Server node architecture design method for separated type high-capacity memory

Abstract

The invention discloses a server node architecture design method for a separated type high-capacity memory and belongs to the technical field of server design. The architecture design method specifically comprises the steps that a memory unit, a calculation unit and an IO unit are designed, wherein the calculation unit comprises a CPU, the memory unit comprises a high-capacity SDRAM memory bank and a memory controller, and the IO unit comprises an IO Subsystem; the CPU controls the memory unit through a DHCM bus and is connected with the IO Subsystem through an IO bus; the SDRAM memory bank is connected with the memory controller through a DDR bus; the calculation unit is in charge of controlling memory operations in the memory unit, the memory unit is in charge of storing and calling data, and the IO unit is in charge of controlling input and output of the data. The server node architecture design method effectively overcomes the defects of traditional server architecture design and improves the total capacity of a server system, and therefore the performance of the system is improved.

Description

technical field [0001] The invention discloses a server node architecture design method, which belongs to the field of server design; in particular, it relates to a server node architecture design method with separate high-capacity memory. Background technique [0002] The server comes with a certain amount of RAM randomly when it leaves the factory, and the amount of memory depends on the needs of the work and the strategy of the manufacturer. Most of the mainboards of servers now have spare memory slots or support memory expansion boards, so that more memory can be installed to expand memory capacity and achieve higher performance. However, with the development of information technology, there are more and more server application scenarios, and the increasing business volume requires higher and higher server performance. Sometimes the server even has to respond to requests from dozens or even hundreds or thousands of computers. In addition to continuously increasing the s...

AI Technical Summary

Problems solved by technology

[0004] The present invention aims at increasing the memory in the traditional server architecture, generally increasing the number of memory sticks, but is limited by the speed of the DDR bus and the space of the computing node board, and provides a server node with a separate high-capacity memory Architectural design method, the shortcomings of effective traditional server architecture design, increase the total memory capacity of the server system to improve system performance

Images