Read more.Quote:
MIT spinoff's "cooling for modern electronics" uses direct-contact fluid microjet technology.
Printable View
Read more.Quote:
MIT spinoff's "cooling for modern electronics" uses direct-contact fluid microjet technology.
I guess, if it's liquid firing at the surface of the CPU, then the seal around the edge will need to be engineered into the CPU, so that it's a perfect seal for ever. I don't forsee many water tight seals, just pushed onto the surface of a CPU, sealing 100% correctly for the life of a CPU under the pressure needs for the jets to work AND the liquid to then pass back out to the reservoir again. Water oozing out of a seal pressed against the CPU would be ... sub optimal
That thing in the picture dun't look 10x smaller, neither...
we don't know the scale...it might be ickleQuote:
Originally Posted by Ttaskmaster
It's probably that it CAN be 10x smaller. Or that it CAN be 10x better.
As for the sealant, it'll depend on a few factors. These jets probably don't need to be proper jets as we're imagining but just enough to get convection working faster. I don't have Solidworks so I can't download and run their simulation. I expect that you won't install the chip on to the mobo and then the cooler on top, but you'll install the chip into the blue section of the cooler (you can see the multiple sealing nuts around the black edge), be expected to run a pressure and leak test before installing / running and then bolting the whole thing into the socket as one. I'd have thought you'll end up modifying the socket to accept this but it's an enthusiast part after all. I can't see you'll be able to make a seal of the standard required any other way.
I don't think making a water-tight seal against the top of a CPU is an especially tricky task to achieve. We always bolt on heat-sinks with an alarming amount of pressure against the chip. Making something reliably water-tight isn't especially hard - just ask all those people who assemble their own water-cooling rigs.
Oooh, just in time for Intels 10th gen 10 core "105w tdp" at up to 5.2GHz!
10th gen needing 10x cooling :P
We can extrapolate, based on known sizes of CPU block mounting plates... of which there is clearly one in the picture, that must be sized to current CPU cooler mounting points, no? :)Quote:
Originally Posted by Zak33
Size is in three dimensions and so whilst the flat area might be the same size, the volume might be a tenth of the size and there's also the question of what their comparator is.... normal cooling might be a giant heatsink and fan. Or it might be the cooling tower in a power station. That's cooling.Quote:
Originally Posted by Ttaskmaster
what no rgb :D.......... smaller (must be rubbish) :D
yeah but next pahse is 5.2 ghz at stock ! idle 95 watts (we all know what that means......... more like 1k at oc) and this time we incuded hyper threading .........so we have 16 coresQuote:
Originally Posted by Tabbykatze
:D
See also: water jet cutter?
That was also my initial thought, if the jets are high enough pressure, surely the surface of what you are cooling is going to degrade over time? Unless they have some way of dispersing the jet to be more widespread, but even so...?Quote:
Originally Posted by AndyM2012
Flat area provides an estimate of the scale. If their depth is 10x smaller, their drawing is not to scale and woefully unrepresentative. I'm not going to waste my childrens' inheritance on that kind of inaccuracy. I'm out.Quote:
Originally Posted by philehidiot
ah that's a little different. That has a water block thermal paste & clamped to the CPU and the waterblock then holds the liquid, with pipes from it with seals.Quote:
Originally Posted by Dashers
EKWB for example
https://www.ekwb.com/shop/water-blocks/cpu-blocks
they're mainly sealed except where the pipe fitting screw in.
most are from most brands
https://www.scan.co.uk/shop/computer-hardware/cooling-water/cpu-water-blocks
But in this case we're looking at clamping a watertight seal to the CPU directly....which is OK if it's utterly flat, utterly clean CPU and the seal pushed ot it doesn't leak. And then there's the pressure that the coolant will need to squirt into the cavity ..... the seals on the "skirts" of this waterblock will need to be very long lasting.
I think the entire idea is to sqirt water onto the CPU heatspreader, and maybe one day onto the CPU itself
and what happens when the nossles block?
This is huge - it's like a new startup claiming to match the performance of a top-end desktop CPU with a chip that only needs 0.1 W. Convective heat transfer is really hard, so the only way to deal with the heat flux from a modern chip is to massively increase the surface area (by ~3 orders of magnitude for a typical fin stack). With this product offering a convective heat transfer coefficient 3 orders greater than air, you can skip the whole thing, and a lot of heat transfer textbooks will have to be re-written.
Normally, the thermal resistance of the heatspreader and such is negligible when you're cooling a CPU/GPU - it's so hard to get the heat into the fins that the bulk of your temperature difference is there, and so you only see a degree or two across the packaging and the like (for a typical GPU heat flux, ~1C/mm of copper heatspreader thickness out of the 60C between the GPU and the air). With this design you can shed GPU-level heat fluxes (~300 kW/m^2) for ~2 degrees temp differential between the chip and the coolant, which is insane and explains why they're complaining about the thermal resistance of the heatspreader.
Whichever GPU manufacturer jumps on this first will have a field day until the patent runs out. Cooling like this covers a multitude of sins in the silicon design - you'll be able to shed silly amounts of heat as the negligible temperature difference between the coolant and the chip means you can run the coolant hotter, which gets you more heat shed from a given radiator size for a given airflow. It should also massively reduce hotspots in the chip - if a section of chip gets 10 C hotter than the rest of the chip it'll be subjected to ~1 MW/m^2 of cooling locally, so will not sustain the temperature difference for long
Those use abrasive grit in the waterQuote:
Originally Posted by AndyM2012
You swear, curse, and then strip everything down over the course of three consecutive evenings.Quote:
Originally Posted by Tattysnuc
Phew! Good job there won't be any cavitation to generate grit.Quote:
Originally Posted by Xlucine
While I agree that it appears to be a great idea, I'd want to see how the chips stand up to it over a few years in real life.
There shouldn't be, so long as the pump isn't pulling a vacuum in the heat transfer chamberQuote:
Originally Posted by AndyM2012