The energy consumption of supercomputers is becoming an ever increasing problem in the high-performance-computing (HPC) world. Some even consider it to be the “one of the greatest challenges in the field of high performance computing” (Balladini, Suppi, Rexachs, & Luque, 2011).
The server’s main memory contributes in two ways to this energy consumption:
This motivates the need for more energy efficient memory systems. One possible solution is hybrid memory, where two different types of memory are combined. This paper explores ways to decide which data is placed in which memory type to maximize energy efficiency and performance.
If you want to create a private blog, a natural choice is WordPress. It can be setup within minutes on the correct webserver. The features you want is usually only one plugin away, you’ve got lots of themes to choose from and writing is easy, thanks to a good media library and a WYSIWYG editor. However, this approach is nothing for the enthusiastic computer scientist. He wants to have complete control over his blog, do some things on his own and use tools he’s already familiar with. This is the setting in which this blog post takes place. A website that uses all the technologies a computer scientist loves, with some aspect of completely unnecessary performance tuning and unneeded redundancy when it comes to serving the website.
Given its size and power consumption, the RaspberryPi is quite an universal talent. When it comes to media consumption, the included processor may lack some computing power if not used properly. For watching video without stuttering, it is required to use a video player that utilizes the build-in h264 decoder. This is not the case when watching Twitch inside the browser. There are dedicated players available that support the hardware decoder. We will use such a player to watch Twitch streams.