System and method for dynamic low-delay optimization

Abstract

The invention provides a system and method for low-delay optimization. The system and method are configured to be executed from a hardware layer of a whole operation system to an application program.Low-delay running implemented according to the embodiment is optimization aiming at a specific application program. The low-delay running is connected to specific hardware, implemented on the system based on a main processor and configured in low-delay optimization. For example, implementation of the low-delay optimization can involve various modules implemented in user space and kernel space, wherein these modules cooperate to obtain information of services and hardware related to application program use; and the information is provided for low-delay running of related application programs. During running in accordance with the embodiment, the low-delay running is started or forbidden dynamically by a type of low-delay optimization implementation; and based on each application program, the low-delay running can be conducted on the application program at proper time or according to needs.

Description

technical field [0001] The present invention relates to the operation of processor-based systems, and more particularly, to dynamic low-latency optimization with respect to the operation of processor-based systems. Background technique [0002] The use of various forms of processor-based platforms has exploded in recent years. For example, the use of processor-based personal communication and data processing platforms, commonly referred to as smartphones, has grown dramatically over the past few years and is expected to continue to grow for the foreseeable future. For example, in addition to providing cellular telephone communication services, smartphones are often used to provide data communications, data processing, viewing and / or creating streaming media, gaming and entertainment, and the like. Applications that can provide the aforementioned functionality range from those that use less processing power and other resources (eg, thin clients and simple web browser applica...

AI Technical Summary

Problems solved by technology

[0007] Similarly, user-space low-latency techniques are not optimized for any particular application, are implemented statically, and may or may not have any significant impact on the latency experienced by the application, depending on the resources used by the application. influences

For example, improving runtime performance by using optimized compilation options (for example, using the "-O3" compilation option of the GNU compiler toolchain) is applicable during compilation, but cannot be reverted to the original option during runtime.

In addition, existing user-space low-latency techniques are not tightly coupled with existing kernel-space low-latency techniques, so instead of providing low-latency optimizations for a combination of operating systems and applications, they try to address the operating system separately and independently and the latency of one of the applications

For example, a command line to be able to start / stop system services may be available to developers, but these command line tools are usually not accessible to application developers

Also, existing user-space low-latency techniques focus on application-related tasks rather than the entire data path

In addition, existing user-space low-latency techniques do not directly access processor-based system hardware or its information, but instead interface through a hardware abstraction layer (HAL) that blocks access to the understanding of specific hardware

Images