Browse Prior Art Database

Dynamic saving energy

IP.com Disclosure Number: IPCOM000245989D
Publication Date: 2016-Apr-22
Document File: 5 page(s) / 66K

Publishing Venue

The IP.com Prior Art Database

Abstract

Multi-core processors are the most viable means to delivering sustainable performance. One of the primary reasons for driving industry towards multi-cores is because of the power wall – the increasing of performance of single processor is being limited by power consumption. State-of-the-art multi-core processors provide per-core level dynamic voltage frequency scaling (DVFS) technique to leverage between the performance and the power consumption. Better performance of applications can be achieved by running them with higher frequency, or lower power consumption will be consumed by running them with lower frequency. This disclosure proposes a language extension with which the programmer/user is able to identify critical parts of the program that amortizes the benefits of DVFS.The key idea of this disclosure is to provide a simple annotation or language extension at the source code level that allows the user to arbitrary indicate which parts of the program are critical.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 27% of the total text.

Page 01 of 5

Dynamic saving energy

Multi-core processors are the most viable means to delivering sustainable performance at least in the following 10 years. One of the primary reasons for driving industry towards multi-cores is because of the power wall - the increasing of performance of single processor is being limited by power consumption. Furthermore, the cost of maintain servers, e.g. data centers, is going up as a result of the increased power capacity and cooling expenses. Authority estimates that for every $1.00 spent on new hardware, an additional $0.50 is spent on power and cooling. This is more than double the amount spent five years ago.

As a result, state-of-the-art multi-core processors provide per-core level dynamic voltage frequency scaling (DVFS) technique to leverage between the performance and the power consumption. This technology automatically allows processor cores to run faster than the base operating frequency by reducing the frequency of inactive cores (or shutting them down). Better performance of applications can be achieved by running them with higher frequency, or lower power consumption will be consumed by running them with lower frequency.

This disclosure proposes a language extension with which the programmer/user is able to identify critical parts of the program that amortizes the benefits of DVFS. The language extension is simple and straightforward while allowing the compiler and the operating system to identify the critical and non-critical parts of the application. When an application (or a thread) is executing critical parts, the OS can dynamically off-load it to a processor with high frequency for the sake of meeting its performance requirement. The OS will also off-load processes/threads that are executing non-critical tasks to lower-frequency processors

which can significantly cut down the power consumption. Moreover, this approach is particularly fit for parallel applications - a prevalent programming pattern in commercial applications (such as databases and web servers). Via the language extension, the OS is aware of the critical thread and non-critical threads in a parallel application, then it can off-load non-critical threads to processors with lower frequency or reducing the frequency of processors that the non-critical threads are running on. The observation is that non-critical threads are likely to run faster than critical threads and have to wait for them to finish (due to the barrier and synchronization requirement). According to researches, migrating non-critical tasks to low-frequency processors can greatly reduce energy. For example, there is a represents an important category of emerging applications such as web search engine, migrating non-critical tasks to lower frequency can achieve more than 40% energy savings without any performance loss on an eight-core system. As we mentioned above, energy cost has became the primary expense in data centuries. Therefore, there is great business va...