#dojo supercomputer

[libertyracer]

Our dojo needed a supercomputer to answer the question of life, universe and everything.

@kingofcaptura

@meleeprowess

[dahlbrendan]

Tesla has a not-so-secret plan to replace a fundamental-but-flawed building block of the Internet — Musk's Exascale DOJO supercomputer will use TTPoE instead of the universal TCP and wants the world to embrace it | TechRadar

[isfeed]

WeRide preps for an IPO, meet the man who built a startup pipeline at CNH and Waymo’s nightly honk-a-thon

Welcome back to TechCrunch Mobility — your central hub for news and insights on the future of transportation. Sign up here for free — just click TechCrunch Mobility! We brought you a Tesla Dojo explainer last week. Now we’re giving you a handy Dojo timeline that tracks the progress of Tesla’s AI supercomputer. The timeline […] © 2024 TechCrunch. All rights reserved. For personal use only. Source:…

[y2fear]

What went wrong at Cruise, a pivot at Vroom and a home for Tesla's Dojo supercomputer

[eclectic-insights]

What went wrong at Cruise, a pivot at Vroom and a home for Tesla’s Dojo supercomputer http://dlvr.it/T20WdK

[worldmarketus]

Tesla’s Market Value Soars on Dojo Supercomputer Estimates $500 Billion

Tesla Inc. is a towering figure in the ever-changing landscape of electric vehicles (EVs), constantly pushing boundaries and redefining the automotive industry. Tesla recently took another giant step forward with its Dojo supercomputer, generating interest in both the tech and financial sectors. According to Morgan Stanley analysts, this innovative technology has the potential to increase Tesla’s market value by $600 billion. Let’s get into the specifics of this game-changing development.

Tesla’s Market Value Soars on Dojo Supercomputer Estimates $500 Billion

Dojo, Tesla’s cutting-edge supercomputer, went into production in July with the primary goal of speeding up the training of artificial intelligence (AI) models for self-driving cars. With over $1 billion earmarked for its development by next year, this supercomputer has the potential to be a game changer in the field of autonomous driving.

Dojo’s ability to process massive amounts of data at unprecedented speeds is one of its key advantages. This capability is expected to significantly improve Tesla’s self-driving technology, making it safer and more dependable. As a result, Tesla will be able to accelerate its entry into the worlds of robotaxis and software services.

Unlocking New Markets Beyond Traditional Vehicle Sales

Dojo’s true potential lies not only in self-driving cars but also in its ability to open up new markets that go far beyond vehicle sales. Analysts at Morgan Stanley envision a future in which Dojo enables any device equipped with a camera to make real-time decisions based on its visual field. This increase in addressable markets has the potential to reshape Tesla’s business model, emphasizing software and services rather than selling vehicles at fixed prices.

A Bullish Outlook

As a result of these developments, Morgan Stanley has upgraded Tesla’s stock from “equal-weight” to “overweight” and named it their “top pick,” replacing Ferrari’s US-listed shares. The Wall Street brokerage also increased its 12-month target price on Tesla stock by 60% to $400, the highest among all Wall Street brokerages. This goal is expected to boost Tesla’s market capitalization to around $1.39 trillion, a whopping 76% increase from its current market value of around $789 billion.

The Future of Tesla

Morgan Stanley analysts, led by Adam Jonas, believe Dojo will be the most valuable asset in Tesla’s software and services division. They also increased their revenue forecast for Tesla’s network services business to $335 billion in 2040, up from $157 billion previously. Furthermore, Jonas predicts that by 2040, this unit will contribute more than 60% of Tesla’s core earnings, nearly doubling its contribution in 2030.

This significant increase can be attributed to new opportunities in third-party fleet licensing as well as higher average monthly revenue per user (ARPU). As it expands its services beyond traditional vehicle sales, Tesla is poised to tap into these revenue streams.

A Market Leader with a Premium Valuation

Tesla’s 12-month forward price-to-earnings ratio of 57.9 outperforms legacy automakers such as Ford (6.31) and General Motors (4.56). This premium valuation reflects not only Tesla’s dominance in the electric vehicle market, but also its global potential to revolutionize transportation, software, and services.

