July 27th 2016 Green Grid Forum: Introduction of liquid-cooled IT and maximizing waste heat reuse (Japanese)

The widespread adoption of smartphones and the proliferation of connected smart systems in cars, homes, wearable devices, and other devices are driving ever-increasing demands on computing capacity and network bandwidth. Over the past decade, data centers, including hyperscale data centers, have pursued horizontal data center footprint growth while reducing rack power levels.There are some signs of change in this direction. For example, virtual reality applications employed by social media companies increasingly rely on high-power GPUs, which when densely packed into chassis and racks require high rack power and efficient air cooling. It's difficult. This problem is a little more obvious in the field of high performance computing (HPC) HPC applications require low latency. Therefore, all servers/racks in a cluster must be close to each other. Server/chassis/rack density is primarily determined by latency constraints. As we move from petascale to exascale computing, rack power densities are expected to continue to soar, given data center space and power limitations.