Nano Labs Debuts FPU3.0 With 5X Power Efficiency Boost: Details
Nano Labs Debuts FPU3.0 With 5X Power Efficiency Boost: Details
The FPU3.0 architecture features stacked 3D memory with a theoretical bandwidth of 24TB/s and an enhanced Smart-NOC on-chip network.Nano Labs Ltd (NASDAQ:NA) shares are trading lower on Thursday. Today, the company unveiled FPU3.0, its latest ASIC architecture aimed at enhancing AI inference and blockchain performance.
毫微實驗室有限公司（納斯達克：NA）股票在週四交易中下跌。今天，該公司發佈了FPU3.0，這是其最新的ASIC架構，旨在提升人工智能推理和區塊鏈性能。
The FPU series is Nano Labs' proprietary ASIC chip architecture tailored for high-bandwidth High Throughput Computing applications.
FPU系列是毫微實驗室專有的ASIC芯片架構，專爲高帶寬高送轉計算應用量身定製。
FPU3.0, utilizing advanced 3D DRAM stacking, achieves a fivefold increase in power efficiency compared to its predecessor, FPU2.0, setting a benchmark for energy-efficient, high-performance ASICs.
FPU3.0採用先進的3D DRAm堆疊，與其前身FPU2.0相比，實現了五倍的能效提升，爲能源高效、高性能的ASIC設定了基準。
The FPU3.0 architecture features stacked 3D memory with a theoretical bandwidth of 24TB/s and an enhanced Smart-NOC on-chip network.
FPU3.0架構具有堆疊的3D內存，理論帶寬爲24TB/s，並增強了芯片內部的Smart-NOC網絡。
This modular design allows for rapid product updates by modifying the FPU core IP while reusing or upgrading other components, enabling the seamless introduction of new features.
這種模塊化設計通過修改FPU核心IP，同時重用或升級其他元件，允許快速的產品更新，使新功能的引入變得無縫。
This network accommodates diverse compute cores and traffic types, including large and small cores, full-crossbar, and feed-through traffic.
這個網絡可以容納多樣的計算核心和流量類型，包括大核心和小核心、全交叉開關和直通流量。
The Nano FPU architecture consists of four key components: the Smart Network-on-Chip, a high-bandwidth memory controller, chip-to-chip interconnect IOs, and the FPU core.
Nano FPU架構由四個關鍵元件組成：Smart Network-on-Chip、高帶寬內存控制器、芯片間互連IO以及FPU核心。
These chips deliver superior computational efficiency and lower power consumption compared to general-purpose CPUs and GP-GPUs, making them ideal for AI inference, edge AI computing, 5G data transmission processing, and network acceleration.
與通用CPU和GP-GPU相比，這些芯片提供了更高的計算效率和更低的功耗，使其非常適合人工智能推理、邊緣AI計算、5G數據傳輸處理和網絡加速。
Price Action: NA shares are down 15.8% at $7.57 at the last check Thursday.
價格走勢：納斯達克的股票在週四最後一次檢查時下跌15.8%，目前爲7.57美元。
Read Next:
下一步閱讀：
- Chinese Fabless Integrated Circuit Company Nano Labs Stock Skyrockets With New V Series Launch: Details
- 中國無晶圓集成電路公司毫微實驗室股票因新V系列推出而暴漲：詳情
Image via Shutterstock.
圖片由Shutterstock提供。
譯文內容由第三人軟體翻譯。