To summarize, Tesla’s Dojo supercomputer marks a watershed moment in the company’s journey toward autonomy and beyond. Tesla’s market value could skyrocket as a result of its potential to transform not only the automotive industry but also a variety of other markets. Investors and enthusiasts alike are eagerly anticipating the future shaped by Dojo’s groundbreaking capabilities as the electric vehicle pioneer continues to break new ground.

[blackholerobots]

Tesla’s push into supercomputers, potentially worth $500 billion in added market value, just suffered a big blow after its chief departs | Fortune

[metamoonshots]

Tesla revealed its Q3 2023 shareholder deck on Oct. 18, indicating that it continues to carry its Bitcoin steadiness because it directs assets to AI efforts. The corporate reported $184 million in digital belongings, recognized to be composed largely or solely of Bitcoin (BTC), for the quarter ending Sept. 30, 2023. That quantity is the same as Tesla’s holdings for the three earlier quarters however represents a long-term lower because of earlier gross sales and worth adjustments. Tesla initially bought $1.5 billion of Bitcoin in 2021, an quantity that after surpassed $2 billion in worth. Nevertheless, the corporate offered most of its Bitcoin in 2022, inflicting it to report simply $218 million in digital belongings for the quarters ending June 30, 2022, and Sept. 30, 2022. In the identical shareholder deck, Tesla detailed its targets regarding its car and vitality companies and, considerably, its synthetic intelligence endeavors. It stated: “We now have greater than doubled the scale of our AI coaching compute to accommodate for our rising dataset in addition to our Optimus robotic mission. Our humanoid robotic is presently being skilled for easy duties by way of AI moderately than hard-coded software program, and its hardware is being additional upgraded with a purpose to deal with datasets.” By way of common earnings, the corporate reported $23.4 billion in income, down roughly 6% from $24.9 billion in Q2 2023. On each events, the corporate’s automotive enterprise accounted for the overwhelming majority of general income. Buyers need to hear about crypto A number of buyers have once more submitted a number of crypto-related questions, together with one on whether or not Tesla would add Bitcoin funds for its charging community and one other asking if Tesla’s Dojo supercomputer would make use of Bitcoin. A number of folks requested whether or not the corporate plans to as soon as once more Bitcoin and crypto funds generally. That is in keeping with Tesla CEO Elon Musk’s earlier promise to just accept Bitcoin as soon as the mining trade is 50% reliant on renewable vitality. Tesla’s shareholder name was ongoing at press time. The put up Tesla maintains Bitcoin holdings while directing resources toward AI appeared first on CryptoSlate.

[isfeed]

Tesla’s Dojo, a timeline

Elon Musk doesn’t want Tesla to be just an automaker. He wants Tesla to be an AI company, one that’s figured out how to make cars drive themselves.  Crucial to that mission is Dojo, Tesla’s custom-built supercomputer designed to train its Full Self-Driving (FSD) neural networks. FSD isn’t actually fully self-driving; it can perform some […] © 2024 TechCrunch. All rights reserved. For personal use…

[nawapon17]

Tesla Dojo: Elon Musk's big plan to build an AI supercomputer, explained | TechCrunch

[wangwill66]

超級電腦

H：徐志摩稱數大便是美，個人電腦從中央處理器CPU16位元，到64位元，電腦所稱的X286，X386，X486，X586之後一直發展。公司的電腦發展到超級電腦是怎樣的光景?「特斯拉2022年有3台Dojo超級電腦，每秒處理40萬部影片容量高達30PB。」「美國橡樹嶺實驗的Frontier系統超越1exa Flops(100萬兆浮點)。」。「最強大的Fugaku (富岳)超級電腦。」特別查一查。20230914W4

維基百科介紹：

超級電腦，指能夠執行一般個人電腦無法處理的高速運算的電腦，規格與效能比個人電腦強大許多。現有的超級電腦運算速度大都可以達到每秒一兆次以上。「超級計算」這名詞第一次出現，是在1929年《紐約世界報》關於IBM為哥倫比亞大學建造大型制表機的報導。 維基百科

