misk@sopuli.xyz to Hardware@lemmy.world · 14 hours agoNvidia's data center Blackwell GPUs reportedly overheat, require rack redesigns and cause delays for customerswww.tomshardware.comexternal-linkmessage-square6fedilinkarrow-up128arrow-down10cross-posted to: technology@lemmy.worldnvidia@lemmy.zip
arrow-up128arrow-down1external-linkNvidia's data center Blackwell GPUs reportedly overheat, require rack redesigns and cause delays for customerswww.tomshardware.commisk@sopuli.xyz to Hardware@lemmy.world · 14 hours agomessage-square6fedilinkcross-posted to: technology@lemmy.worldnvidia@lemmy.zip
minus-squareRekall Incorporatedlinkfedilinkarrow-up3·6 hours ago 120kW per rack I knew GPU compute took a lot of energy, but I didn’t realize it was 120 kW per rack. That is a stupid amount.
I knew GPU compute took a lot of energy, but I didn’t realize it was 120 kW per rack. That is a stupid amount.