,"repostSize":0,"link":"https://laohu8.com/post/189449376608256","isVote":1,"tweetType":1,"viewCount":2084,"authorTweetTopStatus":1,"verified":2,"comments":[{"author":{"id":"3538458176058351","authorId":"3538458176058351","name":"Ruosong","avatar":"https://static.tigerbbs.com/d815eb794e0dfe4ac8a13f8e0d781888","crmLevel":7,"crmLevelSwitch":1,"authorIdStr":"3538458176058351","idStr":"3538458176058351"},"content":"轻轻松松到前高点400","text":"轻轻松松到前高点400","html":"轻轻松松到前高点400"}],"imageCount":0,"langContent":"CN","totalScore":0},{"id":653803802,"gmtCreate":1680702808406,"gmtModify":1680702809924,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"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/653803802","repostId":"2325368910","repostType":2,"repost":{"id":"2325368910","pubTimestamp":1680685560,"share":"https://www.laohu8.com/m/news/2325368910?lang=&edition=full","pubTime":"2023-04-05 17:06","market":"us","language":"zh","title":"谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作","url":"https://stock-news.laohu8.com/highlight/detail?id=2325368910","media":"市场资讯","summary":" 与英伟达的GPU相比,谷歌TPU采用低精度计算,几乎不影响深度学习处理效果的前提,比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍。 我们还没有看到能与 ChatGPT 相匹敌的 AI 大模型,但在算力基础上,领先的可能并不是微软和 OpenAI。 本周二,谷歌公布了其训练语言大模型的超级计算机的细节,基于 TPU 的超算系统已经可以比英伟达的同类更加快速、节能。 TPU v4 的性能比 TPU v3 高 2.1 倍,性能功耗比提高 2.7 倍。scVPU 使用与 TC 的 VPU 相同的","content":"<html><body><div>\n<p cms-style=\"font-L\"> <span><a href=\"https://laohu8.com/S/GOOG\">谷歌</a></span><span></span>TPU超算,大模型性能超<a href=\"https://laohu8.com/S/NVDA\">英伟达</a>,已部署数十台:图灵奖得主新作</p>\n<p cms-style=\"font-L\"> 与英伟达的GPU相比,谷歌TPU采用低精度计算,几乎不影响深度学习处理效果的前提,比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍。</p>\n<p cms-style=\"font-L\"> 我们还没有看到能与 ChatGPT 相匹敌的 AI 大模型,但在算力基础上,领先的可能并不是<span><a href=\"https://laohu8.com/S/MSFT\">微软</a></span><span></span>和 OpenAI。</p>\n<p cms-style=\"font-L\"> 本周二,谷歌公布了其训练语言大模型的超级计算机的细节,基于 TPU 的超算系统已经可以比英伟达的同类更加快速、节能。</p>\n<p cms-style=\"font-L\"> 谷歌张量处理器(tensor processing unit,TPU)是该公司为机器学习定制的专用芯片(ASIC),第一代发布于 2016 年,成为了 AlphaGo 背后的算力。<font cms-style=\"font-L strong-Bold\">与 GPU 相比,TPU采用低精度计算,在几乎不影响深度学习处理效果的前提下大幅降低了功耗、加快运算速度。</font>同时,TPU 使用了脉动阵列等设计来优化矩阵乘法与卷积运算。</p>\n<p cms-style=\"font-L\"> 当前,谷歌 90% 以上的人工智能训练工作都在使用这些芯片,TPU 支撑了包括搜索的谷歌主要业务。作为图灵奖得主、计算机架构巨擘,大卫・<a href=\"https://laohu8.com/S/PDCO\">帕特森</a>(David Patterson)在 2016 年从 UC Berkeley 退休后,以杰出工程师的身份加入了谷歌大脑团队,为几代 TPU 的研发做出了卓越贡献。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/200/w640h360/20230405/7aea-a58abde4bc7a5e5b496021accfccbd15.png\"/><span></span></div>\n<p cms-style=\"font-L\"> 如今 TPU 已经发展到了第四代,谷歌本周二由 Norman Jouppi、大卫・帕特森等人发表的论文《 TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings 》详细介绍了自研的光通信器件是如何将 4000 多块芯片并联成为超级计算机,以提升整体效率的。</p>\n<p cms-style=\"font-L\"> <font cms-style=\"font-L strong-Bold\">TPU v4 的性能比 TPU v3 高 2.1 倍,性能功耗比提高 2.7 倍。基于 TPU v4 的超级计算机拥有 4096 块芯片,整体速度提高了约 10 倍。对于类似大小的系统,谷歌能做到比 Graphcore IPU Bow 快 4.3-4.5 倍,比 <span>Nvidia</span><span></span> A100 快 1.2-1.7 倍,功耗低 1.3-1.9 倍。</font></p>\n<p cms-style=\"font-L\"> 除了芯片本身的算力,芯片间互联已成为构建 AI 超算的公司之间竞争的关键点,最近一段时间,谷歌的 Bard、OpenAI 的 ChatGPT 这样的大语言模型(LLM)规模正在爆炸式增长,算力已经成为明显的瓶颈。</p>\n<p cms-style=\"font-L\"> 由于大模型动辄千亿的参数量,它们必须由数千块芯片共同分担,并持续数周或更长时间进行训练。谷歌的 PaLM 模型 —— 其迄今为止最大的公开披露的语言模型 —— 在训练时被拆分到了两个拥有 4000 块 TPU 芯片的超级计算机上,用时 50 天。</p>\n<p cms-style=\"font-L\"> <font cms-style=\"font-L strong-Bold\">谷歌表示,通过光电路交换机(OCS),其超级计算机可以轻松地动态重新配置芯片之间的连接,有助于避免出现问题并实时调整以提高性能。</font></p>\n<p cms-style=\"font-L\"> 下图展示了 TPU v4 4×3 方式 6 个‘面’的链接。每个面有 16 条链路,每个块总共有 96 条光链路连接到 OCS 上。要提供 3D 环面的环绕链接,相对侧的链接必须连接到相同的 OCS。因此,每个 4×3 块 TPU 连接到 6 × 16 ÷ 2 = 48 个 OCS 上。Palomar OCS 为 136×136(128 个端口加上 8 个用于链路测试和修复的备用端口),因此 48 个 OCS 连接来自 64 个 4×3 块(每个 64 个芯片)的 48 对电缆,总共并联 4096 个 TPU v4 芯片。</p>\n<p cms-style=\"font-L\"> 根据这样的排布,TPU v4(中间的 ASIC 加上 4 个 HBM 堆栈)和带有 4 个液冷封装的印刷电路板 (PCB)。该板的前面板有 4 个顶部 PCIe 连接器和 16 个底部 OSFP 连接器,用于托盘间 ICI 链接。</p>\n<p cms-style=\"font-L\"> 随后,八个 64 芯片机架构成一台 4096 芯片超算。</p>\n<p cms-style=\"font-L\"> 与超级计算机一样,工作负载由不同规模的算力承担,称为切片:64 芯片、128 芯片、256 芯片等。下图显示了当主机可用性从 99.0% 到 99.9% 不等有,及没有 OCS 时切片大小的‘有效输出’。如果没有 OCS,主机可用性必须达到 99.9% 才能提供合理的切片吞吐量。对于大多数切片大小,OCS 也有 99.0% 和 99.5% 的良好输出。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/238/w640h398/20230405/f116-fb029c68861dbe0289f3c17ddbd4141c.png\"/><span></span></div>\n<p cms-style=\"font-L\"> 与 Infiniband 相比,OCS 的成本更低、功耗更低、速度更快,成本不到系统成本的 5%,功率不到系统功率的 3%。每个 TPU v4 都包含 SparseCores 数据流处理器,可将依赖嵌入的模型加速 5 至 7 倍,但仅使用 5% 的裸片面积和功耗。</p>\n<p cms-style=\"font-L\"> ‘这种切换机制使得绕过故障组件变得容易,’谷歌研究员 Norm Jouppi 和谷歌杰出工程师大卫・帕特森在一篇关于该系统的博客文章中写道。‘这种灵活性甚至允许我们改变超级计算机互连的拓扑结构,以加速机器学习模型的性能。’</p>\n<p cms-style=\"font-L\"> 在新论文上,谷歌着重介绍了稀疏核(SparseCore,SC)的设计。在大模型的训练阶段,embedding 可以放在 TensorCore 或超级计算机的主机 CPU 上处理。TensorCore 具有宽 VPU 和矩阵单元,并针对密集操作进行了优化。由于小的聚集 / 分散内存访问和可变长度数据交换,在 TensorCore 上放置嵌入其实并不是最佳选择。在超级计算机的主机 CPU 上放置嵌入会在 CPU DRAM 接口上引发阿姆达尔定律瓶颈,并通过 4:1 TPU v4 与 CPU 主机比率放大。数据中心网络的尾部延迟和带宽限制将进一步限制训练系统。</p>\n<p cms-style=\"font-L\"> 对此,谷歌认为可以使用 TPU 超算的总 HBM 容量优化性能,加入专用 ICI 网络,并提供快速收集 / 分散内存访问支持。这导致了 SparseCore 的协同设计。</p>\n<p cms-style=\"font-L\"> SC 是一种用于嵌入训练的特定领域架构,从 TPU v2 开始,后来在 TPU v3 和 TPU v4 中得到改进。SC 相对划算,只有芯片面积的约 5% 和功率的 5% 左右。SC 结合超算规模的 HBM 和 ICI 来创建一个平坦的、全局可寻址的内存空间(TPU v4 中为 128 TiB)。与密集训练中大参数张量的全部归约相比,较小嵌入向量的全部传输使用 HBM 和 ICI 以及更细粒度的分散 / 聚集访问模式。</p>\n<p cms-style=\"font-L\"> 作为独立的核心,SC 允许跨密集计算、SC 和 ICI 通信进行并行化。下图显示了 SC 框图,谷歌将其视为‘数据流’架构(dataflow),因为数据从内存流向各种直接连接的专用计算单元。</p>\n<p cms-style=\"font-L\"> 最通用的 SC 单元是 16 个计算块(深蓝色框)。每个 tile 都有一个关联的 HBM 通道,并支持多个未完成的内存访问。每个 tile 都有一个 Fetch Unit、一个可编程的 8-wide SIMD Vector Processing Unit 和一个 Flush Unit。获取单元将 HBM 中的激活和参数读取到 2.5 MiB 稀疏向量内存 (Spmem) 的图块切片中。scVPU 使用与 TC 的 VPU 相同的 ALU。Flush Unit 在向后传递期间将更新的参数写入 HBM。此外,五个跨通道单元(金色框)执行特定的嵌入操作,正如它们的名称所解释的那样。</p>\n<p cms-style=\"font-L\"> 与 TPU v1 一样,这些单元执行类似 CISC 的指令并对可变长度输入进行操作,其中每条指令的运行时间都取决于数据。</p>\n<p cms-style=\"font-L\"> 在特定芯片数量下,TPU v3/v4 对分带宽比高 2-4 倍,嵌入速度可以提高 1.1-2.0 倍。</p>\n<p cms-style=\"font-L\"> 下图展示了谷歌自用的推荐模型(DLRM0)在不同芯片上的效率。TPU v3 比 CPU 快 9.8 倍。TPU v4 比 TPU v3 高 3.1 倍,比 CPU 高 30.1 倍。</p>\n<p cms-style=\"font-L\"> 谷歌探索了 TPU v4 超算用于 GPT-3 大语言模型时的性能,展示了预训练阶段专家设计的 1.2 倍改进。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/131/w640h291/20230405/aeaa-a5cf234fc441a2860f280ee9ab2d1572.png\"/><span></span></div>\n<p cms-style=\"font-L\"> 虽然谷歌直到现在才公布有关其超级计算机的详细信息,但自 2020 年以来,基于 TPU 的 AI 超算一直在位于俄克拉荷马州的数据中心发挥作用。谷歌表示,Midjourney 一直在使用该系统训练其模型,最近一段时间,后者已经成为 AI 画图领域最热门的平台。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/200/w640h360/20230405/a6c9-68c9e760a5e8ea3252cebfa0561367c3.png\"/><span></span></div>\n<p cms-style=\"font-L\"> 谷歌在论文中表示,对于同等大小的系统,其芯片比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍,后者与第四代 TPU 同时上市,并被用于 GPT-4 的训练。</p>\n<p cms-style=\"font-L\"> 对此,英伟达发言人拒绝置评。</p>\n<p cms-style=\"font-L\"> 当前英伟达的 AI 芯片已经进入 Hopper 架构的时代。谷歌表示,未对第四代 TPU 与英伟达目前的旗舰 H100 芯片进行比较,因为 H100 在谷歌芯片之后上市,并且采用了更先进的制程。</p>\n<p cms-style=\"font-L\"> 但同样在此,谷歌暗示了下一代 TPU 的计划,其没有提供更多细节。Jouppi 告诉路透社,谷歌拥有开发‘未来芯片的健康管道’。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/168/w640h328/20230405/4f18-fde2e15f58ec4fa8b6dc6e20221faa2c.png\"/><span></span></div>\n<p cms-style=\"font-L\"> <font cms-style=\"font-L strong-Bold\">TPU v4 比当代 DSA 芯片速度更快、功耗更低,如果考虑到互连技术,功率边缘可能会更大。通过使用具有 3D 环面拓扑的 3K TPU v4 切片,与 TPU v3 相比,谷歌的超算也能让 LLM 的训练时间大大减少。</font></p>\n<p cms-style=\"font-L\"> 性能、可扩展性和可用性使 TPU v4 超级计算机成为 LaMDA、MUM 和 PaLM 等大型语言模型 (LLM) 的主要算力。这些功能使 5400 亿参数的 PaLM 模型在 TPU v4 超算上进行训练时,能够在 50 天内维持 57.8% 的峰值硬件浮点性能。</p>\n<p cms-style=\"font-L\"> 谷歌表示,其已经部署了数十台 TPU v4 超级计算机,供内部使用和外部通过谷歌云使用。</p>\n<p cms-style=\"font-L\"> 本文作者:泽南,来源:机器之心,原文标题:《谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作》</p>\n<div>\n<span>炒股开户享福利,送投顾服务60天体验权,一对一指导服务!</span>\n<img src=\"\"/>\n</div>\n<div>\n<div><img src=\"\"/></div>\n<div>海量资讯、精准解读,尽在新浪财经APP</div>\n</div>\n<p>责任编辑:郭明煜 </p>\n</div></body></html>","source":"sina","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>谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作</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谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作\n</h2>\n\n<h4 class=\"meta\">\n\n\n2023-04-05 17:06 北京时间 <a href=https://finance.sina.com.cn/stock/usstock/c/2023-04-05/doc-imypimne9357334.shtml><strong>市场资讯</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作\n 与英伟达的GPU相比,谷歌TPU采用低精度计算,几乎不影响深度学习处理效果的前提,比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍。\n 我们还没有看到能与 ChatGPT 相匹敌的 AI 大模型,但在算力基础上,领先的可能并不是微软和 OpenAI。\n 本周二,谷歌公布了其训练语言大模型的超级计算机的细节...