Multiprocessor system and effectiveness regulation method thereof

Abstract

A multiprocessor system and an efficiency regulation method thereof are provided. The multiprocessor system has a plurality of processing units that at least comprises a first processing unit and a second processing unit. The efficiency regulation method comprises the following steps of: firstly detecting the loads of these processing units to obtain a plurality of relevant detection results; then judging whether the loads are concentrated in one of the processing units according to the detection results; and finally increasing the power supply to the first processing unit or increasing the efficiency of the first processing unit if the loads are concentrated in one of the processing units.

Description

technical field [0001] The invention relates to a technology of a multiprocessor system, and in particular to a multiprocessor system and a performance adjustment method thereof. Background technique [0002] Please refer to figure 1 , which is a schematic diagram of a known multiprocessor system. The multiprocessor system 100 is generally assembled on the motherboard of a computer device, and this multiprocessor system 100 usually includes more than two central processing units (CPU), such as: a first processing unit 110 and a second processing unit 120 , and respectively supply power to the first voltage regulator module (Voltage Regulator Module, VRM) VR1 and the second voltage regulator module VR2 of these processing units 110, 120. [0003] All current voltage regulation modules can adjust the output voltage according to the core voltage level requirement of the processing unit. For example, the first voltage regulation module VR1 can generate a dynamic voltage ident...

AI Technical Summary

Problems solved by technology

[0005] However, as we all know, even at the current hardware level of dual-CPU or multi-core processors, it is still rare to support software programs.

For example, due to the difficulty of program design, game developers almost still write game programs in a single thread (Single Thread) manner, causing the multiprocessor system 100 to use only one processing unit (such as the first processing unit 110) to Executing computer game software while the unused second processing unit 120 is idle

Or, even if the above-mentioned processing units 110, 120 are allocated with a considerable amount of computing data, often because the program is not optimized for the multi-processor architecture when it is written or compiled, so that the data still has correlation rather than completely independent

At this time, the second processing unit 120 may have to wait to receive the output result of the first processing unit 110 before starting to perform the calculation it is responsible for, that is, the above-mentioned processing units 110 and 120 cannot fully exert their computing capabilities at the same time.

Although the above-mentioned processing units 110, 120 theoretically have the computing power that is multiples of that of a single processor, when encountering the above-mentioned computing bottleneck, the improvement of the overall performance of the system is still limited, and it cannot show the expected multi-processing compared with a single processor. Computing advantage

Images