Lanqi Technology releases new fourth-generation Jinja CPU: using Intel fourth-generation Xeon scalable processor Sapphire Rapids core
CTOnews.com, January 12 (Xinhua)-- Lanqi Technology today released its new fourth-generation Jinja CPU, which aims to provide surging computing support for cloud computing, enterprise applications, artificial intelligence and high-performance computing with superior performance.
Lanqi Technology fourth-generation Jinja CPU, with Intel fourth-generation Xeon scalable processor (code: Sapphire Rapids) as its core, has passed the Lancheng Technology Security pre-Test (PrC) test, and is a x86 architecture server processor for the local market. Compared with the previous generation products, the fourth generation Jinnai CPU adopts Intel 7 process technology, with a maximum number of 48 cores, a maximum Rui frequency of 4.2GHz, and a maximum shared cache of 105MB, which greatly improves the key performance indicators.
At the same time, in terms of bandwidth, throughput, latency and other interconnection performance, the fourth generation Jinja CPU has achieved a series of breakthroughs or upgrades: UPI 2.0 supports 4-way interconnection between CPUs at a rate as high as 16GT / s, and the memory type is upgraded to 8-channel DDR5 with a speed up to 4800MT / s, which is 50% higher than that of DDR4, and 16 memory sticks are supported in a single slot. Support PCIe 5.0A single CPU supports up to 80 PCIe channels with a transfer rate up to 32GT / s, double that of the previous generation, and can support higher-speed network cards, GPGPU cards and storage devices. The introduction of CXL 1.1can support a variety of accelerator cards and memory expansion, improving system performance while reducing the total cost of ownership of the data center.
The fourth generation Jinja CPU has built-in AI acceleration engine AMX (Advanced Matrix Extensions), which supports two data types, INT8 and BFloat16, which are widely used in the industry, and can significantly improve the performance of artificial intelligence-related applications. CTOnews.com learned that the fourth-generation Jinja CPU also integrates four on-demand activation accelerators, such as DSA (Data Streaming Accelerator), QAT (Quick Assist Technology), DLB (Dynamic Load Balancer) and IAA (In-memory Analytics Accelerator), which can uninstall special tasks such as data analysis, cryptographic operation, data compression and load balancing, and release CPU core computing resources for key business applications, thereby improving the overall performance of the system.
At present, several server manufacturers have cooperated with Lan Qi Technology to complete the research and development of server models based on the fourth generation Jinnao CPU.
