Big data engine prototype based on multicore heterogeneous CPU-GPU-FPGA (central processing unit, graphics processing unit and field-programmable gate array)

Abstract

The invention discloses a big data engine prototype based on multicore heterogeneous CPU-GPU-FPGA (central processing unit, graphics processing unit and field-programmable gate array); an architecture of the prototype is characterized by including independent parallel CPU and parallel GPU having their respective magnetic random storage subsystems, and the subsystems can access their magnetic random storages mutually and are positioned outside an FPGA structure; the GPU is connected to a chip set through I / O (input / output) bus and is connected with the CPU through an I / O bridge; the CPU is composed of an ALU (arithmetic logic unit, a register file, an intelligent flash cache and a bus interface. System implanting is performed to support the multicore heterogeneous CPU-GPU-FPGA big data engine architecture, forming a big data engine prototype.

Description

technical field [0001] The invention relates to a prototype of a big data engine based on multi-core heterogeneous CPU-GPU-FPGA Background technique [0002] With cloud computing and virtualization, showing the characteristics of "large scale", "high density", "high energy consumption" and "complexity", the construction and development of a new generation of data centers and the improvement of data center infrastructure management will become increasingly Importantly, the infrastructure integration management and intelligence of the data center will become a new trend in the development of the data center. [0003] The ultra-large data center provides the entire application service from infrastructure to subsequent data analysis, screening, and application. Not only data analysis, but also cloud computing dedicated to smart manufacturing, which is different from the generalized services provided by public clouds, and supercomputing, which put forward higher requirements for...

AI Technical Summary

Problems solved by technology

[0006] Although dedicated computing units such as GPUs have low operating frequency, more cores and parallel computing capabilities, the overall performance / chip area ratio and performance / power consumption ratio are high, but they are far from being fully utilized.

Images