</p>\n\n<a href=\"https://finance.sina.com.cn/stock/usstock/c/2023-04-05/doc-imypimne9357334.shtml\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"","relate_stocks":{"LU2237443978.SGD":"Aberdeen Standard SICAV I - Global Dynamic Dividend A Acc SGD-H","LU1914381329.SGD":"Allianz Best Styles Global Equity Cl ET Acc H2-SGD","BK4587":"ChatGPT概念","BK4543":"AI","LU2237443622.USD":"Aberdeen Standard SICAV I - Global Dynamic Dividend A Acc USD","LU0061474705.USD":"THREADNEEDLE (LUX) GLOBAL DYNAMIC REAL RETURN \"AU\" (USD) ACC","LU0648000940.SGD":"Natixis Harris Associates Global Equity RA SGD","BK4588":"碎股","IE00BJJMRX11.SGD":"Janus Henderson Balanced A Acc SGD","LU1066051498.USD":"HSBC GIF GLOBAL EQUITY VOLATILITY FOCUSED \"AM2\" (USD) INC","LU1066053197.SGD":"HSBC GIF GLOBAL EQUITY VOLATILITY FOCUSED \"AM3\" (SGDHDG) INC","SG9999014898.SGD":"United Global Quality Growth Fund Dis SGD","IE00B3S45H60.SGD":"Neuberger Berman US Multicap Opportunities A Acc SGD-H","SG9999018857.SGD":"United Global Quality Growth Fd Cl Acc SGD-H","LU0672654240.SGD":"FTIF - Franklin US Opportunities A Acc SGD-H1","LU1280957306.USD":"THREADNEEDLE (LUX) US CONTRARIAN CORE EQUITIES \"AUP\" (USD) INC","BK4561":"索罗斯持仓","LU1046421795.USD":"富达环球科技A-ACC","LU1720051108.HKD":"ALLIANZ GLOBAL ARTIFICIAL INTELLIGENCE \"AT\" (HKD) ACC","LU0276348264.USD":"THREADNEEDLE (LUX) GLOBAL DYNAMIC REAL RETURN\"AUP\" (USD) INC","LU1548497426.USD":"安联环球人工智能AT Acc","IE00BFSS8Q28.SGD":"Janus Henderson Balanced A Inc SGD-H","LU0820561818.USD":"安联收益及增长平衡基金Cl AM DIS","LU1951198990.SGD":"Natixis Thematics AI & Robotics Fund H-R/A SGD-H","LU0354030511.USD":"ALLSPRING U.S. LARGE CAP GROWTH \"I\" (USD) ACC","LU0957791311.USD":"THREADNEEDLE (LUX) GLOBAL FOCUS \"ZU\" (USD) ACC","LU1201861249.SGD":"Natixis Harris Associates US Equity PA SGD-H","IE0009356076.USD":"JANUS HENDERSON GLOBAL TECHNOLOGY AND INNOVATION \"A2\" (USD) ACC","SG9999017495.SGD":"UGDP UNITED GLOBAL QUALITY GROWTH \"B\" (SGD) ACC","LU0557290698.USD":"施罗德环球可持续增长基金","NVDA":"英伟达","LU2125909247.SGD":"Natixis Thematics Meta H-R/A SGD","LU0061474960.USD":"天利环球焦点基金AU Acc","LU0170899867.USD":"EASTSPRING INVESTMENTS WORLD VALUE EQUITY \"A\" (USD) ACC","LU2087621335.USD":"ALLSPRING GLOBAL FACTOR ENHANCED EQUITY \"A\" (USD) ACC","LU0456855351.SGD":"JPMorgan Funds - Global Equity A (acc) SGD","LU1435385759.SGD":"Natixis Loomis Sayles US Growth Equity RA SGD-H","LU2237443382.USD":"Aberdeen Standard SICAV I - Global Dynamic Dividend A MIncA USD","SG9999001077.SGD":"United International Growth Fund SGD","GB00BDT5M118.USD":"天利环球扩展Alpha基金A Acc","LU0889565833.HKD":"FRANKLIN TECHNOLOGY \"A\" (HKD) ACC","LU0082616367.USD":"摩根大通美国科技A(dist)","BK4567":"ESG概念","BK4534":"瑞士信贷持仓","BK4585":"ETF&股票定投概念","IE0004445015.USD":"JANUS HENDERSON BALANCED \"A2\" (USD) ACC","LU1623119135.USD":"Natixis Mirova Global Sustainable Equity R-NPF/A USD","LU1803068979.SGD":"FTIF - Franklin Technology A (acc) SGD-H1","LU1242518857.USD":"FULLERTON LUX FUNDS - ASIA ABSOLUTE ALPHA \"I\" (USD) ACC","LU1712237335.SGD":"Natixis Mirova Global Sustainable Equity H-R-NPF/A SGD"},"source_url":"https://finance.sina.com.cn/stock/usstock/c/2023-04-05/doc-imypimne9357334.shtml","is_english":false,"share_image_url":"https://static.laohu8.com/b0d1b7e8843deea78cc308b15114de44","article_id":"2325368910","content_text":"谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作\n 与英伟达的GPU相比,谷歌TPU采用低精度计算,几乎不影响深度学习处理效果的前提,比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍。\n 我们还没有看到能与 ChatGPT 相匹敌的 AI 大模型,但在算力基础上,领先的可能并不是微软和 OpenAI。\n 本周二,谷歌公布了其训练语言大模型的超级计算机的细节,基于 TPU 的超算系统已经可以比英伟达的同类更加快速、节能。\n 谷歌张量处理器(tensor processing unit,TPU)是该公司为机器学习定制的专用芯片(ASIC),第一代发布于 2016 年,成为了 AlphaGo 背后的算力。与 GPU 相比,TPU采用低精度计算,在几乎不影响深度学习处理效果的前提下大幅降低了功耗、加快运算速度。同时,TPU 使用了脉动阵列等设计来优化矩阵乘法与卷积运算。\n 当前,谷歌 90% 以上的人工智能训练工作都在使用这些芯片,TPU 支撑了包括搜索的谷歌主要业务。作为图灵奖得主、计算机架构巨擘,大卫・帕特森(David Patterson)在 2016 年从 UC Berkeley 退休后,以杰出工程师的身份加入了谷歌大脑团队,为几代 TPU 的研发做出了卓越贡献。\n\n 如今 TPU 已经发展到了第四代,谷歌本周二由 Norman Jouppi、大卫・帕特森等人发表的论文《 TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings 》详细介绍了自研的光通信器件是如何将 4000 多块芯片并联成为超级计算机,以提升整体效率的。\n TPU v4 的性能比 TPU v3 高 2.1 倍,性能功耗比提高 2.7 倍。基于 TPU v4 的超级计算机拥有 4096 块芯片,整体速度提高了约 10 倍。对于类似大小的系统,谷歌能做到比 Graphcore IPU Bow 快 4.3-4.5 倍,比 Nvidia A100 快 1.2-1.7 倍,功耗低 1.3-1.9 倍。\n 除了芯片本身的算力,芯片间互联已成为构建 AI 超算的公司之间竞争的关键点,最近一段时间,谷歌的 Bard、OpenAI 的 ChatGPT 这样的大语言模型(LLM)规模正在爆炸式增长,算力已经成为明显的瓶颈。\n 由于大模型动辄千亿的参数量,它们必须由数千块芯片共同分担,并持续数周或更长时间进行训练。谷歌的 PaLM 模型 —— 其迄今为止最大的公开披露的语言模型 —— 在训练时被拆分到了两个拥有 4000 块 TPU 芯片的超级计算机上,用时 50 天。\n 谷歌表示,通过光电路交换机(OCS),其超级计算机可以轻松地动态重新配置芯片之间的连接,有助于避免出现问题并实时调整以提高性能。\n 下图展示了 TPU v4 4×3 方式 6 个‘面’的链接。每个面有 16 条链路,每个块总共有 96 条光链路连接到 OCS 上。要提供 3D 环面的环绕链接,相对侧的链接必须连接到相同的 OCS。因此,每个 4×3 块 TPU 连接到 6 × 16 ÷ 2 = 48 个 OCS 上。Palomar OCS 为 136×136(128 个端口加上 8 个用于链路测试和修复的备用端口),因此 48 个 OCS 连接来自 64 个 4×3 块(每个 64 个芯片)的 48 对电缆,总共并联 4096 个 TPU v4 芯片。\n 根据这样的排布,TPU v4(中间的 ASIC 加上 4 个 HBM 堆栈)和带有 4 个液冷封装的印刷电路板 (PCB)。该板的前面板有 4 个顶部 PCIe 连接器和 16 个底部 OSFP 连接器,用于托盘间 ICI 链接。\n 随后,八个 64 芯片机架构成一台 4096 芯片超算。\n 与超级计算机一样,工作负载由不同规模的算力承担,称为切片:64 芯片、128 芯片、256 芯片等。下图显示了当主机可用性从 99.0% 到 99.9% 不等有,及没有 OCS 时切片大小的‘有效输出’。如果没有 OCS,主机可用性必须达到 99.9% 才能提供合理的切片吞吐量。对于大多数切片大小,OCS 也有 99.0% 和 99.5% 的良好输出。\n\n 与 Infiniband 相比,OCS 的成本更低、功耗更低、速度更快,成本不到系统成本的 5%,功率不到系统功率的 3%。每个 TPU v4 都包含 SparseCores 数据流处理器,可将依赖嵌入的模型加速 5 至 7 倍,但仅使用 5% 的裸片面积和功耗。\n ‘这种切换机制使得绕过故障组件变得容易,’谷歌研究员 Norm Jouppi 和谷歌杰出工程师大卫・帕特森在一篇关于该系统的博客文章中写道。‘这种灵活性甚至允许我们改变超级计算机互连的拓扑结构,以加速机器学习模型的性能。’\n 在新论文上,谷歌着重介绍了稀疏核(SparseCore,SC)的设计。在大模型的训练阶段,embedding 可以放在 TensorCore 或超级计算机的主机 CPU 上处理。TensorCore 具有宽 VPU 和矩阵单元,并针对密集操作进行了优化。由于小的聚集 / 分散内存访问和可变长度数据交换,在 TensorCore 上放置嵌入其实并不是最佳选择。在超级计算机的主机 CPU 上放置嵌入会在 CPU DRAM 接口上引发阿姆达尔定律瓶颈,并通过 4:1 TPU v4 与 CPU 主机比率放大。数据中心网络的尾部延迟和带宽限制将进一步限制训练系统。\n 对此,谷歌认为可以使用 TPU 超算的总 HBM 容量优化性能,加入专用 ICI 网络,并提供快速收集 / 分散内存访问支持。这导致了 SparseCore 的协同设计。\n SC 是一种用于嵌入训练的特定领域架构,从 TPU v2 开始,后来在 TPU v3 和 TPU v4 中得到改进。SC 相对划算,只有芯片面积的约 5% 和功率的 5% 左右。SC 结合超算规模的 HBM 和 ICI 来创建一个平坦的、全局可寻址的内存空间(TPU v4 中为 128 TiB)。与密集训练中大参数张量的全部归约相比,较小嵌入向量的全部传输使用 HBM 和 ICI 以及更细粒度的分散 / 聚集访问模式。\n 作为独立的核心,SC 允许跨密集计算、SC 和 ICI 通信进行并行化。下图显示了 SC 框图,谷歌将其视为‘数据流’架构(dataflow),因为数据从内存流向各种直接连接的专用计算单元。\n 最通用的 SC 单元是 16 个计算块(深蓝色框)。每个 tile 都有一个关联的 HBM 通道,并支持多个未完成的内存访问。每个 tile 都有一个 Fetch Unit、一个可编程的 8-wide SIMD Vector Processing Unit 和一个 Flush Unit。获取单元将 HBM 中的激活和参数读取到 2.5 MiB 稀疏向量内存 (Spmem) 的图块切片中。scVPU 使用与 TC 的 VPU 相同的 ALU。Flush Unit 在向后传递期间将更新的参数写入 HBM。此外,五个跨通道单元(金色框)执行特定的嵌入操作,正如它们的名称所解释的那样。\n 与 TPU v1 一样,这些单元执行类似 CISC 的指令并对可变长度输入进行操作,其中每条指令的运行时间都取决于数据。\n 在特定芯片数量下,TPU v3/v4 对分带宽比高 2-4 倍,嵌入速度可以提高 1.1-2.0 倍。\n 下图展示了谷歌自用的推荐模型(DLRM0)在不同芯片上的效率。TPU v3 比 CPU 快 9.8 倍。TPU v4 比 TPU v3 高 3.1 倍,比 CPU 高 30.1 倍。\n 谷歌探索了 TPU v4 超算用于 GPT-3 大语言模型时的性能,展示了预训练阶段专家设计的 1.2 倍改进。\n\n 虽然谷歌直到现在才公布有关其超级计算机的详细信息,但自 2020 年以来,基于 TPU 的 AI 超算一直在位于俄克拉荷马州的数据中心发挥作用。谷歌表示,Midjourney 一直在使用该系统训练其模型,最近一段时间,后者已经成为 AI 画图领域最热门的平台。\n\n 谷歌在论文中表示,对于同等大小的系统,其芯片比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍,后者与第四代 TPU 同时上市,并被用于 GPT-4 的训练。\n 对此,英伟达发言人拒绝置评。\n 当前英伟达的 AI 芯片已经进入 Hopper 架构的时代。谷歌表示,未对第四代 TPU 与英伟达目前的旗舰 H100 芯片进行比较,因为 H100 在谷歌芯片之后上市,并且采用了更先进的制程。\n 但同样在此,谷歌暗示了下一代 TPU 的计划,其没有提供更多细节。Jouppi 告诉路透社,谷歌拥有开发‘未来芯片的健康管道’。\n\n TPU v4 比当代 DSA 芯片速度更快、功耗更低,如果考虑到互连技术,功率边缘可能会更大。通过使用具有 3D 环面拓扑的 3K TPU v4 切片,与 TPU v3 相比,谷歌的超算也能让 LLM 的训练时间大大减少。\n 性能、可扩展性和可用性使 TPU v4 超级计算机成为 LaMDA、MUM 和 PaLM 等大型语言模型 (LLM) 的主要算力。这些功能使 5400 亿参数的 PaLM 模型在 TPU v4 超算上进行训练时,能够在 50 天内维持 57.8% 的峰值硬件浮点性能。\n 谷歌表示,其已经部署了数十台 TPU v4 超级计算机,供内部使用和外部通过谷歌云使用。\n 本文作者:泽南,来源:机器之心,原文标题:《谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作》\n\n炒股开户享福利,送投顾服务60天体验权,一对一指导服务!\n\n\n\n\n海量资讯、精准解读,尽在新浪财经APP\n\n责任编辑:郭明煜","news_type":1},"isVote":1,"tweetType":1,"viewCount":1365,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":627249556,"gmtCreate":1678373866375,"gmtModify":1678373868089,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"与Facebook无关,为什么要挂在这里呢?","listText":"与Facebook无关,为什么要挂在这里呢?","text":"与Facebook无关,为什么要挂在这里呢?","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/627249556","repostId":"2318267778","repostType":2,"repost":{"id":"2318267778","pubTimestamp":1678368190,"share":"https://www.laohu8.com/m/news/2318267778?lang=&edition=full","pubTime":"2023-03-09 21:23","market":"us","language":"zh","title":"美股回购规模创纪录,全靠这5家公司","url":"https://stock-news.laohu8.com/highlight/detail?id=2318267778","media":"智通财经","summary":"摩根大通策略师表示,今年美国企业宣布的回购计划正以创纪录的速度进行,不过在2610亿美元的回购计划中,超过三分之二的回购计划仅分布在五家公司。Dubravko Lakos-Bujas领导的团队在一份报","content":"<html><head></head><body><p><a href=\"https://laohu8.com/S/JPM\">摩根大通</a>策略师表示,今年美国企业宣布的回购计划正以创纪录的速度进行,不过在2610亿美元的回购计划中,超过三分之二的回购计划仅分布在五家公司。</p><p>Dubravko Lakos-Bujas领导的团队在一份报告中写道,<a href=\"https://laohu8.com/S/CVX\">雪佛龙</a>以750亿美元的回购额领先,其次是Meta(400亿美元)、<a href=\"https://laohu8.com/S/GS\">高盛</a>(300亿美元)和<a href=\"https://laohu8.com/S/BKNG\">Booking Holdings</a>和<a href=\"https://laohu8.com/S/CRM\">赛富时</a>(各200亿美元)。</p><p><img src=\"https://static.tigerbbs.com/71f72a76205cbdde2d9cef53e826b138\" tg-width=\"528\" tg-height=\"363\" referrerpolicy=\"no-referrer\" width=\"100%\" height=\"auto\"/></p><p>不过,该行策略师指出,尽管今年迄今为止宣布的回购计划屡创新高,但执行数量一直在下降。去年第四季度股票回购减少了20%,自去年第一季度以来,回购速度有所放缓。</p><p>过去十年,回购一直是美国股票需求的一个关键来源。但鉴于当前经济增长的不确定性,同时回购成为了拜登政府增税的目标,企业可能会寻求保留现金。</p><p>摩根大通策略师表示,过去12个月,标普500指数成份股公司宣布的股票回购总额为8530亿美元,相比之下,2019年5月的峰值约为1万亿美元。</p><p><img src=\"https://static.tigerbbs.com/3fc42d54c2a592cf08658b947d95f178\" tg-width=\"530\" tg-height=\"364\" referrerpolicy=\"no-referrer\" width=\"100%\" height=\"auto\"/></p><p><b>其他观点</b></p><p>另外,摩根大通还对美国上市公司的第四季度业绩表现发表了观点。该行策略师认为,美国第四季度财报季并没有想象中的那么糟糕,但证实了企业基本面正在出现裂痕。</p><p>标普500指数成分股公司中,有64%的公司的第四季度业绩超出了预期,低于过去四季度70%的平均水平。</p><p>报告还指出,每股收益继续向下修正,2023年每股收益向下修正8美元至222美元。</p><p>该行策略师认为,下半年的盈利预期仍然过高,利润率上升的预期与劳动力成本的粘性、资本成本压力的上升以及需求放缓的风险相矛盾。</p></body></html>","source":"stock_zhitongcaijing","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>美股回购规模创纪录,全靠这5家公司</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美股回购规模创纪录,全靠这5家公司\n</h2>\n\n<h4 class=\"meta\">\n\n\n2023-03-09 21:23 北京时间 <a href=http://www.zhitongcaijing.com/content/detail/889419.html><strong>智通财经</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>摩根大通策略师表示,今年美国企业宣布的回购计划正以创纪录的速度进行,不过在2610亿美元的回购计划中,超过三分之二的回购计划仅分布在五家公司。