Computing Power Upgrade: Feng's Architecture "Breaker", Breaking Through the Bottleneck of Storage and Computing Integration

Computing Power Upgrade: Feng's Architecture "Breaker", Breaking Through the Bottleneck of Storage and Computing Integration

The Fung architecture is centered on computing, and computing and storage are separated, and the two cooperate to complete data access and computing. However, since the design of the processor is mainly to increase the calculation speed, the storage is more focused on capacity improvement and cost optimization, and the performance mismatch between "storage" and "calculation", resulting in low memory access bandwidth, extended time, high power consumption, etc. Problems, commonly referred to as "storage wall" and "power consumption wall".

As a new computing architecture, the integration of storage and computing is considered to be a revolutionary technology with potential. The core is to fully integrate storage and computing, effectively overcome the bottleneck of the von Neumann architecture, and combine technologies such as advanced packaging and new storage devices in the post-Moore era to reduce invalid data transfer and improve computing efficiency. China Mobile has incorporated the integration of storage and computing into the ten key technologies of the computing power network.

At present, traditional devices such as NOR Flash and SRAM are relatively mature and can take the lead in promoting the commercialization of in-memory computing. Among the new devices, the overall performance of RRAM is better, and the lifespan and read/write performance of MRAM are better. Both have their own unique advantages and development potential to continuously promote the maturity of devices and simultaneously carry out in-memory computing exploration.


Samsung Electronics, SK Hynix, #TSMC, Micron, #IBM, Intel, etc. are all conducting research on storage-computing integration technology. Among Chinese companies, Yizhu Technology, Qianxin Technology, and Houmo Intelligent focus on integrated chips with large computing power. Count as one chip.

要查看或添加评论,请登录

Lansheng Technology Limited的更多文章

社区洞察

其他会员也浏览了