跳到主要內容

The Rapid Rise of RISC‑V

Jack Kang, SVP, Business Development, Customer Experience, and Corporate Marketing at SiFive

SiFive is aiming high with bold new technology for performance-driven applications
SiFive transformed in 2021 and grew from leading RISC-V for embedded products into performance-demanding markets, creating real choice in the semiconductor processor IP market. Now, the SiFive portfolio features three distinct, market-focused product families, based on market requirements ranging from high-performance applications, machine learning and artificial intelligence processing, to embedded real-time deterministic and small application processors. Our RISC-V IP portfolio is designed around our focus to deliver processor IP for these demanding markets, combining the unlimited potential of RISC-V with the experience and talent of our team.

This rapid progress was made possible by our partners and customers who believe and invest in RISC-V and the freedom it represents for technology. RISC-V has no limits, and all of us at SiFive are thankful and grateful for all our partner’s support in creating new products for the untapped potential of new and fast growing markets. In 2021, we reached significant milestones for our company with over 300 design wins at more than 100 companies, including 8 of the top 10 semiconductor companies; and we launched our new SiFive Intelligence™, SiFive Performance™, and SiFive Essential™ product brands with class-leading products in each family.

A Breakout Year
In 2021 we launched several new products. Most recently, we expanded the SiFive Essential family with the introduction of the SiFive Essential 6-Series processors, designed for mid-range applications; we also introduced the SiFive Performance P650, our fastest RISC-V processor, which will be available to lead customers in Q1 2022. Earlier this year we launched the SiFive Intelligence X280 for AI/ML markets, featuring SiFive Intelligence Extensions that build upon the RISC-V Vector extension to provide a platform for AI & ML with enhanced data types and support for TensorFlow Lite. The SiFive Intelligence X280 is a great inference processor and may be combined with existing hardened AI IP to enhance programmability and scalability by offering scalar and vector capabilities. The SiFive Performance P550 and SiFive Performance P270 processors for performance driven markets were also introduced in 2021, with up to 8-core coherent clusters and multiple clusters in an SoC, with heterogeneous mix+match options to allow for the right mix of processing capabilities.

The RISC-V ecosystem has moved forward significantly in 2021 with important specifications ratified that unlock and enable hardware and software development in key markets. Working alongside the RISC-V community has been a pleasure for the many SiFive employees who contribute to the technical groups that develop RISC-V specifications as we all work together to build the technology and tools needed for virtualization, advanced interrupts, vector processing, and more.

Building IP based on these new specifications is a key goal for SiFive, enabling our customers to quickly adopt the ratified standards. In 2021, SiFive offered three releases using our relentless innovation methodology to deliver new features in a software-like manner across our portfolio. SiFive increased efficiency and performance across our RISC-V processors, and introduced new features for cluster scalability, security, advanced power monitoring and control, and virtualization.

Raising The Bar in 2022
In 2022, we’ll continue to deliver high performance processor IP that will enable the creation of new computing platforms. From the metaverse to autonomous vehicles, natural language processing, recommendation systems, client computing devices and more, there is massive demand for RISC-V. We’re making the most of this demand by continuing to deliver high-performance processor IP packed full of performance and accompanied by the features our customers and partners require. If you didn’t catch them already, you can see some great pointers on what we’re doing in these talks from the 2021 RISC-V Summit:

If you’d like to be part of the rapid rise of RISC-V, SiFive is looking for talented individuals in many areas. You’ll work alongside the inventors of RISC-V and leaders in the semiconductor industry as we design, develop, and deploy RISC-V technology for a variety of computing challenges. Please look at our careers page here for more information and take the next step to be part of the unlimited future of RISC-V!

Keep up with SiFive news and announcements on social media: FacebookInstagramLinkedInTwitter, and YouTube.

留言

這個網誌中的熱門文章

鑽石級贊助商 - KKBOX 帶你打造具備 NLP 功能的 Telegram Bot (上)

打造具備 NLP 功能的 Telegram Bot(上) 最近因為一些契機學了 Python 3,用它做了一個 Telegram Bot ( GitHub 連結 ),裡面用到 NLP Service,用上下兩篇文章記錄一下實作過程還有眉角。上篇首先教大家如何做一個最基本的回聲 Chatbot,接下來我們可以透過 NLP 服務,讓 Chatbot 根據使用者不同的訊息做回答,這樣就變成更加人性化的聊天機器人囉! 使用的工具及服務: Python 3 (for develop) pipenv (for dependency management) OLAMI (for NLP) ngrok (for testing) Step 1. Creating new bot Telegram 很有趣的地方在於,與其他通訊軟體(Line、Messenger)相比,開發者管理 Bot 的方式也是透過官方提供的一位 Bot 在處理的,它叫做 BotFather (眾 Bot 之父 XD)。如果已經有 Telegram 帳號,只要加 BotFather 為好友,就可以開始管理你的 Bot。 加入 BotFather 好友後,它會親切地問候,並告訴你他能為你提供什麼服務。 I can help you create and manage Telegram bots. If you're new to the Bot API, please see the manual ([https://core.telegram.org/bots](https://core.telegram.org/bots)). You can control me by sending these commands: /newbot - create a new bot /mybots - edit your bots [beta] /mygames - edit your games ([https://core.telegram.org/bots/games](https://core.telegram.org/bots/games)) [beta] Edit Bots /setname - change a bot's name /setdescr...