Dubravko Lakos-Bujas领导的团队在一份报告中写道,雪佛龙以750亿美元的回购额领先,其次是Meta(400亿美元)、高盛(300亿美元)和Booking Holdings和赛富时(各200亿美元)。不过,该行策略师指出,尽管今年迄今为止宣布...</p>\n\n<a href=\"http://www.zhitongcaijing.com/content/detail/889419.html\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/c318bcd91a109139b7d70c76c30bb154","relate_stocks":{"CVX":"雪佛龙","CRM":"赛富时","BKNG":"Booking Holdings","META":"Meta Platforms, Inc.","GS":"高盛"},"source_url":"http://www.zhitongcaijing.com/content/detail/889419.html","is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"2318267778","content_text":"摩根大通策略师表示,今年美国企业宣布的回购计划正以创纪录的速度进行,不过在2610亿美元的回购计划中,超过三分之二的回购计划仅分布在五家公司。Dubravko Lakos-Bujas领导的团队在一份报告中写道,雪佛龙以750亿美元的回购额领先,其次是Meta(400亿美元)、高盛(300亿美元)和Booking Holdings和赛富时(各200亿美元)。不过,该行策略师指出,尽管今年迄今为止宣布的回购计划屡创新高,但执行数量一直在下降。去年第四季度股票回购减少了20%,自去年第一季度以来,回购速度有所放缓。过去十年,回购一直是美国股票需求的一个关键来源。但鉴于当前经济增长的不确定性,同时回购成为了拜登政府增税的目标,企业可能会寻求保留现金。摩根大通策略师表示,过去12个月,标普500指数成份股公司宣布的股票回购总额为8530亿美元,相比之下,2019年5月的峰值约为1万亿美元。其他观点另外,摩根大通还对美国上市公司的第四季度业绩表现发表了观点。该行策略师认为,美国第四季度财报季并没有想象中的那么糟糕,但证实了企业基本面正在出现裂痕。标普500指数成分股公司中,有64%的公司的第四季度业绩超出了预期,低于过去四季度70%的平均水平。报告还指出,每股收益继续向下修正,2023年每股收益向下修正8美元至222美元。该行策略师认为,下半年的盈利预期仍然过高,利润率上升的预期与劳动力成本的粘性、资本成本压力的上升以及需求放缓的风险相矛盾。","news_type":1},"isVote":1,"tweetType":1,"viewCount":1279,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":172137108,"gmtCreate":1626943379204,"gmtModify":1626944287942,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"title":"2021.07.22 复盘笔记 by 凯瑞昂","htmlText":"2021-07-22 成本 现价 浮动收益 老虎证券(TIGR) 16.98 20.15 +18.65% 买入时机要求再严格一些,终于等来了一个不错的位置。","listText":"2021-07-22 成本 现价 浮动收益 老虎证券(TIGR) 16.98 20.15 +18.65% 买入时机要求再严格一些,终于等来了一个不错的位置。","text":"2021-07-22 成本 现价 浮动收益 老虎证券(TIGR) 16.98 20.15 +18.65% 买入时机要求再严格一些,终于等来了一个不错的位置。","images":[],"top":1,"highlighted":1,"essential":1,"paper":2,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/172137108","isVote":1,"tweetType":1,"viewCount":1615,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":151209490,"gmtCreate":1625092306310,"gmtModify":1625092306310,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/MF\">$每日优鲜(MF)$</a>看图就好","listText":"<a href=\"https://laohu8.com/S/MF\">$每日优鲜(MF)$</a>看图就好","text":"$每日优鲜(MF)$看图就好","images":[{"img":"https://static.tigerbbs.com/87c510c68045f9df083e7415cf26fb1b","width":"1125","height":"2436"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":1,"commentSize":1,"repostSize":0,"link":"https://laohu8.com/post/151209490","isVote":1,"tweetType":1,"viewCount":1778,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":2,"langContent":"CN","totalScore":0},{"id":153967908,"gmtCreate":1625005927905,"gmtModify":1625005927905,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/DDL\">$叮咚(DDL)$</a>夜里两点开始交易,就那么几股,真是无语。","listText":"<a href=\"https://laohu8.com/S/DDL\">$叮咚(DDL)$</a>夜里两点开始交易,就那么几股,真是无语。","text":"$叮咚(DDL)$夜里两点开始交易,就那么几股,真是无语。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":1,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/153967908","isVote":1,"tweetType":1,"viewCount":1867,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":121867542,"gmtCreate":1624459116886,"gmtModify":1624459116886,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"刚下载了APP去储值,参加活动。发现竟然还有几块钱的余额,你看看,感觉盈利多了几块钱。这活动好。","listText":"刚下载了APP去储值,参加活动。发现竟然还有几块钱的余额,你看看,感觉盈利多了几块钱。这活动好。","text":"刚下载了APP去储值,参加活动。发现竟然还有几块钱的余额,你看看,感觉盈利多了几块钱。这活动好。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/121867542","isVote":1,"tweetType":1,"viewCount":1800,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":121865686,"gmtCreate":1624459041445,"gmtModify":1624459041445,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"申购后发现了活动页,然后发现可用资金不足了,下次要先看活动再出手。","listText":"申购后发现了活动页,然后发现可用资金不足了,下次要先看活动再出手。","text":"申购后发现了活动页,然后发现可用资金不足了,下次要先看活动再出手。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/121865686","isVote":1,"tweetType":1,"viewCount":906,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":129493271,"gmtCreate":1624379754693,"gmtModify":1624379754693,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/YMM\">$满帮(YMM)$</a>为什么这么晚不开盘呢。","listText":"<a href=\"https://laohu8.com/S/YMM\">$满帮(YMM)$</a>为什么这么晚不开盘呢。","text":"$满帮(YMM)$为什么这么晚不开盘呢。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":2,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/129493271","isVote":1,"tweetType":1,"viewCount":1161,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":187800881,"gmtCreate":1623748164343,"gmtModify":1623748164343,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"看起来手机成色不错的样子,爱回收的服务也很棒,接触过几次,工作人员都会引导使用优惠券。相信会大涨。","listText":"看起来手机成色不错的样子,爱回收的服务也很棒,接触过几次,工作人员都会引导使用优惠券。相信会大涨。","text":"看起来手机成色不错的样子,爱回收的服务也很棒,接触过几次,工作人员都会引导使用优惠券。相信会大涨。","images":[{"img":"https://static.tigerbbs.com/749c8456a0ae384685226d95d1b8b25d","width":"1125","height":"2436"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/187800881","isVote":1,"tweetType":1,"viewCount":1476,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":1,"langContent":"CN","totalScore":0},{"id":100281410,"gmtCreate":1619617223930,"gmtModify":1619617223930,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"怪不得","listText":"怪不得","text":"怪不得","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/100281410","repostId":"2130308159","repostType":2,"repost":{"id":"2130308159","pubTimestamp":1619613300,"share":"https://www.laohu8.com/m/news/2130308159?lang=&edition=full","pubTime":"2021-04-28 20:35","market":"hk","language":"zh","title":"特斯拉已完全偿还上海超级工厂6.14亿美元贷款","url":"https://stock-news.laohu8.com/highlight/detail?id=2130308159","media":"e公司","summary":"【特斯拉已完全偿还上海超级工厂6.14亿美元贷款】4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止...","content":"<html><body><div>\n<p>\r\n 原标题:<a href=\"https://laohu8.com/S/TSLA\">特斯拉</a>已完全偿还上海超级工厂6.14亿美元贷款\r\n </p>\n<div>摘要</div>\n<div>【特斯拉已完全偿还上海超级工厂6.14亿美元贷款】4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。(e公司)</div>\n<p><img border=\"0\" height=\"276\" src=\"https://webquoteklinepic.eastmoney.com/GetPic.aspx?nid=105.TSLA&imageType=k&token=28dfeb41d35cc81d84b4664d7c23c49f&at=1\" width=\"578\"/></p><p> 4月28日晚间,根据<span>特斯拉</span><span></span>向美国证券交易委员会递交的文件,<span href=\"http://quote.eastmoney.com/unify/r/105.TSLA\" target=\"_blank\" web=\"1\">特斯拉</span>披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款<span>合同</span>已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。</p><p>(文章来源:e公司)</p>\n<p>\r\n (责任编辑:DF537)\r\n </p>\n</div></body></html>","source":"stock_eastmoney","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>特斯拉已完全偿还上海超级工厂6.14亿美元贷款</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特斯拉已完全偿还上海超级工厂6.14亿美元贷款\n</h2>\n\n<h4 class=\"meta\">\n\n\n2021-04-28 20:35 北京时间 <a href=http://finance.eastmoney.com/a/202104281905257786.html><strong>e公司</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>原标题:特斯拉已完全偿还上海超级工厂6.14亿美元贷款\r\n \n摘要\n【特斯拉已完全偿还上海超级工厂6.14亿美元贷款】4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。(e公司)\n ...</p>\n\n<a href=\"http://finance.eastmoney.com/a/202104281905257786.html\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/8350896f4f33c86bc28f200b67ab82b4","relate_stocks":{"TSLA":"特斯拉"},"source_url":"http://finance.eastmoney.com/a/202104281905257786.html","is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"2130308159","content_text":"原标题:特斯拉已完全偿还上海超级工厂6.14亿美元贷款\r\n \n摘要\n【特斯拉已完全偿还上海超级工厂6.14亿美元贷款】4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。(e公司)\n 4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。(文章来源:e公司)\n\r\n (责任编辑:DF537)","news_type":1},"isVote":1,"tweetType":1,"viewCount":1602,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":374543326,"gmtCreate":1619468373965,"gmtModify":1619468373965,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"多打了个万吧?","listText":"多打了个万吧?","text":"多打了个万吧?","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/374543326","repostId":"2130343683","repostType":2,"repost":{"id":"2130343683","weMediaInfo":{"introduction":"Stock Market Quotes, Business News, Financial News, Trading Ideas, and Stock Research by Professionals","home_visible":0,"media_name":"Benzinga","id":"1052270027","head_image":"https://static.tigerbbs.com/d08bf7808052c0ca9deb4e944cae32aa"},"pubTimestamp":1619467700,"share":"https://www.laohu8.com/m/news/2130343683?lang=&edition=full","pubTime":"2021-04-27 04:08","market":"us","language":"en","title":"Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY","url":"https://stock-news.laohu8.com/highlight/detail?id=2130343683","media":"Benzinga","summary":"Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY","content":"<html><body><p>Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY</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>Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY</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\">\nTesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY\n</h2>\n\n<h4 class=\"meta\">\n\n\n<div class=\"head\" \">\n\n\n<div class=\"h-thumb\" style=\"background-image:url(https://static.tigerbbs.com/d08bf7808052c0ca9deb4e944cae32aa);background-size:cover;\"></div>\n\n<div class=\"h-content\">\n<p class=\"h-name\">Benzinga </p>\n<p class=\"h-time\">2021-04-27 04:08</p>\n</div>\n\n</div>\n\n\n</h4>\n\n</header>\n<article>\n<html><body><p>Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY</p></body></html>\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"","relate_stocks":{"TSLA":"特斯拉"},"source_url":"https://www.benzinga.com/node/20799832","is_english":true,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"2130343683","content_text":"Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY","news_type":1},"isVote":1,"tweetType":1,"viewCount":652,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":376936429,"gmtCreate":1619078676061,"gmtModify":1619078676061,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"会不会是玩游戏玩的兴奋了一跺脚,踩到油门了。","listText":"会不会是玩游戏玩的兴奋了一跺脚,踩到油门了。","text":"会不会是玩游戏玩的兴奋了一跺脚,踩到油门了。