There is SO MUCH single-phase liquid immersion cooling misinformation in this presentation it's hard to know where to begin. So let's tackle this one bite at a time. Yes, there is more coolant in a single-phase liquid immersion cooling system (SPLIC) but the coolant last 10+ years with no degradation or volume loss. The same cannot be said for cold plate refrigerant, or two-phase immersion. Your performance figures are absolutely false. SPLIC performance is far beyond 250W processor TDP limitations. A 1U server can have X number of sockets to affix CPUs (let's say two) meaning your 1U number would theoretically have two CPUs @ 425W TDP ea. = 850W (a ridiculously high TDP, BTW). SPLIC runs 400W TDP CPUs without performance loss (so, 800W), and analysis shows it can cool CPUs beyond 750W TDP CPUs, which would be 1,500W in your example. The reason I am calculating two CPUs, is that each would need to be plumbed in a 1U scenario. That's a lot of inlets and outlets. Your number for PUE is false. SPLIC is
There is SO MUCH single-phase liquid immersion cooling misinformation in this presentation it's hard to know where to begin. So let's tackle this one bite at a time. Yes, there is more coolant in a single-phase liquid immersion cooling system (SPLIC) but the coolant last 10+ years with no degradation or volume loss. The same cannot be said for cold plate refrigerant, or two-phase immersion.
Your performance figures are absolutely false. SPLIC performance is far beyond 250W processor TDP limitations. A 1U server can have X number of sockets to affix CPUs (let's say two) meaning your 1U number would theoretically have two CPUs @ 425W TDP ea. = 850W (a ridiculously high TDP, BTW). SPLIC runs 400W TDP CPUs without performance loss (so, 800W), and analysis shows it can cool CPUs beyond 750W TDP CPUs, which would be 1,500W in your example. The reason I am calculating two CPUs, is that each would need to be plumbed in a 1U scenario. That's a lot of inlets and outlets.
Your number for PUE is false. SPLIC is