はじめに
Xの投稿にて、RAIN AI なる会社を知りました。
RAIN AI
どんなものを作っているかの詳細は、まだわかりませんが、ビジネスとしては、下記のものがあるようです。説明のために下図を引用します。
パッと見、tenstorrent みたいなビジネスモデルですね。。。
Blog を覗いてみると、
なるものがあります。
LinkedInで調べても、Silicon関連の人は少ないですね。
- Cagri E. : SoC Architect at Rain.ai ( - 2023.3)
- Daniel Talmon : Principal Engineer Physical Design ( - 2023.11)
- Ismael Garcia : ASIC Design Enginee ( - 2024.2)
職
$220K/年棒ですか。。。なんか凄いですね。。。
消えると思うので、記録として残します。全文を引用します。
About Rain: At Rain AI, we’re creating a future with abundant and scalable artificial intelligence by building the world’s most cost and energy efficient hardware for AI. Our ultimate goal is to become the dominant hardware company of the AI era. Rain AI is currently a Series A stage startup and backed by world leaders in AI. Our seed round was led by Sam Altman (OpenAI). In addition, our current VC partners include Y Combinator, Daniel Gross, Jaan Tallinn, Founders X Fund, Airbus Ventures, Liquid 2 Ventures and Deepwater. About the role: Rain AI is looking for a ML Hardware Architect to help define the architecture of Rain’s next-generation AI accelerators. Given the increased customer demand and ambitious timelines, we are growing our architecture team and open to candidates at various levels (staff to principal). You would have a large degree of independence and influence at Rain, with an immense opportunity to directly impact one of the most important topics across AI & semiconductor. This role will be based in the Bay Area, and will report to our Lead Architect. Responsibilities: Provide technical assistance to the architecture team in development of next-generation AI accelerators to enable novel machine learning algorithms. Work closely with the algorithms teams to understand and evaluate local and global data movement in different potential architectures. Work closely with the system software team to develop compilers and ISA extensions to enable new hardware functional units and features at the software level in an SoC setting. Work closely with the IC Design team to evaluate various microarchitectures and digital and mixed-signal design collaterals in synchronous and asynchronous architectures and collaborate on microarchitecture development and specification hand-off architecture specifications to the IC Design team. Develop functional and/or performance simulators for architecture evaluation, system design, verification, software prototyping and hardware-software co-optimizations. Integrate proprietary accelerators and design macros ARM or RISC-V cores. Develop and maintain architectural specification and documentation to be used by architecture, design, verification and software teams. Qualifications: MSEE with 10+ years or PhD in Electrical Engineering and/or Computer Engineering with a focus on SoC architecture, ML accelerator design, system and performance modeling, and simulation. Familiarity with compute architecture, programming models, compiler optimizations and code generation. Familiarity with deep learning models and willingness to learn novel algorithms and translate them into specifications for hardware accelerators. Familiarity with algorithmic techniques for on-device learning (online/continuous learning) is a plus. Deep understanding of processor microarchitectures such as x86, ARM, and RISC-V. Strong understanding of memory subsystem functionality, on-chip and off-chip data movement and associated tradeoffs, logic-memory optimizations and efficient code/date placement techniques. Experience with SystemC, Python, C/C++ and the ability to write production-ready, annotated code. Familiarity with the integration of novel accelerators and design macros with ARM or RISC-V cores. Exhibit a high degree of motivation and independence. Strong communication skills, both written and verbal. Preferred qualifications: PhD in Electrical Engineering and/or Computer Engineering with focus on near- and in-memory computing architectures for AI acceleration. Demonstrated track record of owning architecture of an AI accelerator from conception to tape-out. Demonstration of programmability. Our Benefits: Medical Insurance with 100% coverage of employee premiums Dental and Vision Insurance 401k match Unlimited vacation + all federal holidays Company wide time off: Time between Christmas and New Years, first week of August Health & Wellness Perk The anticipated annual base salary for this position in the United States is $220,000 to $320,000. This range does not include any other compensation components or other benefits that an individual may be eligible for.
どんなチップ？
によると、「アナログAIチップ」
あれ、この時の会社名は、「Rain Neuromorphics」だったんですね。
同社が2022年初めにテープアウトしたNPUテストチップは、180nmのCMOSプロセスで構築されており、1万個のニューロンを備えている。このチップは実証された学習ならびに推論能力を既に備えている。
とあるので、テストチップはあるんですね。。。
この記事によると、
Rain AIは今年に入ってCEOをひっそりと交代させたようだ。創業CEOのゴードン・ウィルソンは、現在はエグゼクティブ・アドバイザーとしてウェブサイトに掲載されており、名門法律事務所出身の元弁護士のウィリアム・パッソがCOOからCEOへと昇進している。
Rain AIの初期のチップは、グーグルやクアルコムなどのテック企業が支持しているオープンソースのチップアーキテクチャー「RISC-V」をベースにしている。スマートフォンやドローン、自動車、ロボットなど、データセンターから遠く離れた場所で動作する“エッジデバイス”と呼ばれるものがターゲットだ。
おわりに
Sam Altman san が出資している Rain AI 、どうなんでしょうかね。。
下記のような記事もありました。
2024.06.19 の記事で、来年初めに準備が整うと。。。
うーん、日本でRISC-V関連のIPビジネス。。。。。。。。。。。。。。。。。。ダメだ。。。