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/376936429","repostId":"2129555388","repostType":2,"repost":{"id":"2129555388","pubTimestamp":1619077800,"share":"https://www.laohu8.com/m/news/2129555388?lang=&edition=full","pubTime":"2021-04-22 15:50","market":"us","language":"zh","title":"女车主自称特斯拉失控自动倒车!地库被撞出一个大洞","url":"https://stock-news.laohu8.com/highlight/detail?id=2129555388","media":"快科技","summary":"而今,郑州又有一位女车主曝料称,自己的特斯拉车辆突然无故倒车,把地库的墙也撞了一个大洞,车辆也严重受损。此外,对于荣女士所称的,事故时一分钟左右的行车记录仪视频缺失,和满屏错误代码消失的问题,工作人员表示,特斯拉是一个负责任的企业,不可能去修改、篡改、删除客户任何的数据,这是特斯拉在中国市场的一个基本底线。","content":"<html><body><article><p>原标题:女车主自称<a href=\"https://laohu8.com/S/TSLA\">特斯拉</a>失控自动倒车!地库被撞出一个大洞</p><img src=\"https://fid-75186.picgzc.qpic.cn/20210422155358144v1731ffqjnrv2n9\"/><p>近日,特斯拉在国内已经成为了车圈热点顶流,一边是特斯拉不断致歉,希望第三方进行检测证明清白,另一方面,是越来越多的车主质疑特斯拉“失控”事故。</p><p>而今,郑州又有一位女车主曝料称,<strong>自己的特斯拉车辆突然无故倒车,把地库的墙也撞了一个大洞,车辆也严重受损。</strong>更诡异的是,事故发生时的行车记录仪视频也不翼而飞。</p><p>据媒体报道,今年3月29日下午,荣女士在小区车库停车时,突然遭遇失控倒车,荣女士称自己当时在等其他车辆挪位置,就在车里玩了会儿王者荣耀,谁知车辆突然失控,一下子撞到墙上。</p><p>视频</p><p>荣女士称虽然当时挂的是R档,但是处于驻车状况,自己并没有踩油门。而且令荣女士更加不解的是,当时满屏的错误代码后来变成了三行。<strong>并且,行车记录仪中也缺失了出事故时一分钟左右的记录,自己在电脑上恢复怎么都不能恢复成功。</strong></p><p>带着这些疑问,荣女士和记者一同来到了郑州特斯拉中原福塔店,特斯拉工作人员向他们表示,<strong>后台数据显示,在事故发生时,系统检测到加速踏板电门被深踩。</strong>因此,是驾驶员误踩电门导致的事故。</p><p>但是,对于这样的回应,荣女士并不满意,她坚称自己没有误踩电门,同时向特斯拉方面要求提供后台数据,让她自己亲眼查看。但是,特斯拉员工又称,这些数据是被加密存储在服务器端,他们也没有权限查看。</p><p>此外,对于荣女士所称的,事故时一分钟左右的行车记录仪视频缺失,和满屏错误代码消失的问题,工作人员表示,特斯拉是一个负责任的企业,不可能去修改、篡改、删除客户任何的数据,这是特斯拉在中国市场的一个基本底线。</p><p>但是,面对记者的提问“那为何行车记录仪视频真实消失了呢?”,<strong>特斯拉工作员工只能表示,这个现象确实没有见到,因此无法回复。</strong></p><p>当前,荣女士和特斯拉方间的协商还在进行当中,对此,我们也会保持关注。</p><p>(文章来源:快科技)</p></article></body></html>","source":"tencent","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>女车主自称特斯拉失控自动倒车!地库被撞出一个大洞</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女车主自称特斯拉失控自动倒车!地库被撞出一个大洞\n</h2>\n\n<h4 class=\"meta\">\n\n\n2021-04-22 15:50 北京时间 <a href=http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN202104221553587dd3279e&s=b><strong>快科技</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>原标题:女车主自称特斯拉失控自动倒车!地库被撞出一个大洞近日,特斯拉在国内已经成为了车圈热点顶流,一边是特斯拉不断致歉,希望第三方进行检测证明清白,另一方面,是越来越多的车主质疑特斯拉“失控”事故。而今,郑州又有一位女车主曝料称,自己的特斯拉车辆突然无故倒车,把地库的墙也撞了一个大洞,车辆也严重受损。更诡异的是,事故发生时的行车记录仪视频也不翼而飞。据媒体报道,今年3月29日下午,荣女士在小区车库...</p>\n\n<a href=\"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN202104221553587dd3279e&s=b\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/9749762c5ee11a52030b4c66a89c8026","relate_stocks":{"TSLA":"特斯拉"},"source_url":"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN202104221553587dd3279e&s=b","is_english":false,"share_image_url":"https://static.laohu8.com/9a95c1376e76363c1401fee7d3717173","article_id":"2129555388","content_text":"原标题:女车主自称特斯拉失控自动倒车!地库被撞出一个大洞近日,特斯拉在国内已经成为了车圈热点顶流,一边是特斯拉不断致歉,希望第三方进行检测证明清白,另一方面,是越来越多的车主质疑特斯拉“失控”事故。而今,郑州又有一位女车主曝料称,自己的特斯拉车辆突然无故倒车,把地库的墙也撞了一个大洞,车辆也严重受损。更诡异的是,事故发生时的行车记录仪视频也不翼而飞。据媒体报道,今年3月29日下午,荣女士在小区车库停车时,突然遭遇失控倒车,荣女士称自己当时在等其他车辆挪位置,就在车里玩了会儿王者荣耀,谁知车辆突然失控,一下子撞到墙上。视频荣女士称虽然当时挂的是R档,但是处于驻车状况,自己并没有踩油门。而且令荣女士更加不解的是,当时满屏的错误代码后来变成了三行。并且,行车记录仪中也缺失了出事故时一分钟左右的记录,自己在电脑上恢复怎么都不能恢复成功。带着这些疑问,荣女士和记者一同来到了郑州特斯拉中原福塔店,特斯拉工作人员向他们表示,后台数据显示,在事故发生时,系统检测到加速踏板电门被深踩。因此,是驾驶员误踩电门导致的事故。但是,对于这样的回应,荣女士并不满意,她坚称自己没有误踩电门,同时向特斯拉方面要求提供后台数据,让她自己亲眼查看。但是,特斯拉员工又称,这些数据是被加密存储在服务器端,他们也没有权限查看。此外,对于荣女士所称的,事故时一分钟左右的行车记录仪视频缺失,和满屏错误代码消失的问题,工作人员表示,特斯拉是一个负责任的企业,不可能去修改、篡改、删除客户任何的数据,这是特斯拉在中国市场的一个基本底线。但是,面对记者的提问“那为何行车记录仪视频真实消失了呢?”,特斯拉工作员工只能表示,这个现象确实没有见到,因此无法回复。当前,荣女士和特斯拉方间的协商还在进行当中,对此,我们也会保持关注。(文章来源:快科技)","news_type":1},"isVote":1,"tweetType":1,"viewCount":2331,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":315116420,"gmtCreate":1612220296360,"gmtModify":1703758921666,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"title":"2021.02.02 复盘笔记 by 凯瑞昂","htmlText":"2021-02-02 成本 现价 浮动收益 声网(API) 41.42 74.50 +81.08% 有时候,炒股是需要一些运气的。大涨追涨买入了几天,赶上风头正热的clubhouse了。","listText":"2021-02-02 成本 现价 浮动收益 声网(API) 41.42 74.50 +81.08% 有时候,炒股是需要一些运气的。大涨追涨买入了几天,赶上风头正热的clubhouse了。","text":"2021-02-02 成本 现价 浮动收益 声网(API) 41.42 74.50 +81.08% 有时候,炒股是需要一些运气的。大涨追涨买入了几天,赶上风头正热的clubhouse了。","images":[],"top":1,"highlighted":1,"essential":1,"paper":2,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/315116420","isVote":1,"tweetType":1,"viewCount":1617,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":302800460,"gmtCreate":1604586438150,"gmtModify":1703833932220,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"authorIdStr":"3453412771192240","idStr":"3453412771192240"},"themes":[],"htmlText":"问题是就是不涨啊","listText":"问题是就是不涨啊","text":"问题是就是不涨啊","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":1,"repostSize":0,"link":"https://laohu8.com/post/302800460","repostId":"1126874258","repostType":4,"repost":{"id":"1126874258","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":1604582255,"share":"https://www.laohu8.com/m/news/1126874258?lang=&edition=full","pubTime":"2020-11-05 21:17","market":"sh","language":"zh","title":"特斯拉在全球范围内拥有超过20000个超级充电桩","url":"https://stock-news.laohu8.com/highlight/detail?id=1126874258","media":"老虎资讯综合","summary":"特斯拉官方公众号:截至目前,特斯拉在全球范围内拥有超过20000个超级充电桩。","content":"<p>特斯拉官方公众号:截至目前,特斯拉在全球范围内拥有超过20000个超级充电桩。</p>\n<p><img src=\"https://static.tigerbbs.com/0364c1190c211d3dfa88e734e40d21da\" tg-width=\"511\" tg-height=\"756\"></p>","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>特斯拉在全球范围内拥有超过20000个超级充电桩</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特斯拉在全球范围内拥有超过20000个超级充电桩\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\">2020-11-05 21:17</p>\n</div>\n\n</a>\n\n\n</h4>\n\n</header>\n<article>\n<p>特斯拉官方公众号:截至目前,特斯拉在全球范围内拥有超过20000个超级充电桩。</p>\n<p><img src=\"https://static.tigerbbs.com/0364c1190c211d3dfa88e734e40d21da\" tg-width=\"511\" tg-height=\"756\"></p>\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/cf056c93b86b4b78405c574b04f01c45","relate_stocks":{"TSLA":"特斯拉"},"is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"1126874258","content_text":"特斯拉官方公众号:截至目前,特斯拉在全球范围内拥有超过20000个超级充电桩。","news_type":1},"isVote":1,"tweetType":1,"viewCount":1203,"authorTweetTopStatus":1,"verified":2,"comments":[{"author":{"id":"3527667803686145","authorId":"3527667803686145","name":"社区成长助手","avatar":"https://static.tigerbbs.com/2b7c7106b5c0c8b0037faa67439d898f","crmLevel":1,"crmLevelSwitch":0,"authorIdStr":"3527667803686145","idStr":"3527667803686145"},"content":"终于等到了您的初发帖[比心][比心]发帖时关联相关股票或者相关话题,可以获得更多曝光哦~如果您想创作优质文章,请查看老虎社区创作指引","text":"终于等到了您的初发帖[比心][比心]发帖时关联相关股票或者相关话题,可以获得更多曝光哦~如果您想创作优质文章,请查看老虎社区创作指引","html":"终于等到了您的初发帖[比心][比心]发帖时关联相关股票或者相关话题,可以获得更多曝光哦~如果您想创作优质文章,请查看老虎社区创作指引"}],"imageCount":0,"langContent":"CN","totalScore":0}],"hots":[{"id":281796405420344,"gmtCreate":1709826414550,"gmtModify":1709826544837,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/TSM\">$台积电(TSM)$ </a> 2月6号买的,拿着没动到今天。","listText":"<a href=\"https://laohu8.com/S/TSM\">$台积电(TSM)$ </a> 2月6号买的,拿着没动到今天。","text":"$台积电(TSM)$ 2月6号买的,拿着没动到今天。","images":[{"img":"https://static.tigerbbs.com/877a3dac84f84a961ebece92ae6525bf","width":"927","height":"1599"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":1,"commentSize":4,"repostSize":0,"link":"https://laohu8.com/post/281796405420344","isVote":1,"tweetType":1,"viewCount":2148,"authorTweetTopStatus":1,"verified":2,"comments":[{"author":{"id":"3544073188249485","authorId":"3544073188249485","name":"jius","avatar":"https://static.laohu8.com/default-avatar.jpg","crmLevel":1,"crmLevelSwitch":0,"idStr":"3544073188249485","authorIdStr":"3544073188249485"},"content":"这是开了几倍杠杆?","text":"这是开了几倍杠杆?","html":"这是开了几倍杠杆?"}],"imageCount":1,"langContent":"CN","totalScore":0},{"id":246780873236504,"gmtCreate":1701270483487,"gmtModify":1701270484823,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/PDD\">$拼多多(PDD)$ </a><v-v data-views=\"0\"></v-v>是不是该买put了?这个加速有点不可持续的感觉。另外就是,坦白讲,拼多多对于卖家来讲太伤了,买家是重要,但买家如果叛逃了呢?卖家会一直情愿被这么剥削吗?就一直愿意被白嫖吗?不好说。","listText":"<a href=\"https://laohu8.com/S/PDD\">$拼多多(PDD)$ </a><v-v data-views=\"0\"></v-v>是不是该买put了?这个加速有点不可持续的感觉。另外就是,坦白讲,拼多多对于卖家来讲太伤了,买家是重要,但买家如果叛逃了呢?卖家会一直情愿被这么剥削吗?就一直愿意被白嫖吗?不好说。","text":"$拼多多(PDD)$ 是不是该买put了?这个加速有点不可持续的感觉。另外就是,坦白讲,拼多多对于卖家来讲太伤了,买家是重要,但买家如果叛逃了呢?卖家会一直情愿被这么剥削吗?就一直愿意被白嫖吗?不好说。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":5,"commentSize":3,"repostSize":0,"link":"https://laohu8.com/post/246780873236504","isVote":1,"tweetType":1,"viewCount":2649,"authorTweetTopStatus":1,"verified":2,"comments":[{"author":{"id":"3537567136208150","authorId":"3537567136208150","name":"阿Joe哥哥","avatar":"https://static.tigerbbs.com/4b837ef0611c78a61e6a699699e530e5","crmLevel":1,"crmLevelSwitch":0,"idStr":"3537567136208150","authorIdStr":"3537567136208150"},"content":"我就受不了一直给拼多多各种规则剥削血汗钱还有一直给白嫖,我已经退店了。