最強大的Fugaku (富岳)超級電腦。

非凡新聞台報導(20230914)：

特斯拉2022年有3台Dojo超級電腦，每秒處理40萬部影片容量高達30PB。

美國橡樹嶺實驗的Frontier系統超越1exa Flops(100萬兆浮點)。

網路資料

Dojo可開啟「全新的潛在市場」，就如同亞馬遜(Amazon.com Inc.)旗下的雲端運算服務子公司AWS。AWS目前為亞馬遜貢獻70%的息稅前利潤(EBIT)。

特斯拉的Dojo超級電腦由幾個「系統托盤」組成，採用該公司內部的D1晶片，這些晶片內置在機櫃中，然後合併成一個「ExaPOD」超級電腦。ExaPOD可以串聯在一起形成集群，使超級電腦更加強大。

維基百科介紹：

艾可薩（英語：exa-），又譯為艾，是一個國際單位制詞頭，符號E，代表1018倍。100萬兆，10的8次方是億，10的18次方是百萬兆。

十進位單位符號

詞冠符號10的因數petaP10的15次方teraT10的12次乡gigaG10的9次方megaM10的6次方

每秒浮點運算次數

每秒浮點運算次數，亦稱每秒峰值速度，（英語：Floating-point operations per second；縮寫：FLOPS），即每秒所執行的浮點運算次數。浮點（floating-point）指的是帶有小數的數值，浮點運算即是小數的四則運算

，常用來測量電腦運算速度或被用來估算電腦效能，尤其是在使用到大量浮點運算的科學計算領域中。因為FLOPS字尾的那個S代表秒，而不是複數，所以不能夠省略。在多數情況下，測算FLOPS比測算每秒指令數（IPS）要準確。

前沿 (超級電腦)

Frontier（OLCF-5）是由HPE Cray和AMD建造，託管於美國田納西州橡樹嶺國家實驗室的超級計算機，它基於Cray EX架構，是Summit（OLCF-4）的繼任者，也是全球首台百億億次級計算機[1]。2022年5月，Frontier獲得超級計算機運算速度全球排名第一名[2]。目前早期運營性能達到了每秒1.194百億億次每秒浮點運算次數（PFlop/s），其性能代表了整個榜單計算能力的四分之一，但該計算機尚未達到最佳優化，雖然已經超越百億億次目標但是未達到設計性能指標。

介紹

Frontier的設計目標計算性能是大於1.5 百億億每秒浮點運算次數（PFlop/s）[3][4]。預計成本6億美元。

Frontier計算系統被安裝在74個19英寸（48厘米）機櫃中，每個機櫃安裝了64個包含了2個節點的刀片服務器，每個節點帶有1個高性能和AI計算負載優化的AMD EPYC 7453s 64核CPU和4個AMD Radeon Instinct MI250X GPU 整套系統包括9,472個CPU和37,888個GPU，總計CPU內核數達606,208個，GPU內核數達8,335,360個[5]。現今Frontier的功率為22,703千瓦，為降低其在運行時產生的熱量，該系統配備水冷系統散熱，其水冷系統擁有四個大功率水泵，每分鐘可推動超過25噸水在機器周圍流動[1]。Frontier採用慧與科技子公司克雷公司開發的Cray OS做為作業系統。

橡樹嶺國家實驗室

橡樹嶺國家實驗室（英語：Oak Ridge National Laboratory，縮寫為ORNL）是美國能源部所屬的一個大型多學科研究國家實驗室[2]，位於田納西州橡樹嶺。此實驗室成立於1943年，最初是作為曼哈頓計劃的一部分，以生產和分離鈾和鈽為主要目的建造的，原稱克林頓實驗室。2000年4月以後由田納西大學和巴特勒紀念研究所（英語：Battelle Memorial Institute）共同管理。

拍字節PB

PB千兆位元組

單位位元的倍數

拍位元組（英語：Petabyte，縮寫為PB）（又稱千兆位元組）是一種資訊計量單位，現今通常在標示網路硬碟、伺服器農場總容量，或具有大容量的儲存媒介時使用。

Quick Facts 位元組的次方單位, 十進制前綴 (SI) ...

