Georgia Tech unveils direct-to-silicon liquid cooling
October 6, 2015 | 11:49
Companies: #georgia-institute-of-technology #georgia-tech
Researchers at the Georgia Institute of Technology have announced a new era of liquid cooling, with the first silicon chips which can be directly integrated into a cooling loop without the need for a water block.
Liquid cooling is a common sight in the enthusiast market, pulling heat away from high-powered CPU and GPU chips in water-filled tubes and taking it to a radiator which offers a far larger surface area than would be possible with direct air cooling over a traditional heatsink. Traditionally, this is achieved with the use of a waterblock - a sealed heatsink which draws the heat away from the chip and into the water which flows through its centre. The chip seen in Georgia Tech's lab, though, is different: it connects directly to the liquid loop with no waterblock in sight.
'We believe we have eliminated one of the major barriers to building high-performance systems that are more compact and energy efficient,' claimed Muhannad Bakir, an associate professor in the Georgia Tech School of Electrical and Computer Engineering, of his team's research. 'We have eliminated the heat sink atop the silicon die by moving liquid cooling just a few hundred microns away from the transistors. We believe that reliably integrating microfluidic cooling directly on the silicon will be a disruptive technology for a new generation of electronics.'
The system works by etching microfluidic channels directly into the silcon of a standard, off-the-shelf semiconductor - an Altera FPGA built on a 28nm process node - after removing the traditional metal heatspreader. Each channel features silicon cylinders around 100 microns in diameter to boost the surface area, and is then covered over in more silicon to provide a water-tight closure. Fitting ports to the top, it's then possible to connect the chip directly to standard liquid cooling systems.
The results were impressive: compared to cooling the same chip with air, the liquid-cooled FPGA ran 60 per cent cooler. It was enough to please the Defence Advanced Research Projects Agency (DARPA), which funded the research, and its creators believe the technique has real commercial value for cooling everything from CPUs and GPUs through to power amplifiers. 'We have created a real electronic platform to evaluate the benefits of liquid cooling versus air cooling,' explained Bakir. 'This may open the door to stacking multiple chips, potentially multiple FPGA chips or FPGA chips with other chips that are high in power consumption. We are seeing a significant reduction in the temperature of these liquid-cooled chips.'
More information on the project is available on the Georgia Tech website.
Liquid cooling is a common sight in the enthusiast market, pulling heat away from high-powered CPU and GPU chips in water-filled tubes and taking it to a radiator which offers a far larger surface area than would be possible with direct air cooling over a traditional heatsink. Traditionally, this is achieved with the use of a waterblock - a sealed heatsink which draws the heat away from the chip and into the water which flows through its centre. The chip seen in Georgia Tech's lab, though, is different: it connects directly to the liquid loop with no waterblock in sight.
'We believe we have eliminated one of the major barriers to building high-performance systems that are more compact and energy efficient,' claimed Muhannad Bakir, an associate professor in the Georgia Tech School of Electrical and Computer Engineering, of his team's research. 'We have eliminated the heat sink atop the silicon die by moving liquid cooling just a few hundred microns away from the transistors. We believe that reliably integrating microfluidic cooling directly on the silicon will be a disruptive technology for a new generation of electronics.'
The system works by etching microfluidic channels directly into the silcon of a standard, off-the-shelf semiconductor - an Altera FPGA built on a 28nm process node - after removing the traditional metal heatspreader. Each channel features silicon cylinders around 100 microns in diameter to boost the surface area, and is then covered over in more silicon to provide a water-tight closure. Fitting ports to the top, it's then possible to connect the chip directly to standard liquid cooling systems.
The results were impressive: compared to cooling the same chip with air, the liquid-cooled FPGA ran 60 per cent cooler. It was enough to please the Defence Advanced Research Projects Agency (DARPA), which funded the research, and its creators believe the technique has real commercial value for cooling everything from CPUs and GPUs through to power amplifiers. 'We have created a real electronic platform to evaluate the benefits of liquid cooling versus air cooling,' explained Bakir. 'This may open the door to stacking multiple chips, potentially multiple FPGA chips or FPGA chips with other chips that are high in power consumption. We are seeing a significant reduction in the temperature of these liquid-cooled chips.'
More information on the project is available on the Georgia Tech website.
MSI MPG Velox 100R Chassis Review
October 14 2021 | 15:04
Want to comment? Please log in.