在拼多多做商家根本就像奴隶一样!","text":"我就受不了一直给拼多多各种规则剥削血汗钱还有一直给白嫖,我已经退店了。在拼多多做商家根本就像奴隶一样!","html":"我就受不了一直给拼多多各种规则剥削血汗钱还有一直给白嫖,我已经退店了。在拼多多做商家根本就像奴隶一样!"}],"imageCount":0,"langContent":"CN","totalScore":0},{"id":189449376608256,"gmtCreate":1687277811767,"gmtModify":1687277812723,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a>上300去","listText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a>上300去","text":"$特斯拉(TSLA)$ 上300去","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":2,"commentSize":2,"repostSize":0,"link":"https://laohu8.com/post/189449376608256","isVote":1,"tweetType":1,"viewCount":2084,"authorTweetTopStatus":1,"verified":2,"comments":[{"author":{"id":"3538458176058351","authorId":"3538458176058351","name":"Ruosong","avatar":"https://static.tigerbbs.com/d815eb794e0dfe4ac8a13f8e0d781888","crmLevel":7,"crmLevelSwitch":1,"idStr":"3538458176058351","authorIdStr":"3538458176058351"},"content":"轻轻松松到前高点400","text":"轻轻松松到前高点400","html":"轻轻松松到前高点400"}],"imageCount":0,"langContent":"CN","totalScore":0},{"id":269716813479976,"gmtCreate":1706856803875,"gmtModify":1706856804925,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a> 150恐怕是很难到了","listText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a> 150恐怕是很难到了","text":"$特斯拉(TSLA)$ 150恐怕是很难到了","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":1,"commentSize":2,"repostSize":0,"link":"https://laohu8.com/post/269716813479976","isVote":1,"tweetType":1,"viewCount":2729,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":151209490,"gmtCreate":1625092306310,"gmtModify":1625092306310,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/MF\">$每日优鲜(MF)$</a>看图就好","listText":"<a href=\"https://laohu8.com/S/MF\">$每日优鲜(MF)$</a>看图就好","text":"$每日优鲜(MF)$看图就好","images":[{"img":"https://static.tigerbbs.com/87c510c68045f9df083e7415cf26fb1b","width":"1125","height":"2436"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":1,"commentSize":1,"repostSize":0,"link":"https://laohu8.com/post/151209490","isVote":1,"tweetType":1,"viewCount":1778,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":2,"langContent":"CN","totalScore":0},{"id":129493271,"gmtCreate":1624379754693,"gmtModify":1624379754693,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/YMM\">$满帮(YMM)$</a>为什么这么晚不开盘呢。","listText":"<a href=\"https://laohu8.com/S/YMM\">$满帮(YMM)$</a>为什么这么晚不开盘呢。","text":"$满帮(YMM)$为什么这么晚不开盘呢。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":2,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/129493271","isVote":1,"tweetType":1,"viewCount":1161,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":302800460,"gmtCreate":1604586438150,"gmtModify":1703833932220,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"问题是就是不涨啊","listText":"问题是就是不涨啊","text":"问题是就是不涨啊","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":1,"repostSize":0,"link":"https://laohu8.com/post/302800460","repostId":"1126874258","repostType":4,"repost":{"id":"1126874258","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":1604582255,"share":"https://www.laohu8.com/m/news/1126874258?lang=&edition=full","pubTime":"2020-11-05 21:17","market":"sh","language":"zh","title":"特斯拉在全球范围内拥有超过20000个超级充电桩","url":"https://stock-news.laohu8.com/highlight/detail?id=1126874258","media":"老虎资讯综合","summary":"特斯拉官方公众号:截至目前,特斯拉在全球范围内拥有超过20000个超级充电桩。","content":"<p>特斯拉官方公众号:截至目前,特斯拉在全球范围内拥有超过20000个超级充电桩。</p>\n<p><img src=\"https://static.tigerbbs.com/0364c1190c211d3dfa88e734e40d21da\" tg-width=\"511\" tg-height=\"756\"></p>","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>特斯拉在全球范围内拥有超过20000个超级充电桩</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特斯拉在全球范围内拥有超过20000个超级充电桩\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\">2020-11-05 21:17</p>\n</div>\n\n</a>\n\n\n</h4>\n\n</header>\n<article>\n<p>特斯拉官方公众号:截至目前,特斯拉在全球范围内拥有超过20000个超级充电桩。</p>\n<p><img src=\"https://static.tigerbbs.com/0364c1190c211d3dfa88e734e40d21da\" tg-width=\"511\" tg-height=\"756\"></p>\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/cf056c93b86b4b78405c574b04f01c45","relate_stocks":{"TSLA":"特斯拉"},"is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"1126874258","content_text":"特斯拉官方公众号:截至目前,特斯拉在全球范围内拥有超过20000个超级充电桩。","news_type":1},"isVote":1,"tweetType":1,"viewCount":1203,"authorTweetTopStatus":1,"verified":2,"comments":[{"author":{"id":"3527667803686145","authorId":"3527667803686145","name":"社区成长助手","avatar":"https://static.tigerbbs.com/2b7c7106b5c0c8b0037faa67439d898f","crmLevel":1,"crmLevelSwitch":0,"idStr":"3527667803686145","authorIdStr":"3527667803686145"},"content":"终于等到了您的初发帖[比心][比心]发帖时关联相关股票或者相关话题,可以获得更多曝光哦~如果您想创作优质文章,请查看老虎社区创作指引","text":"终于等到了您的初发帖[比心][比心]发帖时关联相关股票或者相关话题,可以获得更多曝光哦~如果您想创作优质文章,请查看老虎社区创作指引","html":"终于等到了您的初发帖[比心][比心]发帖时关联相关股票或者相关话题,可以获得更多曝光哦~如果您想创作优质文章,请查看老虎社区创作指引"}],"imageCount":0,"langContent":"CN","totalScore":0},{"id":153967908,"gmtCreate":1625005927905,"gmtModify":1625005927905,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/DDL\">$叮咚(DDL)$</a>夜里两点开始交易,就那么几股,真是无语。","listText":"<a href=\"https://laohu8.com/S/DDL\">$叮咚(DDL)$</a>夜里两点开始交易,就那么几股,真是无语。","text":"$叮咚(DDL)$夜里两点开始交易,就那么几股,真是无语。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":1,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/153967908","isVote":1,"tweetType":1,"viewCount":1867,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":298754890772528,"gmtCreate":1713967478994,"gmtModify":1713967480493,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a><v-v data-views=\"1\"></v-v> 这是要拉爆空头了吗?","listText":"<a href=\"https://laohu8.com/S/TSLA\">$特斯拉(TSLA)$ </a><v-v data-views=\"1\"></v-v> 这是要拉爆空头了吗?","text":"$特斯拉(TSLA)$ 这是要拉爆空头了吗?","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/298754890772528","isVote":1,"tweetType":1,"viewCount":513,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":653803802,"gmtCreate":1680702808406,"gmtModify":1680702809924,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"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/653803802","repostId":"2325368910","repostType":2,"repost":{"id":"2325368910","pubTimestamp":1680685560,"share":"https://www.laohu8.com/m/news/2325368910?lang=&edition=full","pubTime":"2023-04-05 17:06","market":"us","language":"zh","title":"谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作","url":"https://stock-news.laohu8.com/highlight/detail?id=2325368910","media":"市场资讯","summary":" 与英伟达的GPU相比,谷歌TPU采用低精度计算,几乎不影响深度学习处理效果的前提,比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍。 我们还没有看到能与 ChatGPT 相匹敌的 AI 大模型,但在算力基础上,领先的可能并不是微软和 OpenAI。 本周二,谷歌公布了其训练语言大模型的超级计算机的细节,基于 TPU 的超算系统已经可以比英伟达的同类更加快速、节能。 TPU v4 的性能比 TPU v3 高 2.1 倍,性能功耗比提高 2.7 倍。scVPU 使用与 TC 的 VPU 相同的","content":"<html><body><div>\n<p cms-style=\"font-L\"> <span><a href=\"https://laohu8.com/S/GOOG\">谷歌</a></span><span></span>TPU超算,大模型性能超<a href=\"https://laohu8.com/S/NVDA\">英伟达</a>,已部署数十台:图灵奖得主新作</p>\n<p cms-style=\"font-L\"> 与英伟达的GPU相比,谷歌TPU采用低精度计算,几乎不影响深度学习处理效果的前提,比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍。</p>\n<p cms-style=\"font-L\"> 我们还没有看到能与 ChatGPT 相匹敌的 AI 大模型,但在算力基础上,领先的可能并不是<span><a href=\"https://laohu8.com/S/MSFT\">微软</a></span><span></span>和 OpenAI。</p>\n<p cms-style=\"font-L\"> 本周二,谷歌公布了其训练语言大模型的超级计算机的细节,基于 TPU 的超算系统已经可以比英伟达的同类更加快速、节能。</p>\n<p cms-style=\"font-L\"> 谷歌张量处理器(tensor processing unit,TPU)是该公司为机器学习定制的专用芯片(ASIC),第一代发布于 2016 年,成为了 AlphaGo 背后的算力。<font cms-style=\"font-L strong-Bold\">与 GPU 相比,TPU采用低精度计算,在几乎不影响深度学习处理效果的前提下大幅降低了功耗、加快运算速度。</font>同时,TPU 使用了脉动阵列等设计来优化矩阵乘法与卷积运算。</p>\n<p cms-style=\"font-L\"> 当前,谷歌 90% 以上的人工智能训练工作都在使用这些芯片,TPU 支撑了包括搜索的谷歌主要业务。作为图灵奖得主、计算机架构巨擘,大卫・<a href=\"https://laohu8.com/S/PDCO\">帕特森</a>(David Patterson)在 2016 年从 UC Berkeley 退休后,以杰出工程师的身份加入了谷歌大脑团队,为几代 TPU 的研发做出了卓越贡献。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/200/w640h360/20230405/7aea-a58abde4bc7a5e5b496021accfccbd15.png\"/><span></span></div>\n<p cms-style=\"font-L\"> 如今 TPU 已经发展到了第四代,谷歌本周二由 Norman Jouppi、大卫・帕特森等人发表的论文《 TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings 》详细介绍了自研的光通信器件是如何将 4000 多块芯片并联成为超级计算机,以提升整体效率的。</p>\n<p cms-style=\"font-L\"> <font cms-style=\"font-L strong-Bold\">TPU v4 的性能比 TPU v3 高 2.1 倍,性能功耗比提高 2.7 倍。基于 TPU v4 的超级计算机拥有 4096 块芯片,整体速度提高了约 10 倍。对于类似大小的系统,谷歌能做到比 Graphcore IPU Bow 快 4.3-4.5 倍,比 <span>Nvidia</span><span></span> A100 快 1.2-1.7 倍,功耗低 1.3-1.9 倍。</font></p>\n<p cms-style=\"font-L\"> 除了芯片本身的算力,芯片间互联已成为构建 AI 超算的公司之间竞争的关键点,最近一段时间,谷歌的 Bard、OpenAI 的 ChatGPT 这样的大语言模型(LLM)规模正在爆炸式增长,算力已经成为明显的瓶颈。</p>\n<p cms-style=\"font-L\"> 由于大模型动辄千亿的参数量,它们必须由数千块芯片共同分担,并持续数周或更长时间进行训练。谷歌的 PaLM 模型 —— 其迄今为止最大的公开披露的语言模型 —— 在训练时被拆分到了两个拥有 4000 块 TPU 芯片的超级计算机上,用时 50 天。</p>\n<p cms-style=\"font-L\"> <font cms-style=\"font-L strong-Bold\">谷歌表示,通过光电路交换机(OCS),其超级计算机可以轻松地动态重新配置芯片之间的连接,有助于避免出现问题并实时调整以提高性能。</font></p>\n<p cms-style=\"font-L\"> 下图展示了 TPU v4 4×3 方式 6 个‘面’的链接。每个面有 16 条链路,每个块总共有 96 条光链路连接到 OCS 上。要提供 3D 环面的环绕链接,相对侧的链接必须连接到相同的 OCS。因此,每个 4×3 块 TPU 连接到 6 × 16 ÷ 2 = 48 个 OCS 上。Palomar OCS 为 136×136(128 个端口加上 8 个用于链路测试和修复的备用端口),因此 48 个 OCS 连接来自 64 个 4×3 块(每个 64 个芯片)的 48 对电缆,总共并联 4096 个 TPU v4 芯片。</p>\n<p cms-style=\"font-L\"> 根据这样的排布,TPU v4(中间的 ASIC 加上 4 个 HBM 堆栈)和带有 4 个液冷封装的印刷电路板 (PCB)。该板的前面板有 4 个顶部 PCIe 连接器和 16 个底部 OSFP 连接器,用于托盘间 ICI 链接。</p>\n<p cms-style=\"font-L\"> 随后,八个 64 芯片机架构成一台 4096 芯片超算。