jaykrown@lemmy.world to Technology@lemmy.worldEnglish · 2 days agoDDR6 RAM shrinks by 44% while delivering 4X DDR5's baseline bandwidth — JEDEC previews 12.8 GT/s modules with up to 1 terabyte of capacity per slotwww.tomshardware.comexternal-linkmessage-square119linkfedilinkarrow-up1385arrow-down16
arrow-up1379arrow-down1external-linkDDR6 RAM shrinks by 44% while delivering 4X DDR5's baseline bandwidth — JEDEC previews 12.8 GT/s modules with up to 1 terabyte of capacity per slotwww.tomshardware.comjaykrown@lemmy.world to Technology@lemmy.worldEnglish · 2 days agomessage-square119linkfedilink
minus-squaremantricx@lemmy.worldlinkfedilinkEnglisharrow-up16·1 day agoSo many non-gaming workloads these days are capacity driven more than bandwidth driven. Any AI, LLM, image generation, video editing, etc. would be much better off with 32GB of DDR3 than 8GB of DDR4/5. 32GB of DDR4 is a good sweet spot to be in today.
minus-squareaksdb@lemmy.worldlinkfedilinkEnglisharrow-up11·edit-21 day agoLLM workloads become faster with higher bandwidth, which directly translates to token per second. So it does matter for daily use. But you are right that more and slightly slower is still better than less but fast.
minus-squareaim_at_me@lemmy.nzlinkfedilinkEnglisharrow-up1·17 hours agoI wish I bought 64gb when I built my rig.
minus-squareboonhet@sopuli.xyzlinkfedilinkEnglisharrow-up6·1 day ago32 GB of DDR4 here, would much rather have more and faster tbh. LLMs are slow when running cpu-moe mode or Strata
So many non-gaming workloads these days are capacity driven more than bandwidth driven. Any AI, LLM, image generation, video editing, etc. would be much better off with 32GB of DDR3 than 8GB of DDR4/5.
32GB of DDR4 is a good sweet spot to be in today.
LLM workloads become faster with higher bandwidth, which directly translates to token per second. So it does matter for daily use.
But you are right that more and slightly slower is still better than less but fast.
I wish I bought 64gb when I built my rig.
32 GB of DDR4 here, would much rather have more and faster tbh. LLMs are slow when running cpu-moe mode or Strata