Metrics
Eurotech HPC division encourages the usage of some metrics in order to let our customers evaluate the solution we propose.
The most important are
TCO, PUE, ERE and CUE.
TCO
TCO means total cost of ownership and equals the full cost of a solution during its lifetime including the cost of purchase, maintenance, support, energy consumed and disposal. In the recent years, the cost of energy has been affecting TCO almost at the same level than the initial purchase. Also, there are capital costs, related to space occupancy, which weight quite considerably in the final TCO calculation.
Aurora supercomputers benefit from a TCO that makes them economically wise investments. Energy efficiency, density, liquid cooling, reliability are all factors that contribute to low maintenance cost and less non IT related capital investments. For instance, Aurora HPC systems allow electricity bills savings that recover a distinctive percentage of the initial cost of purchase. At European average electricity prices per Kwh, it would be enough to have 1 Aurora rack to save as much as 500,000 € (750,000$) in 5 years.
PUE
PUE stands for power usage effectiveness. It measures how much of the electrical power entering a data center is effectively used for the IT load, which is the energy absorbed by the server and usefully used to compute. The definition of PUE in formula is as follows:
The perfect theoretical PUE is equal to 1. Average datacenters have nowadays a PUE of 2.13.
A datacenter based on Aurora clusters can reach a PUE of 1.05. This can be achieved by hot water cooling, high efficient power conversion, optimal synchronization and, optionally, the use of 3D Torus.
ERE
The temperature gap between inlet and outlet water flow in each Aurora rack is between 3C and 5C. Putting in series more racks and considering that the coolant can reach more than 50 C, it is possible to add these temperature steps to get an overall temperature gap, between datacenter inlet and outlet, which is enough to heat a building or be used in industrial processes. This technique is called thermal energy reuse and implies PUE<1 which mathematically makes no sense.
Energy resue calls for a new metric, ERE, to better capture the idea of energy saved in the overall datacenter balance.
The ERE is defined as follows:
CUE
CUE stands for carbon usage effectiveness. It is a Green Grid metric and stands for carbon usage effectiveness. It measures: the total CO2 emissions caused by the total data center divided by the IT load, that is the energy consumed by the servers. The formulae can be expressed as follows:
That means:
CEF is the carbon emission factor (kgCO2eq/kWh) of the site, based on the government’s published data for the region of operation for that year.
The CEF depends on the energy production mix that ultimately feeds the datacenter. It could very low, for instance, for a datacenter entirely powered by a hydroelectric plant. In reality, the CEF changes from country to country. In USA it changes from state to state. The USA average is 0,59 kgCO2eq/kWh. Using this value it possible to estimate that in 5 years of operation an Aurora rack could save as much as 1500 tons of CO2 compared to a standard 2U server rack.