使用示例

在2009年，阿凡達3D版在維塔數碼使用超過1 PB來儲存。

Google的服務器集群於2004年時的估計容量大約是5.625 PB。

美國國會圖書館的書籍與非書籍藏品數量，於2005年的估計近似值大約是10 PB。

在2012年的8月，Facebook共使用約100 PB儲存空間。

2012年為止，人類生產的所有印刷材料的數據量是200PB。[1]

Google相簿在發布1年之後，2016年5月的統計為用戶節省共約13.7 PB的存儲空間[2]。

匯豐銀行儲存全球70個國家和地區，共3,700萬客戶的資料和交易紀錄，資料總量由2014年的56 PB，激增至2017年的93 PB[3]。

《微軟模擬飛行2020》的全球地景模擬系統資料庫約為2 PB。

與其他儲存單位的換算

1PB = 1,000TB = (10)3 TB

1PB = 1,000,000GB = (10)6 GB

1PB = 1,000,000,000MB = (10)9 MB

1PB = 1,000,000,000,000KB = (10)12 KB

1PB = 1,000,000,000,000,000B = (10)15 B

另外，Petabyte往往亦可以指Pebibyte，其換算公式是：

1PiB = 1,024 TiB

1PiB = 1,048,576GiB = (1024)2 GiB

1PiB = 1,073,741,824MiB = (1024)3 MiB

1PiB = 1,099,511,627,776KiB = (1024)4 KiB

1PiB = 1,125,899,906,842,624B = (1024)5 B

富岳 (超級計算機)

日本超級電腦

富岳（日語：富岳／ふがく Fugaku */?、英語：Supercomputer Fugaku）是富士通與日本理化學研究所共同開發的超級電腦，作為「京」的後繼機型。2014年開始研發，2021年正式啟用。[2]富岳部署在兵庫縣神戶市中央區港灣人工島上的理化學研究所計算科學研究中心內，其運算能力為京的100至120倍，耗電為30百萬瓦至40百萬瓦，京的耗電為12.7百萬瓦[3]。��富岳」是富士山的別稱[4]。

Quick Facts 啟用時間, 承建商 ...



京/富岳 - SC 18展覽會



PRIMEHPC FX1000 (富岳節點) - SC 19展覽會

富岳是全球首度奪冠的ARM架構超級電腦，採用富士通48核心A64FX SoC，與過往超級電腦大多採用的Intel或AMD的x86、x64主流平台不同。富岳共有158,976個節點，尖峰性能可達到1 exaFLOPS（1,000 petaFLOPS）。富岳除了在Linpack中拿到好成績，也在HPL-AI中獲得1.421 exaFLOPS。

2020年6月23日，富岳正式獲認證，以415 PFLOPS計算速度成為TOP500排名第一的超級電腦[5]。之後同年11月17日發表的TOP 500排行榜成功蟬聯第一[6]。

神戶新交通港灣人工島線的京電腦前站於2021年6月19日更名為計算科學中心站時，將副站名同時變更為『神戶動物王國・「富岳」前』，以紀念富岳取得TOP500排行第一[7][8][9]。

2023年5月22日，日本東京工業大學、富士通、理化學研究所和東北大學宣布將利用「富岳」於研發日語生成式AI[10]。

徐志摩

數大便是美，碧綠的山坡前幾千隻綿羊，挨成一片的雪絨，是美； 一天的繁星，千萬隻閃亮的眼神，從無極的藍空中下窺大地，是美； 泰山頂上的雲海，巨萬的雲峰在晨光裏靜定著，是美； 大海萬頃的波浪，戴著各式的白帽，在日光裏動盪著，起落著，是美； 愛爾蘭附近的那個羽毛島上棲著幾千萬的飛禽， 夕陽西沉時只見一個羽化的

CPU中央處理器

蘋果電腦開始轉而採用英特爾x86處理器，2020年蘋果推出自家中央處理器M1，蘋果採用ARM架構並且CPU+GPU+RAM做在同一顆晶片。

CPU（如16，32，64甚至128位），