</p>\n<p cms-style=\"font-L\"> 与超级计算机一样,工作负载由不同规模的算力承担,称为切片:64 芯片、128 芯片、256 芯片等。下图显示了当主机可用性从 99.0% 到 99.9% 不等有,及没有 OCS 时切片大小的‘有效输出’。如果没有 OCS,主机可用性必须达到 99.9% 才能提供合理的切片吞吐量。对于大多数切片大小,OCS 也有 99.0% 和 99.5% 的良好输出。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/238/w640h398/20230405/f116-fb029c68861dbe0289f3c17ddbd4141c.png\"/><span></span></div>\n<p cms-style=\"font-L\"> 与 Infiniband 相比,OCS 的成本更低、功耗更低、速度更快,成本不到系统成本的 5%,功率不到系统功率的 3%。每个 TPU v4 都包含 SparseCores 数据流处理器,可将依赖嵌入的模型加速 5 至 7 倍,但仅使用 5% 的裸片面积和功耗。</p>\n<p cms-style=\"font-L\"> ‘这种切换机制使得绕过故障组件变得容易,’谷歌研究员 Norm Jouppi 和谷歌杰出工程师大卫・帕特森在一篇关于该系统的博客文章中写道。‘这种灵活性甚至允许我们改变超级计算机互连的拓扑结构,以加速机器学习模型的性能。’</p>\n<p cms-style=\"font-L\"> 在新论文上,谷歌着重介绍了稀疏核(SparseCore,SC)的设计。在大模型的训练阶段,embedding 可以放在 TensorCore 或超级计算机的主机 CPU 上处理。TensorCore 具有宽 VPU 和矩阵单元,并针对密集操作进行了优化。由于小的聚集 / 分散内存访问和可变长度数据交换,在 TensorCore 上放置嵌入其实并不是最佳选择。在超级计算机的主机 CPU 上放置嵌入会在 CPU DRAM 接口上引发阿姆达尔定律瓶颈,并通过 4:1 TPU v4 与 CPU 主机比率放大。数据中心网络的尾部延迟和带宽限制将进一步限制训练系统。</p>\n<p cms-style=\"font-L\"> 对此,谷歌认为可以使用 TPU 超算的总 HBM 容量优化性能,加入专用 ICI 网络,并提供快速收集 / 分散内存访问支持。这导致了 SparseCore 的协同设计。</p>\n<p cms-style=\"font-L\"> SC 是一种用于嵌入训练的特定领域架构,从 TPU v2 开始,后来在 TPU v3 和 TPU v4 中得到改进。SC 相对划算,只有芯片面积的约 5% 和功率的 5% 左右。SC 结合超算规模的 HBM 和 ICI 来创建一个平坦的、全局可寻址的内存空间(TPU v4 中为 128 TiB)。与密集训练中大参数张量的全部归约相比,较小嵌入向量的全部传输使用 HBM 和 ICI 以及更细粒度的分散 / 聚集访问模式。</p>\n<p cms-style=\"font-L\"> 作为独立的核心,SC 允许跨密集计算、SC 和 ICI 通信进行并行化。下图显示了 SC 框图,谷歌将其视为‘数据流’架构(dataflow),因为数据从内存流向各种直接连接的专用计算单元。</p>\n<p cms-style=\"font-L\"> 最通用的 SC 单元是 16 个计算块(深蓝色框)。每个 tile 都有一个关联的 HBM 通道,并支持多个未完成的内存访问。每个 tile 都有一个 Fetch Unit、一个可编程的 8-wide SIMD Vector Processing Unit 和一个 Flush Unit。获取单元将 HBM 中的激活和参数读取到 2.5 MiB 稀疏向量内存 (Spmem) 的图块切片中。scVPU 使用与 TC 的 VPU 相同的 ALU。Flush Unit 在向后传递期间将更新的参数写入 HBM。此外,五个跨通道单元(金色框)执行特定的嵌入操作,正如它们的名称所解释的那样。</p>\n<p cms-style=\"font-L\"> 与 TPU v1 一样,这些单元执行类似 CISC 的指令并对可变长度输入进行操作,其中每条指令的运行时间都取决于数据。</p>\n<p cms-style=\"font-L\"> 在特定芯片数量下,TPU v3/v4 对分带宽比高 2-4 倍,嵌入速度可以提高 1.1-2.0 倍。</p>\n<p cms-style=\"font-L\"> 下图展示了谷歌自用的推荐模型(DLRM0)在不同芯片上的效率。TPU v3 比 CPU 快 9.8 倍。TPU v4 比 TPU v3 高 3.1 倍,比 CPU 高 30.1 倍。</p>\n<p cms-style=\"font-L\"> 谷歌探索了 TPU v4 超算用于 GPT-3 大语言模型时的性能,展示了预训练阶段专家设计的 1.2 倍改进。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/131/w640h291/20230405/aeaa-a5cf234fc441a2860f280ee9ab2d1572.png\"/><span></span></div>\n<p cms-style=\"font-L\"> 虽然谷歌直到现在才公布有关其超级计算机的详细信息,但自 2020 年以来,基于 TPU 的 AI 超算一直在位于俄克拉荷马州的数据中心发挥作用。谷歌表示,Midjourney 一直在使用该系统训练其模型,最近一段时间,后者已经成为 AI 画图领域最热门的平台。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/200/w640h360/20230405/a6c9-68c9e760a5e8ea3252cebfa0561367c3.png\"/><span></span></div>\n<p cms-style=\"font-L\"> 谷歌在论文中表示,对于同等大小的系统,其芯片比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍,后者与第四代 TPU 同时上市,并被用于 GPT-4 的训练。</p>\n<p cms-style=\"font-L\"> 对此,英伟达发言人拒绝置评。</p>\n<p cms-style=\"font-L\"> 当前英伟达的 AI 芯片已经进入 Hopper 架构的时代。谷歌表示,未对第四代 TPU 与英伟达目前的旗舰 H100 芯片进行比较,因为 H100 在谷歌芯片之后上市,并且采用了更先进的制程。</p>\n<p cms-style=\"font-L\"> 但同样在此,谷歌暗示了下一代 TPU 的计划,其没有提供更多细节。Jouppi 告诉路透社,谷歌拥有开发‘未来芯片的健康管道’。</p>\n<div><img src=\"http://n.sinaimg.cn/spider20230405/168/w640h328/20230405/4f18-fde2e15f58ec4fa8b6dc6e20221faa2c.png\"/><span></span></div>\n<p cms-style=\"font-L\"> <font cms-style=\"font-L strong-Bold\">TPU v4 比当代 DSA 芯片速度更快、功耗更低,如果考虑到互连技术,功率边缘可能会更大。通过使用具有 3D 环面拓扑的 3K TPU v4 切片,与 TPU v3 相比,谷歌的超算也能让 LLM 的训练时间大大减少。</font></p>\n<p cms-style=\"font-L\"> 性能、可扩展性和可用性使 TPU v4 超级计算机成为 LaMDA、MUM 和 PaLM 等大型语言模型 (LLM) 的主要算力。这些功能使 5400 亿参数的 PaLM 模型在 TPU v4 超算上进行训练时,能够在 50 天内维持 57.8% 的峰值硬件浮点性能。</p>\n<p cms-style=\"font-L\"> 谷歌表示,其已经部署了数十台 TPU v4 超级计算机,供内部使用和外部通过谷歌云使用。</p>\n<p cms-style=\"font-L\"> 本文作者:泽南,来源:机器之心,原文标题:《谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作》</p>\n<div>\n<span>炒股开户享福利,送投顾服务60天体验权,一对一指导服务!</span>\n<img src=\"\"/>\n</div>\n<div>\n<div><img src=\"\"/></div>\n<div>海量资讯、精准解读,尽在新浪财经APP</div>\n</div>\n<p>责任编辑:郭明煜 </p>\n</div></body></html>","source":"sina","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>谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作</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谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作\n</h2>\n\n<h4 class=\"meta\">\n\n\n2023-04-05 17:06 北京时间 <a href=https://finance.sina.com.cn/stock/usstock/c/2023-04-05/doc-imypimne9357334.shtml><strong>市场资讯</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作\n 与英伟达的GPU相比,谷歌TPU采用低精度计算,几乎不影响深度学习处理效果的前提,比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍。\n 我们还没有看到能与 ChatGPT 相匹敌的 AI 大模型,但在算力基础上,领先的可能并不是微软和 OpenAI。\n 本周二,谷歌公布了其训练语言大模型的超级计算机的细节...</p>\n\n<a href=\"https://finance.sina.com.cn/stock/usstock/c/2023-04-05/doc-imypimne9357334.shtml\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"","relate_stocks":{"LU2237443978.SGD":"Aberdeen Standard SICAV I - Global Dynamic Dividend A Acc SGD-H","LU1914381329.SGD":"Allianz Best Styles Global Equity Cl ET Acc H2-SGD","BK4587":"ChatGPT概念","BK4543":"AI","LU2237443622.USD":"Aberdeen Standard SICAV I - Global Dynamic Dividend A Acc USD","LU0061474705.USD":"THREADNEEDLE (LUX) GLOBAL DYNAMIC REAL RETURN \"AU\" (USD) ACC","LU0648000940.SGD":"Natixis Harris Associates Global Equity RA SGD","BK4588":"碎股","IE00BJJMRX11.SGD":"Janus Henderson Balanced A Acc SGD","LU1066051498.USD":"HSBC GIF GLOBAL EQUITY VOLATILITY FOCUSED \"AM2\" (USD) INC","LU1066053197.SGD":"HSBC GIF GLOBAL EQUITY VOLATILITY FOCUSED \"AM3\" (SGDHDG) INC","SG9999014898.SGD":"United Global Quality Growth Fund Dis SGD","IE00B3S45H60.SGD":"Neuberger Berman US Multicap Opportunities A Acc SGD-H","SG9999018857.SGD":"United Global Quality Growth Fd Cl Acc SGD-H","LU0672654240.SGD":"FTIF - Franklin US Opportunities A Acc SGD-H1","LU1280957306.USD":"THREADNEEDLE (LUX) US CONTRARIAN CORE EQUITIES \"AUP\" (USD) INC","BK4561":"索罗斯持仓","LU1046421795.USD":"富达环球科技A-ACC","LU1720051108.HKD":"ALLIANZ GLOBAL ARTIFICIAL INTELLIGENCE \"AT\" (HKD) ACC","LU0276348264.USD":"THREADNEEDLE (LUX) GLOBAL DYNAMIC REAL RETURN\"AUP\" (USD) INC","LU1548497426.USD":"安联环球人工智能AT Acc","IE00BFSS8Q28.SGD":"Janus Henderson Balanced A Inc SGD-H","LU0820561818.USD":"安联收益及增长平衡基金Cl AM DIS","LU1951198990.SGD":"Natixis Thematics AI & Robotics Fund H-R/A SGD-H","LU0354030511.USD":"ALLSPRING U.S. LARGE CAP GROWTH \"I\" (USD) ACC","LU0957791311.USD":"THREADNEEDLE (LUX) GLOBAL FOCUS \"ZU\" (USD) ACC","LU1201861249.SGD":"Natixis Harris Associates US Equity PA SGD-H","IE0009356076.USD":"JANUS HENDERSON GLOBAL TECHNOLOGY AND INNOVATION \"A2\" (USD) ACC","SG9999017495.SGD":"UGDP UNITED GLOBAL QUALITY GROWTH \"B\" (SGD) ACC","LU0557290698.USD":"施罗德环球可持续增长基金","NVDA":"英伟达","LU2125909247.SGD":"Natixis Thematics Meta H-R/A SGD","LU0061474960.USD":"天利环球焦点基金AU Acc","LU0170899867.USD":"EASTSPRING INVESTMENTS WORLD VALUE EQUITY \"A\" (USD) ACC","LU2087621335.USD":"ALLSPRING GLOBAL FACTOR ENHANCED EQUITY \"A\" (USD) ACC","LU0456855351.SGD":"JPMorgan Funds - Global Equity A (acc) SGD","LU1435385759.SGD":"Natixis Loomis Sayles US Growth Equity RA SGD-H","LU2237443382.USD":"Aberdeen Standard SICAV I - Global Dynamic Dividend A MIncA USD","SG9999001077.SGD":"United International Growth Fund SGD","GB00BDT5M118.USD":"天利环球扩展Alpha基金A Acc","LU0889565833.HKD":"FRANKLIN TECHNOLOGY \"A\" (HKD) ACC","LU0082616367.USD":"摩根大通美国科技A(dist)","BK4567":"ESG概念","BK4534":"瑞士信贷持仓","BK4585":"ETF&股票定投概念","IE0004445015.USD":"JANUS HENDERSON BALANCED \"A2\" (USD) ACC","LU1623119135.USD":"Natixis Mirova Global Sustainable Equity R-NPF/A USD","LU1803068979.SGD":"FTIF - Franklin Technology A (acc) SGD-H1","LU1242518857.USD":"FULLERTON LUX FUNDS - ASIA ABSOLUTE ALPHA \"I\" (USD) ACC","LU1712237335.SGD":"Natixis Mirova Global Sustainable Equity H-R-NPF/A SGD"},"source_url":"https://finance.sina.com.cn/stock/usstock/c/2023-04-05/doc-imypimne9357334.shtml","is_english":false,"share_image_url":"https://static.laohu8.com/b0d1b7e8843deea78cc308b15114de44","article_id":"2325368910","content_text":"谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作\n 与英伟达的GPU相比,谷歌TPU采用低精度计算,几乎不影响深度学习处理效果的前提,比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍。\n 我们还没有看到能与 ChatGPT 相匹敌的 AI 大模型,但在算力基础上,领先的可能并不是微软和 OpenAI。\n 本周二,谷歌公布了其训练语言大模型的超级计算机的细节,基于 TPU 的超算系统已经可以比英伟达的同类更加快速、节能。\n 谷歌张量处理器(tensor processing unit,TPU)是该公司为机器学习定制的专用芯片(ASIC),第一代发布于 2016 年,成为了 AlphaGo 背后的算力。与 GPU 相比,TPU采用低精度计算,在几乎不影响深度学习处理效果的前提下大幅降低了功耗、加快运算速度。同时,TPU 使用了脉动阵列等设计来优化矩阵乘法与卷积运算。\n 当前,谷歌 90% 以上的人工智能训练工作都在使用这些芯片,TPU 支撑了包括搜索的谷歌主要业务。作为图灵奖得主、计算机架构巨擘,大卫・帕特森(David Patterson)在 2016 年从 UC Berkeley 退休后,以杰出工程师的身份加入了谷歌大脑团队,为几代 TPU 的研发做出了卓越贡献。\n\n 如今 TPU 已经发展到了第四代,谷歌本周二由 Norman Jouppi、大卫・帕特森等人发表的论文《 TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings 》详细介绍了自研的光通信器件是如何将 4000 多块芯片并联成为超级计算机,以提升整体效率的。\n TPU v4 的性能比 TPU v3 高 2.1 倍,性能功耗比提高 2.7 倍。基于 TPU v4 的超级计算机拥有 4096 块芯片,整体速度提高了约 10 倍。对于类似大小的系统,谷歌能做到比 Graphcore IPU Bow 快 4.3-4.5 倍,比 Nvidia A100 快 1.2-1.7 倍,功耗低 1.3-1.9 倍。\n 除了芯片本身的算力,芯片间互联已成为构建 AI 超算的公司之间竞争的关键点,最近一段时间,谷歌的 Bard、OpenAI 的 ChatGPT 这样的大语言模型(LLM)规模正在爆炸式增长,算力已经成为明显的瓶颈。\n 由于大模型动辄千亿的参数量,它们必须由数千块芯片共同分担,并持续数周或更长时间进行训练。谷歌的 PaLM 模型 —— 其迄今为止最大的公开披露的语言模型 —— 在训练时被拆分到了两个拥有 4000 块 TPU 芯片的超级计算机上,用时 50 天。\n 谷歌表示,通过光电路交换机(OCS),其超级计算机可以轻松地动态重新配置芯片之间的连接,有助于避免出现问题并实时调整以提高性能。\n 下图展示了 TPU v4 4×3 方式 6 个‘面’的链接。每个面有 16 条链路,每个块总共有 96 条光链路连接到 OCS 上。要提供 3D 环面的环绕链接,相对侧的链接必须连接到相同的 OCS。因此,每个 4×3 块 TPU 连接到 6 × 16 ÷ 2 = 48 个 OCS 上。