The Picard Maneuver@lemmy.worldM to memes@lemmy.world · 17 days agoThey're all vitallemmy.worldimagemessage-square138fedilinkarrow-up11.24Karrow-down125
arrow-up11.21Karrow-down1imageThey're all vitallemmy.worldThe Picard Maneuver@lemmy.worldM to memes@lemmy.world · 17 days agomessage-square138fedilink
minus-squareHonytawklinkfedilinkarrow-up6·16 days agoI mean, you can never have too much ram. But if you don’t use it in its entirety, it is kind of useless.
minus-squareAnUnusualRelic@lemmy.worldlinkfedilinkarrow-up1·16 days agoThat’s the system’s job. It can cache stuff or whatever.
minus-squarePossibly linuxlinkfedilinkEnglisharrow-up1·edit-216 days ago192GB is way to big to be used as a cache However, you probably could run some of the huge language models
minus-squareAnUnusualRelic@lemmy.worldlinkfedilinkarrow-up2·16 days agoYou’re underestimating the cache.
minus-squarePossibly linuxlinkfedilinkEnglisharrow-up1·16 days agoYou are underestimating writebacks. You won’t use that much on a normal filesystem and with ZFS you will only use that with bigger amounts of data
I mean, you can never have too much ram.
But if you don’t use it in its entirety, it is kind of useless.
That’s the system’s job. It can cache stuff or whatever.
192GB is way to big to be used as a cache
However, you probably could run some of the huge language models
You’re underestimating the cache.
You are underestimating writebacks. You won’t use that much on a normal filesystem and with ZFS you will only use that with bigger amounts of data