Canada Condemns Executions of Four Citizens in China Amid Diplomatic Strain

A team at the University of Waterloo has developed a method to reduce energy consumption in data centers by up to 30%, a significant achievement given that computing consumes approximately 5% of the world's daily energy supply. The researchers, led by computer science professor Martin Karsten and graduate student Peter Cai, restructured the data processing flow, improving the efficiency of how data packets are sorted within data centers, which typically operate on Linux software.
Karsten described the change as "rearranging the pipeline at a manufacturing plant," stating that it optimizes the use of CPU caches without adding any new components. The solution, which involves about 30 lines of code, has been integrated into the latest Linux software update and can be adopted by major tech companies such as Amazon, Google, and Meta.
The implications of this innovation are vast, particularly given the rising demands of AI technology. Karsten noted that activating this protocol could lead to substantial reductions in energy use globally, benefiting both businesses and the environment.