Is a lack of standards holding immersion cooling back? Liquid and immersion cooling have undergone something of a renaissance in the datacenter in recent years as components have grown ever hotter. This trend has only accelerated over the past few months as we’ve seen a fervor of innovation and development around everything from liquid-cooled servers and components for vendors …
With the era of AI writing parametrized product descriptions (A/B testing and optimising for conversions using various psychological manipulation strategies), the only way to actually find the differences and if the product suits you is to order it, "give it a go" and then send it back if it doesn't work as expected.
What impact does it have on the environment? Let's not think about that.
In my opinion should practice should be illegal.
External dry cooler is a term of art in industrial cooling systems as opposed to a chiller.
Examples in our real lives - fans on a PC with liquid cooled chips are external dry coolers while refrigerators are chillers but neither works at the data center level. Currently data centers use chillers, commonly referred to as air conditioning, which have a lot of downsides.
Detailed specifications for all the listed alternatives are readily available if you're in the market for one.
Heat exchangers / refrigeration / cooling are things that are in widespread use across many industries, and found nearly everywhere that can be described as a factory in the manufacturing sector.
Why not go and talk to them about how best to do it, instead of thinking you need to reinvent the wheel just because it's new shiny cloud tech?
>Why not go and talk to them about how best to do it, instead of thinking you need to reinvent the wheel just because it's new shiny cloud tech?
You could do both and patent it!
Dyson effective did this with its cyclone vacuum cleaner.
The use of cyclone technology wasn't new, Dyson took it from the wood mill, did a lot of R&D and miniaturised it and put it in a vacuum cleaner.
I think you will find Hoover et al lost and had to pay Dyson royalties...
That specific patent has expired, hence why there are now so many cyclone-based vacuum cleaners from manufacturers other than Dyson.
However, Dyson have continued to innovate and create a forest of patents around their products.
Dyson invented and patented the cyclone technology. As there were no special tools or tech required to manufacture it, it was promptly stolen and regurgitated by everybody from Hoover to XiangJiang Shenzhen. Dyson very nearly went bankrupt trying to fight the counterfeiters, got some royalties but ultimately decided to move on and invent other stuff.
Liquid cooling for datacenters could produce higher-quality waste heat for growing tomatoes, or heating (or even cooling) houses and offices. But people seldom think even of this. So I'm hoping that non-Intel (and non-Nvidia, non-Radeon) approaches will drive energy requirements down.
I got a tour of a new university datacentre not long ago and asked about recycling the heat. Apparently the techies wanted to do this; but the architects asked if they could guarantee that the DC would still be in use and similar in a couple of decades time. Sadly the heat now just dumped outside in the city...
You need to get 1000W of heat out of a couple of square cm of CPU/GPU in the middle of a box other parts, while keeping the hot end below 80C.
Air doesn't have much heat capacity, a little bit of energy heats the air above the temperature of the chip, at which point no more heat flows. So you need to start with the air really-really cold and blow lots-and-lots of air passed the chip.
Getting lots-and-lots of really-really cold air and delivering it to the chip and removing it afterwards is bloody expensive.
In theory it's much easier to have a pipe with outside temperature water going in, and 80C water coming out, through a radiator and being cooled to the outside temperature again
"I think some of the Cray supercomputers were designed that way"
Indeed. The CDC 6x00 and (I think) the 7x00 were water cooled. An engineer asked me to shut down my testing and go out and enjoy the sunshine one Saturday afternoon in Minnesota Seems the CDC 6400 we were developing ABM benchmark software on had sprung a leak and he needed to fix the plumbing.
Some of us who worked at a certain tea-to-luxury 4x4s adjacent consultancy well remember when we had a rack of hard drives park outside the server room for six months because the excellent* managers hadn't taken the weight of the drives themselves into account when speccing the floor....I dread to think what would have happened if they'd tried their hand at water cooling.
But the sheer noise from the many windy noise generators has annoyed the neighbours so much that we were pushed towards liquid cooling/immersive cooling. This just has the problem of not performing reliably for the months needed to ‘solve’ whatever cryptic problem was loaded. This short-term reliability problem will be overlayed with the long term reliability issues that will inevitably arise. tanstaafl applies.
Solution: keep simple, use air, place the systems where there are few neighbours….
The Register - Independent news and views for the tech community. Part of Situation Publishing
Biting the hand that feeds IT © 1998–2022