Palomar OCS 为 136×136(128 个端口加上 8 个用于链路测试和修复的备用端口),因此 48 个 OCS 连接来自 64 个 4×3 块(每个 64 个芯片)的 48 对电缆,总共并联 4096 个 TPU v4 芯片。\n 根据这样的排布,TPU v4(中间的 ASIC 加上 4 个 HBM 堆栈)和带有 4 个液冷封装的印刷电路板 (PCB)。该板的前面板有 4 个顶部 PCIe 连接器和 16 个底部 OSFP 连接器,用于托盘间 ICI 链接。\n 随后,八个 64 芯片机架构成一台 4096 芯片超算。\n 与超级计算机一样,工作负载由不同规模的算力承担,称为切片:64 芯片、128 芯片、256 芯片等。下图显示了当主机可用性从 99.0% 到 99.9% 不等有,及没有 OCS 时切片大小的‘有效输出’。如果没有 OCS,主机可用性必须达到 99.9% 才能提供合理的切片吞吐量。对于大多数切片大小,OCS 也有 99.0% 和 99.5% 的良好输出。\n\n 与 Infiniband 相比,OCS 的成本更低、功耗更低、速度更快,成本不到系统成本的 5%,功率不到系统功率的 3%。每个 TPU v4 都包含 SparseCores 数据流处理器,可将依赖嵌入的模型加速 5 至 7 倍,但仅使用 5% 的裸片面积和功耗。\n ‘这种切换机制使得绕过故障组件变得容易,’谷歌研究员 Norm Jouppi 和谷歌杰出工程师大卫・帕特森在一篇关于该系统的博客文章中写道。‘这种灵活性甚至允许我们改变超级计算机互连的拓扑结构,以加速机器学习模型的性能。’\n 在新论文上,谷歌着重介绍了稀疏核(SparseCore,SC)的设计。在大模型的训练阶段,embedding 可以放在 TensorCore 或超级计算机的主机 CPU 上处理。TensorCore 具有宽 VPU 和矩阵单元,并针对密集操作进行了优化。由于小的聚集 / 分散内存访问和可变长度数据交换,在 TensorCore 上放置嵌入其实并不是最佳选择。在超级计算机的主机 CPU 上放置嵌入会在 CPU DRAM 接口上引发阿姆达尔定律瓶颈,并通过 4:1 TPU v4 与 CPU 主机比率放大。数据中心网络的尾部延迟和带宽限制将进一步限制训练系统。\n 对此,谷歌认为可以使用 TPU 超算的总 HBM 容量优化性能,加入专用 ICI 网络,并提供快速收集 / 分散内存访问支持。这导致了 SparseCore 的协同设计。\n SC 是一种用于嵌入训练的特定领域架构,从 TPU v2 开始,后来在 TPU v3 和 TPU v4 中得到改进。SC 相对划算,只有芯片面积的约 5% 和功率的 5% 左右。SC 结合超算规模的 HBM 和 ICI 来创建一个平坦的、全局可寻址的内存空间(TPU v4 中为 128 TiB)。与密集训练中大参数张量的全部归约相比,较小嵌入向量的全部传输使用 HBM 和 ICI 以及更细粒度的分散 / 聚集访问模式。\n 作为独立的核心,SC 允许跨密集计算、SC 和 ICI 通信进行并行化。下图显示了 SC 框图,谷歌将其视为‘数据流’架构(dataflow),因为数据从内存流向各种直接连接的专用计算单元。\n 最通用的 SC 单元是 16 个计算块(深蓝色框)。每个 tile 都有一个关联的 HBM 通道,并支持多个未完成的内存访问。每个 tile 都有一个 Fetch Unit、一个可编程的 8-wide SIMD Vector Processing Unit 和一个 Flush Unit。获取单元将 HBM 中的激活和参数读取到 2.5 MiB 稀疏向量内存 (Spmem) 的图块切片中。scVPU 使用与 TC 的 VPU 相同的 ALU。Flush Unit 在向后传递期间将更新的参数写入 HBM。此外,五个跨通道单元(金色框)执行特定的嵌入操作,正如它们的名称所解释的那样。\n 与 TPU v1 一样,这些单元执行类似 CISC 的指令并对可变长度输入进行操作,其中每条指令的运行时间都取决于数据。\n 在特定芯片数量下,TPU v3/v4 对分带宽比高 2-4 倍,嵌入速度可以提高 1.1-2.0 倍。\n 下图展示了谷歌自用的推荐模型(DLRM0)在不同芯片上的效率。TPU v3 比 CPU 快 9.8 倍。TPU v4 比 TPU v3 高 3.1 倍,比 CPU 高 30.1 倍。\n 谷歌探索了 TPU v4 超算用于 GPT-3 大语言模型时的性能,展示了预训练阶段专家设计的 1.2 倍改进。\n\n 虽然谷歌直到现在才公布有关其超级计算机的详细信息,但自 2020 年以来,基于 TPU 的 AI 超算一直在位于俄克拉荷马州的数据中心发挥作用。谷歌表示,Midjourney 一直在使用该系统训练其模型,最近一段时间,后者已经成为 AI 画图领域最热门的平台。\n\n 谷歌在论文中表示,对于同等大小的系统,其芯片比基于英伟达 A100 芯片的系统快 1.7 倍,能效高 1.9 倍,后者与第四代 TPU 同时上市,并被用于 GPT-4 的训练。\n 对此,英伟达发言人拒绝置评。\n 当前英伟达的 AI 芯片已经进入 Hopper 架构的时代。谷歌表示,未对第四代 TPU 与英伟达目前的旗舰 H100 芯片进行比较,因为 H100 在谷歌芯片之后上市,并且采用了更先进的制程。\n 但同样在此,谷歌暗示了下一代 TPU 的计划,其没有提供更多细节。Jouppi 告诉路透社,谷歌拥有开发‘未来芯片的健康管道’。\n\n TPU v4 比当代 DSA 芯片速度更快、功耗更低,如果考虑到互连技术,功率边缘可能会更大。通过使用具有 3D 环面拓扑的 3K TPU v4 切片,与 TPU v3 相比,谷歌的超算也能让 LLM 的训练时间大大减少。\n 性能、可扩展性和可用性使 TPU v4 超级计算机成为 LaMDA、MUM 和 PaLM 等大型语言模型 (LLM) 的主要算力。这些功能使 5400 亿参数的 PaLM 模型在 TPU v4 超算上进行训练时,能够在 50 天内维持 57.8% 的峰值硬件浮点性能。\n 谷歌表示,其已经部署了数十台 TPU v4 超级计算机,供内部使用和外部通过谷歌云使用。\n 本文作者:泽南,来源:机器之心,原文标题:《谷歌TPU超算,大模型性能超英伟达,已部署数十台:图灵奖得主新作》\n\n炒股开户享福利,送投顾服务60天体验权,一对一指导服务!\n\n\n\n\n海量资讯、精准解读,尽在新浪财经APP\n\n责任编辑:郭明煜","news_type":1},"isVote":1,"tweetType":1,"viewCount":1365,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":627249556,"gmtCreate":1678373866375,"gmtModify":1678373868089,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"与Facebook无关,为什么要挂在这里呢?","listText":"与Facebook无关,为什么要挂在这里呢?","text":"与Facebook无关,为什么要挂在这里呢?","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/627249556","repostId":"2318267778","repostType":2,"repost":{"id":"2318267778","pubTimestamp":1678368190,"share":"https://www.laohu8.com/m/news/2318267778?lang=&edition=full","pubTime":"2023-03-09 21:23","market":"us","language":"zh","title":"美股回购规模创纪录,全靠这5家公司","url":"https://stock-news.laohu8.com/highlight/detail?id=2318267778","media":"智通财经","summary":"摩根大通策略师表示,今年美国企业宣布的回购计划正以创纪录的速度进行,不过在2610亿美元的回购计划中,超过三分之二的回购计划仅分布在五家公司。Dubravko Lakos-Bujas领导的团队在一份报","content":"<html><head></head><body><p><a href=\"https://laohu8.com/S/JPM\">摩根大通</a>策略师表示,今年美国企业宣布的回购计划正以创纪录的速度进行,不过在2610亿美元的回购计划中,超过三分之二的回购计划仅分布在五家公司。</p><p>Dubravko Lakos-Bujas领导的团队在一份报告中写道,<a href=\"https://laohu8.com/S/CVX\">雪佛龙</a>以750亿美元的回购额领先,其次是Meta(400亿美元)、<a href=\"https://laohu8.com/S/GS\">高盛</a>(300亿美元)和<a href=\"https://laohu8.com/S/BKNG\">Booking Holdings</a>和<a href=\"https://laohu8.com/S/CRM\">赛富时</a>(各200亿美元)。</p><p><img src=\"https://static.tigerbbs.com/71f72a76205cbdde2d9cef53e826b138\" tg-width=\"528\" tg-height=\"363\" referrerpolicy=\"no-referrer\" width=\"100%\" height=\"auto\"/></p><p>不过,该行策略师指出,尽管今年迄今为止宣布的回购计划屡创新高,但执行数量一直在下降。去年第四季度股票回购减少了20%,自去年第一季度以来,回购速度有所放缓。</p><p>过去十年,回购一直是美国股票需求的一个关键来源。但鉴于当前经济增长的不确定性,同时回购成为了拜登政府增税的目标,企业可能会寻求保留现金。</p><p>摩根大通策略师表示,过去12个月,标普500指数成份股公司宣布的股票回购总额为8530亿美元,相比之下,2019年5月的峰值约为1万亿美元。</p><p><img src=\"https://static.tigerbbs.com/3fc42d54c2a592cf08658b947d95f178\" tg-width=\"530\" tg-height=\"364\" referrerpolicy=\"no-referrer\" width=\"100%\" height=\"auto\"/></p><p><b>其他观点</b></p><p>另外,摩根大通还对美国上市公司的第四季度业绩表现发表了观点。该行策略师认为,美国第四季度财报季并没有想象中的那么糟糕,但证实了企业基本面正在出现裂痕。</p><p>标普500指数成分股公司中,有64%的公司的第四季度业绩超出了预期,低于过去四季度70%的平均水平。</p><p>报告还指出,每股收益继续向下修正,2023年每股收益向下修正8美元至222美元。</p><p>该行策略师认为,下半年的盈利预期仍然过高,利润率上升的预期与劳动力成本的粘性、资本成本压力的上升以及需求放缓的风险相矛盾。</p></body></html>","source":"stock_zhitongcaijing","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>美股回购规模创纪录,全靠这5家公司</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美股回购规模创纪录,全靠这5家公司\n</h2>\n\n<h4 class=\"meta\">\n\n\n2023-03-09 21:23 北京时间 <a href=http://www.zhitongcaijing.com/content/detail/889419.html><strong>智通财经</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>摩根大通策略师表示,今年美国企业宣布的回购计划正以创纪录的速度进行,不过在2610亿美元的回购计划中,超过三分之二的回购计划仅分布在五家公司。Dubravko Lakos-Bujas领导的团队在一份报告中写道,雪佛龙以750亿美元的回购额领先,其次是Meta(400亿美元)、高盛(300亿美元)和Booking Holdings和赛富时(各200亿美元)。不过,该行策略师指出,尽管今年迄今为止宣布...</p>\n\n<a href=\"http://www.zhitongcaijing.com/content/detail/889419.html\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/c318bcd91a109139b7d70c76c30bb154","relate_stocks":{"CVX":"雪佛龙","CRM":"赛富时","BKNG":"Booking Holdings","META":"Meta Platforms, Inc.","GS":"高盛"},"source_url":"http://www.zhitongcaijing.com/content/detail/889419.html","is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"2318267778","content_text":"摩根大通策略师表示,今年美国企业宣布的回购计划正以创纪录的速度进行,不过在2610亿美元的回购计划中,超过三分之二的回购计划仅分布在五家公司。Dubravko Lakos-Bujas领导的团队在一份报告中写道,雪佛龙以750亿美元的回购额领先,其次是Meta(400亿美元)、高盛(300亿美元)和Booking Holdings和赛富时(各200亿美元)。不过,该行策略师指出,尽管今年迄今为止宣布的回购计划屡创新高,但执行数量一直在下降。去年第四季度股票回购减少了20%,自去年第一季度以来,回购速度有所放缓。过去十年,回购一直是美国股票需求的一个关键来源。但鉴于当前经济增长的不确定性,同时回购成为了拜登政府增税的目标,企业可能会寻求保留现金。摩根大通策略师表示,过去12个月,标普500指数成份股公司宣布的股票回购总额为8530亿美元,相比之下,2019年5月的峰值约为1万亿美元。其他观点另外,摩根大通还对美国上市公司的第四季度业绩表现发表了观点。该行策略师认为,美国第四季度财报季并没有想象中的那么糟糕,但证实了企业基本面正在出现裂痕。标普500指数成分股公司中,有64%的公司的第四季度业绩超出了预期,低于过去四季度70%的平均水平。报告还指出,每股收益继续向下修正,2023年每股收益向下修正8美元至222美元。该行策略师认为,下半年的盈利预期仍然过高,利润率上升的预期与劳动力成本的粘性、资本成本压力的上升以及需求放缓的风险相矛盾。","news_type":1},"isVote":1,"tweetType":1,"viewCount":1279,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":172137108,"gmtCreate":1626943379204,"gmtModify":1626944287942,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"title":"2021.07.22 复盘笔记 by 凯瑞昂","htmlText":"2021-07-22 成本 现价 浮动收益 老虎证券(TIGR) 16.98 20.15 +18.65% 买入时机要求再严格一些,终于等来了一个不错的位置。","listText":"2021-07-22 成本 现价 浮动收益 老虎证券(TIGR) 16.98 20.15 +18.65% 买入时机要求再严格一些,终于等来了一个不错的位置。","text":"2021-07-22 成本 现价 浮动收益 老虎证券(TIGR) 16.98 20.15 +18.65% 买入时机要求再严格一些,终于等来了一个不错的位置。","images":[],"top":1,"highlighted":1,"essential":1,"paper":2,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/172137108","isVote":1,"tweetType":1,"viewCount":1615,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":121867542,"gmtCreate":1624459116886,"gmtModify":1624459116886,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"刚下载了APP去储值,参加活动。发现竟然还有几块钱的余额,你看看,感觉盈利多了几块钱。这活动好。","listText":"刚下载了APP去储值,参加活动。发现竟然还有几块钱的余额,你看看,感觉盈利多了几块钱。这活动好。","text":"刚下载了APP去储值,参加活动。发现竟然还有几块钱的余额,你看看,感觉盈利多了几块钱。这活动好。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/121867542","isVote":1,"tweetType":1,"viewCount":1800,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":121865686,"gmtCreate":1624459041445,"gmtModify":1624459041445,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"申购后发现了活动页,然后发现可用资金不足了,下次要先看活动再出手。","listText":"申购后发现了活动页,然后发现可用资金不足了,下次要先看活动再出手。","text":"申购后发现了活动页,然后发现可用资金不足了,下次要先看活动再出手。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/121865686","isVote":1,"tweetType":1,"viewCount":906,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":187800881,"gmtCreate":1623748164343,"gmtModify":1623748164343,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"看起来手机成色不错的样子,爱回收的服务也很棒,接触过几次,工作人员都会引导使用优惠券。相信会大涨。","listText":"看起来手机成色不错的样子,爱回收的服务也很棒,接触过几次,工作人员都会引导使用优惠券。相信会大涨。","text":"看起来手机成色不错的样子,爱回收的服务也很棒,接触过几次,工作人员都会引导使用优惠券。相信会大涨。","images":[{"img":"https://static.tigerbbs.com/749c8456a0ae384685226d95d1b8b25d","width":"1125","height":"2436"}],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/187800881","isVote":1,"tweetType":1,"viewCount":1476,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":1,"langContent":"CN","totalScore":0},{"id":100281410,"gmtCreate":1619617223930,"gmtModify":1619617223930,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"怪不得","listText":"怪不得","text":"怪不得","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/100281410","repostId":"2130308159","repostType":2,"repost":{"id":"2130308159","pubTimestamp":1619613300,"share":"https://www.laohu8.com/m/news/2130308159?lang=&edition=full","pubTime":"2021-04-28 20:35","market":"hk","language":"zh","title":"特斯拉已完全偿还上海超级工厂6.14亿美元贷款","url":"https://stock-news.laohu8.com/highlight/detail?id=2130308159","media":"e公司","summary":"【特斯拉已完全偿还上海超级工厂6.14亿美元贷款】4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止...","content":"<html><body><div>\n<p>\r\n 原标题:<a href=\"https://laohu8.com/S/TSLA\">特斯拉</a>已完全偿还上海超级工厂6.14亿美元贷款\r\n </p>\n<div>摘要</div>\n<div>【特斯拉已完全偿还上海超级工厂6.14亿美元贷款】4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。