Although the future is bright for direct-to-chip two-phase cooling in data centers, the transition to liquid cooling, mostly in the form of direct-to-chip single-phase water, has only just begun.
Stakeholders making up the data center ecosystem have a tough choice to make regarding liquid cooling, needing to solve the here and now but also desiring a liquid cooling technology that is future proof to the next generation of processors.
In this experimental study, we attempt to demonstrate the feasibility of the universal cold plate concept by performance testing a cold plate optimized for single-phase water and subsequently performance testing the identical cold plate with several two-phase working fluids. The study also includes some insights on some of the other requirements that must be met for a universal cold plate.
Comments