stumu415@lemmy.zip to Technology@lemmy.worldEnglish · 1 month agoJensen Huang says Nvidia now has 'zero percent' market share in China — says US export policy 'has already largely backfired'www.tomshardware.comexternal-linkmessage-square132linkfedilinkarrow-up11arrow-down10
arrow-up11arrow-down1external-linkJensen Huang says Nvidia now has 'zero percent' market share in China — says US export policy 'has already largely backfired'www.tomshardware.comstumu415@lemmy.zip to Technology@lemmy.worldEnglish · 1 month agomessage-square132linkfedilink
minus-squarefrongt@lemmy.ziplinkfedilinkEnglisharrow-up0·1 month agoThere are a bunch, and you can’t buy them retail as far as I know, but I’m sure you can find them on aliexpress and eBay. But they’re currently several years behind current nvidia chips. It’ll take a couple more before they’re globally competitive.
minus-squarefuckwit_mcbumcrumble@lemmy.dbzer0.comlinkfedilinkEnglisharrow-up0·1 month agoBut what they do have is fuck tons of vram which is very important for AI workloads.
minus-squareRioting Pacifist@lemmy.worldlinkfedilinkEnglisharrow-up0·1 month agoSeveral years behind in what sense? AI workloads in datacenters don’t care if the silicone is slightly bigger or less efficient, an AI workload doesn’t need fancy instruction sets. for use in desktops or laptops, sure NVIDIA can be years ahead but for AI stuff do their improvements really matter?
minus-squarezbyte64@awful.systemslinkfedilinkEnglisharrow-up0·1 month agoWell we all don’t have a government pursuing renewable that makes power costs predictably drop over time. Efficiency matters when I’m looking at my power bill thay went up because of a near by data center being built
There are a bunch, and you can’t buy them retail as far as I know, but I’m sure you can find them on aliexpress and eBay.
But they’re currently several years behind current nvidia chips. It’ll take a couple more before they’re globally competitive.
But what they do have is fuck tons of vram which is very important for AI workloads.
Several years behind in what sense?
AI workloads in datacenters don’t care if the silicone is slightly bigger or less efficient, an AI workload doesn’t need fancy instruction sets.
for use in desktops or laptops, sure NVIDIA can be years ahead but for AI stuff do their improvements really matter?
Well we all don’t have a government pursuing renewable that makes power costs predictably drop over time. Efficiency matters when I’m looking at my power bill thay went up because of a near by data center being built