(e公司)</div>\n<p><img border=\"0\" height=\"276\" src=\"https://webquoteklinepic.eastmoney.com/GetPic.aspx?nid=105.TSLA&imageType=k&token=28dfeb41d35cc81d84b4664d7c23c49f&at=1\" width=\"578\"/></p><p> 4月28日晚间,根据<span>特斯拉</span><span></span>向美国证券交易委员会递交的文件,<span href=\"http://quote.eastmoney.com/unify/r/105.TSLA\" target=\"_blank\" web=\"1\">特斯拉</span>披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款<span>合同</span>已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。</p><p>(文章来源:e公司)</p>\n<p>\r\n (责任编辑:DF537)\r\n </p>\n</div></body></html>","source":"stock_eastmoney","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>特斯拉已完全偿还上海超级工厂6.14亿美元贷款</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特斯拉已完全偿还上海超级工厂6.14亿美元贷款\n</h2>\n\n<h4 class=\"meta\">\n\n\n2021-04-28 20:35 北京时间 <a href=http://finance.eastmoney.com/a/202104281905257786.html><strong>e公司</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>原标题:特斯拉已完全偿还上海超级工厂6.14亿美元贷款\r\n \n摘要\n【特斯拉已完全偿还上海超级工厂6.14亿美元贷款】4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。(e公司)\n ...</p>\n\n<a href=\"http://finance.eastmoney.com/a/202104281905257786.html\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/8350896f4f33c86bc28f200b67ab82b4","relate_stocks":{"TSLA":"特斯拉"},"source_url":"http://finance.eastmoney.com/a/202104281905257786.html","is_english":false,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"2130308159","content_text":"原标题:特斯拉已完全偿还上海超级工厂6.14亿美元贷款\r\n \n摘要\n【特斯拉已完全偿还上海超级工厂6.14亿美元贷款】4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。(e公司)\n 4月28日晚间,根据特斯拉向美国证券交易委员会递交的文件,特斯拉披露称公司已完全偿还与上海超级工厂支出相关的6.14亿美元贷款,相关的贷款合同已经终止。该合同终止后,公司债务和融资租赁表中包含的固定资产授信项下未使用的7.58亿美元将不可再用。(文章来源:e公司)\n\r\n (责任编辑:DF537)","news_type":1},"isVote":1,"tweetType":1,"viewCount":1602,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":374543326,"gmtCreate":1619468373965,"gmtModify":1619468373965,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"多打了个万吧?","listText":"多打了个万吧?","text":"多打了个万吧?","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/374543326","repostId":"2130343683","repostType":2,"repost":{"id":"2130343683","weMediaInfo":{"introduction":"Stock Market Quotes, Business News, Financial News, Trading Ideas, and Stock Research by Professionals","home_visible":0,"media_name":"Benzinga","id":"1052270027","head_image":"https://static.tigerbbs.com/d08bf7808052c0ca9deb4e944cae32aa"},"pubTimestamp":1619467700,"share":"https://www.laohu8.com/m/news/2130343683?lang=&edition=full","pubTime":"2021-04-27 04:08","market":"us","language":"en","title":"Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY","url":"https://stock-news.laohu8.com/highlight/detail?id=2130343683","media":"Benzinga","summary":"Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY","content":"<html><body><p>Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY</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>Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY</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\">\nTesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY\n</h2>\n\n<h4 class=\"meta\">\n\n\n<div class=\"head\" \">\n\n\n<div class=\"h-thumb\" style=\"background-image:url(https://static.tigerbbs.com/d08bf7808052c0ca9deb4e944cae32aa);background-size:cover;\"></div>\n\n<div class=\"h-content\">\n<p class=\"h-name\">Benzinga </p>\n<p class=\"h-time\">2021-04-27 04:08</p>\n</div>\n\n</div>\n\n\n</h4>\n\n</header>\n<article>\n<html><body><p>Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY</p></body></html>\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"","relate_stocks":{"TSLA":"特斯拉"},"source_url":"https://www.benzinga.com/node/20799832","is_english":true,"share_image_url":"https://static.laohu8.com/e9f99090a1c2ed51c021029395664489","article_id":"2130343683","content_text":"Tesla Q1 Deliveries 184.877K, Up 109% Year Over Year, Production 180.338K, Up 76% YoY","news_type":1},"isVote":1,"tweetType":1,"viewCount":652,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":376936429,"gmtCreate":1619078676061,"gmtModify":1619078676061,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"htmlText":"会不会是玩游戏玩的兴奋了一跺脚,踩到油门了。","listText":"会不会是玩游戏玩的兴奋了一跺脚,踩到油门了。","text":"会不会是玩游戏玩的兴奋了一跺脚,踩到油门了。","images":[],"top":1,"highlighted":1,"essential":1,"paper":1,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/376936429","repostId":"2129555388","repostType":2,"repost":{"id":"2129555388","pubTimestamp":1619077800,"share":"https://www.laohu8.com/m/news/2129555388?lang=&edition=full","pubTime":"2021-04-22 15:50","market":"us","language":"zh","title":"女车主自称特斯拉失控自动倒车!地库被撞出一个大洞","url":"https://stock-news.laohu8.com/highlight/detail?id=2129555388","media":"快科技","summary":"而今,郑州又有一位女车主曝料称,自己的特斯拉车辆突然无故倒车,把地库的墙也撞了一个大洞,车辆也严重受损。此外,对于荣女士所称的,事故时一分钟左右的行车记录仪视频缺失,和满屏错误代码消失的问题,工作人员表示,特斯拉是一个负责任的企业,不可能去修改、篡改、删除客户任何的数据,这是特斯拉在中国市场的一个基本底线。","content":"<html><body><article><p>原标题:女车主自称<a href=\"https://laohu8.com/S/TSLA\">特斯拉</a>失控自动倒车!地库被撞出一个大洞</p><img src=\"https://fid-75186.picgzc.qpic.cn/20210422155358144v1731ffqjnrv2n9\"/><p>近日,特斯拉在国内已经成为了车圈热点顶流,一边是特斯拉不断致歉,希望第三方进行检测证明清白,另一方面,是越来越多的车主质疑特斯拉“失控”事故。</p><p>而今,郑州又有一位女车主曝料称,<strong>自己的特斯拉车辆突然无故倒车,把地库的墙也撞了一个大洞,车辆也严重受损。</strong>更诡异的是,事故发生时的行车记录仪视频也不翼而飞。</p><p>据媒体报道,今年3月29日下午,荣女士在小区车库停车时,突然遭遇失控倒车,荣女士称自己当时在等其他车辆挪位置,就在车里玩了会儿王者荣耀,谁知车辆突然失控,一下子撞到墙上。</p><p>视频</p><p>荣女士称虽然当时挂的是R档,但是处于驻车状况,自己并没有踩油门。而且令荣女士更加不解的是,当时满屏的错误代码后来变成了三行。<strong>并且,行车记录仪中也缺失了出事故时一分钟左右的记录,自己在电脑上恢复怎么都不能恢复成功。</strong></p><p>带着这些疑问,荣女士和记者一同来到了郑州特斯拉中原福塔店,特斯拉工作人员向他们表示,<strong>后台数据显示,在事故发生时,系统检测到加速踏板电门被深踩。</strong>因此,是驾驶员误踩电门导致的事故。</p><p>但是,对于这样的回应,荣女士并不满意,她坚称自己没有误踩电门,同时向特斯拉方面要求提供后台数据,让她自己亲眼查看。但是,特斯拉员工又称,这些数据是被加密存储在服务器端,他们也没有权限查看。</p><p>此外,对于荣女士所称的,事故时一分钟左右的行车记录仪视频缺失,和满屏错误代码消失的问题,工作人员表示,特斯拉是一个负责任的企业,不可能去修改、篡改、删除客户任何的数据,这是特斯拉在中国市场的一个基本底线。</p><p>但是,面对记者的提问“那为何行车记录仪视频真实消失了呢?”,<strong>特斯拉工作员工只能表示,这个现象确实没有见到,因此无法回复。</strong></p><p>当前,荣女士和特斯拉方间的协商还在进行当中,对此,我们也会保持关注。</p><p>(文章来源:快科技)</p></article></body></html>","source":"tencent","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>女车主自称特斯拉失控自动倒车!地库被撞出一个大洞</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女车主自称特斯拉失控自动倒车!地库被撞出一个大洞\n</h2>\n\n<h4 class=\"meta\">\n\n\n2021-04-22 15:50 北京时间 <a href=http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN202104221553587dd3279e&s=b><strong>快科技</strong></a>\n\n\n</h4>\n\n</header>\n<article>\n<div>\n<p>原标题:女车主自称特斯拉失控自动倒车!地库被撞出一个大洞近日,特斯拉在国内已经成为了车圈热点顶流,一边是特斯拉不断致歉,希望第三方进行检测证明清白,另一方面,是越来越多的车主质疑特斯拉“失控”事故。而今,郑州又有一位女车主曝料称,自己的特斯拉车辆突然无故倒车,把地库的墙也撞了一个大洞,车辆也严重受损。更诡异的是,事故发生时的行车记录仪视频也不翼而飞。据媒体报道,今年3月29日下午,荣女士在小区车库...</p>\n\n<a href=\"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN202104221553587dd3279e&s=b\">Web Link</a>\n\n</div>\n\n\n</article>\n</div>\n</body>\n</html>\n","type":0,"thumbnail":"https://static.tigerbbs.com/9749762c5ee11a52030b4c66a89c8026","relate_stocks":{"TSLA":"特斯拉"},"source_url":"http://gu.qq.com/resources/shy/news/detail-v2/index.html#/?id=nesSN202104221553587dd3279e&s=b","is_english":false,"share_image_url":"https://static.laohu8.com/9a95c1376e76363c1401fee7d3717173","article_id":"2129555388","content_text":"原标题:女车主自称特斯拉失控自动倒车!地库被撞出一个大洞近日,特斯拉在国内已经成为了车圈热点顶流,一边是特斯拉不断致歉,希望第三方进行检测证明清白,另一方面,是越来越多的车主质疑特斯拉“失控”事故。而今,郑州又有一位女车主曝料称,自己的特斯拉车辆突然无故倒车,把地库的墙也撞了一个大洞,车辆也严重受损。更诡异的是,事故发生时的行车记录仪视频也不翼而飞。据媒体报道,今年3月29日下午,荣女士在小区车库停车时,突然遭遇失控倒车,荣女士称自己当时在等其他车辆挪位置,就在车里玩了会儿王者荣耀,谁知车辆突然失控,一下子撞到墙上。视频荣女士称虽然当时挂的是R档,但是处于驻车状况,自己并没有踩油门。而且令荣女士更加不解的是,当时满屏的错误代码后来变成了三行。并且,行车记录仪中也缺失了出事故时一分钟左右的记录,自己在电脑上恢复怎么都不能恢复成功。带着这些疑问,荣女士和记者一同来到了郑州特斯拉中原福塔店,特斯拉工作人员向他们表示,后台数据显示,在事故发生时,系统检测到加速踏板电门被深踩。因此,是驾驶员误踩电门导致的事故。但是,对于这样的回应,荣女士并不满意,她坚称自己没有误踩电门,同时向特斯拉方面要求提供后台数据,让她自己亲眼查看。但是,特斯拉员工又称,这些数据是被加密存储在服务器端,他们也没有权限查看。此外,对于荣女士所称的,事故时一分钟左右的行车记录仪视频缺失,和满屏错误代码消失的问题,工作人员表示,特斯拉是一个负责任的企业,不可能去修改、篡改、删除客户任何的数据,这是特斯拉在中国市场的一个基本底线。但是,面对记者的提问“那为何行车记录仪视频真实消失了呢?”,特斯拉工作员工只能表示,这个现象确实没有见到,因此无法回复。当前,荣女士和特斯拉方间的协商还在进行当中,对此,我们也会保持关注。(文章来源:快科技)","news_type":1},"isVote":1,"tweetType":1,"viewCount":2331,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0},{"id":315116420,"gmtCreate":1612220296360,"gmtModify":1703758921666,"author":{"id":"3453412771192240","authorId":"3453412771192240","name":"凯瑞昂","avatar":"https://static.tigerbbs.com/8d3c623b92b9d36252abd7d1fa6b3aba","crmLevel":2,"crmLevelSwitch":0,"followedFlag":false,"idStr":"3453412771192240","authorIdStr":"3453412771192240"},"themes":[],"title":"2021.02.02 复盘笔记 by 凯瑞昂","htmlText":"2021-02-02 成本 现价 浮动收益 声网(API) 41.42 74.50 +81.08% 有时候,炒股是需要一些运气的。大涨追涨买入了几天,赶上风头正热的clubhouse了。","listText":"2021-02-02 成本 现价 浮动收益 声网(API) 41.42 74.50 +81.08% 有时候,炒股是需要一些运气的。大涨追涨买入了几天,赶上风头正热的clubhouse了。","text":"2021-02-02 成本 现价 浮动收益 声网(API) 41.42 74.50 +81.08% 有时候,炒股是需要一些运气的。大涨追涨买入了几天,赶上风头正热的clubhouse了。","images":[],"top":1,"highlighted":1,"essential":1,"paper":2,"likeSize":0,"commentSize":0,"repostSize":0,"link":"https://laohu8.com/post/315116420","isVote":1,"tweetType":1,"viewCount":1617,"authorTweetTopStatus":1,"verified":2,"comments":[],"imageCount":0,"langContent":"CN","totalScore":0}],"lives":[]}