Daim ntawv thov tshiab 3D Printing Technology hauv Liquid Cooled Plate Field
Cov kua txias yog kim dua li cua txias. Yog li ntawd, muaj ntau txoj kev tshawb fawb txog kev nqis peev loj tshaj plaws thaum ua kev hloov pauv. Cov qauv sab hauv ntawm cov neeg rau zaub mov ua kua txias phaj muaj qhov cuam tshuam loj rau kev ua haujlwm ntawm cov cua sov. Qhov kev tsim kho kom zoo tuaj yeem ua kom qhov chaw sib pauv cua sov ntawm cov phaj txias thiab cov khoom kub xws li CPU lossis GPU, yog li ua kom cov cua sov hloov tau zoo.
Piv txwv li, microchannels los yog fins hauv lub phaj txias tuaj yeem txhim kho qhov sib txawv ntawm cov cua sov, yog li ua tiav cov cua kub zoo dua. Cov qauv ntws thiab turbulence induced yam ntxwv nyob rau hauv lub txias phaj yog ua tib zoo tsim los xyuas kom meej tias lub coolant zoo absorbs thiab tshem tawm tshav kub. Ua kom qhov chaw sib chwv ntau tshaj plaws, ua kom thaj chaw saum npoo, ua kom zoo dua cov qauv ntws, thiab xaiv cov ntaub ntawv thermal conductive uas tsim nyog tuaj yeem txhim kho kev ua haujlwm txias.
Txoj kev ua kom txias txias tam sim no siv nyob rau hauv cov chaw khaws ntaub ntawv yog cov phaj txias, thiab cov ntawv sib xyaw ua kua txias feem ntau siv microchannels nrog 100 micron fins. Hlau additive manufacturing tuaj yeem tsim cov qauv tsim, feem ntau yog tus nqi siab dua li cov microchannels ncaj qha. Txoj kev tsim khoom siv tshuaj ntxiv yog siv los luam ntau cov qauv tsim thiab yuav tsum tau muab cov hmoov tshem tawm ua ntej siv. Los ntawm kev siv nws cov tshuab electrochemical additive manufacturing technology, tsis tas yuav muaj hmoov, yog li nws tuaj yeem siv rau cov kev daws teeb meem txias.
3D luam ntawv tso cai rau cov qauv tsim ntawm cov duab geometric nyuaj nyob rau hauv lub phaj txias, xws li peb lub sij hawm tsawg kawg nkaus (TPMS) lattice microchannels thiab turbulence induced nta. Qhov no tso cai rau kev tsim cov txheej txheem tsim kho kom zoo, ua kom zoo dua qhov kev sib pauv cua sov ntawm cov qauv sab hauv ntawm cov phaj txias thiab cov coolant.
More efficient liquid cooled cold plates can help improve performance and reduce cooling costs, especially as the next generation of chips approaches 500W TDP CPUs. In terms of AI accelerators, we have seen designs for 1kW accelerators per socket. Two CPUs, eight accelerators, along with network and memory, will mean that each node system is>10 kWm. Yuav tsum tau ua kua txias.
