Parallel information processing system

Abstract

The invention relates to a parallel information processing system comprising a plurality of heterogeneous PMC (...) modules. Each PMC module comprises a processing part, a memory part and a communication part. Tasks are divided into subtasks with different algorithm characteristics and are transmitted to the PMC modules matched with the algorithm characteristics of the subtasks for parallel processing, and during processing, the processing parts, the memory parts and the communication parts of the PMC modules process the subtasks simultaneously. Compared with the prior art, the parallel information processing system has the advantages of high cost performance and processing speed and variable system structure and the like.

Description

technical field [0001] The invention relates to an information processing system, in particular to a parallel information processing system based on space-time heterogeneous reconstruction. Background technique [0002] With the development of computer parallel technology, the parallel computing model has always been a hot topic of exploration and research by parallel processing experts and scholars at home and abroad. In the early stage of computer development, the structure of the parallel machine system was relatively simple, and the difference between storage access and CPU speed was small, and performance analysis focused on calculation, such as the PRAM model. With the development of parallel processing technology, the complexity of the structure and the increase of the gap between the storage access speed and the CPU computing speed, etc., make the parallel computing model have to face the performance factors that were not taken seriously. The expansion of descriptio...

AI Technical Summary

Problems solved by technology

However, the traditional serial computing model also has many shortcomings: ① A large number of transistors on the chip are wasted

With the calculator as the basis of calculation or processing, no matter how complex the processing is, the time and cost of the result calculation and processing are far less than the cost of data handling, which has become a serious bottleneck hindering the development of computers

③Multiple CPUs and many cores cannot be realized

④ Calculation granularity is too small

Using instructions as the basic computing granularity to realize various processing tasks, the computing granularity is too small, which seriously restricts the efficient use of computer hardware resources

⑤Processor personality can not play

Although there are many types of existing architectures, once they are set, they are often difficult to change and cannot meet the diversity of applications in the objective world. The rigid architecture has become a major obstacle to the development of computer systems, and it is necessary to find diverse solutions according to the needs of users. computing particles, and then develop new architectures and new solutions based on the characteristics of computing particles

Images