COSCUP 2025 Call for Proposals / 徵稿辦法

COSCUP 常規徵稿已於 2025-05-10 截止,接下來進入加碼徵稿階段。加碼徵稿是為了提升大會的稿件品質,依據投稿狀況(數量、品質)部分徵稿主題可能提前喊停。最遲請於 05 月 24 日(AoE) 前投稿,徵稿主題可參考下方列表。 The regular call for proposals (CFP) for COSCUP closed on May 10, 2025. We are now entering the bonus CFP phase to improve the quality of submissions. Some topics may close earlier than expected, depending on the current status of submissions (in terms of quantity and quality). Please submit by May 24 (AoE) at the latest. You may refer to the topic list below for inspiration. 開始投稿 Submit Your Proposal 提案須知 Things you may need to know 演講形式:預設為現場30分鐘演講包含QA,若有其他需求可於提案系統註明,由各主題主辦單位決定如何安排。 Talk Format : The default format is a 30-minute on-site talk, including Q&A. If you have other requirements, please indicate them while submitting your proposal. The final arrangement will be decided independently by the organizers of each topic. 語言:COSCUP 受眾包含海內外與會者,大會不限制發表語言但鼓勵以英語發表。大會將公布雙語議程表,請提供中英文版議程介紹。 L...

機器學習的五大實務問題:對企業的影響與相應的化解方式

Appier 首席機器學習科學家 林守德博士 正如 Jason Jennings 及 Laurence Haughton 在《以快吃慢–如何藉速度在商戰中克敵制勝》一書中指出──未來,不是大公司吃掉小公司,而是速度快的公司吃掉速度慢的公司。 從現在開始,唯有善用適當的資訊快速做出決策的企業,才能成為戰場上的贏家。 機器學習技術驅動了這場變革。無論企業是嘗試向顧客提出建議、改進生產製造流程或應對市場的變動,都能運用機器學習技術處理大量的資料,進而提高自身的競爭優勢。 然而,機器學習雖能創造大好機會,卻也同時帶來了相應的挑戰。機器學習系統需要大量的資料,以及執行複雜的運算能力。顧客期望改變、出乎意料的市場波動等等外部因素,都意味著機器學習模型的運作並不是百分之百的自動,往往仰賴許多外部的資源來作監控及維護。 此外,機器學習也有不少尚待解決的實務問題。以下將深入探討機器學習的五大實務問題,以及這些問題對企業應用會產生的影響。 1. 資料品質 機器學習系統仰賴資料進行訓練,而訓練資料在廣義上可分為「特徵」及「標籤」兩種類別。 「特徵」是輸入機器學習模型的資料,像是來自感測器、顧客問卷、網站 cookie 或歷史資訊等等。 然而這些特徵的品質可能良莠不齊。舉例而言,顧客在填寫問卷時可能會隨便填寫,或對題目略而不答;感測器可能因失靈而回傳錯誤資料;即使使用者的網頁行為明確,網站 cookie 回報的資訊也可能不完整。 此外,資料也可能包含雜訊,當無謂的資訊夾雜其中時,機器學習模型將會受到誤導而做出不正確的預測。 相較於「特徵」,「標籤」的正確性與穩定度更為重要。標籤是機器學習模型最後輸出的結果。所以需要在訓練的時間利用正確的結果教導機器學習模型。標籤的稀疏性也是個問題,這是當系統已掌握大量輸入的資料卻對輸出的結果沒有把握時出現的現象。在這樣的情況下,將難以針對該模型偵測其特徵與標籤之間的關聯性優化,甚至需要耗費額外的人力干預,將標籤與輸入資料關聯起來。 機器學習需仰賴輸入與輸出資料的關聯,才能具備足夠的泛化能力以預測未來行動並提供相關建議。因此,如果輸入資料過於雜亂、殘缺或有所偏差時,將可能難以理解某輸出/標籤的產出原因。近年來機器學習也開發出許多先進的方法如半指導式學習,轉移學習來處理這樣的問題。 2. 複雜性與品質的取捨 建立強大的機器學習模型需要大量的計算資源來處理特徵和...