Thread

1. fooker · 1789929721 · focus · HN ↗

   What's the main blocker (other than the current inflated cost) for using HBM instead of DRAM as the primary memory for consumer electronics?

   1. chessgecko · 1789930440 · focus · HN ↗

      I think people might prefer the lower idle power consumption from lpddr over the better bandwidth in hbm in battery powered stuff. That said right now the price is definitely preventing us from finding out.

      1. fooker · 1789930658 · focus · HN ↗

         Idle yes, but HBM energy consumption / memory operations seems to be a bit better than DRAM.

         1. vlovich123 · 1789930837 · focus · HN ↗

            Only if you’re running at 100%. Consumers generally do not.

            1. monster_truck · 1789931758 · focus · HN ↗

               That hasn't been true since early HBM2 days, before the controllers standardized on power/voltage management and did things like leave them in P0 to ship on time

               1. vlovich123 · 1789939050 · focus · HN ↗

                  LPDDR/DDR/GDDR generally still win over HBM when there’s no data being transferred. HBM is primarily better in watts/byte transferred. Consumer electronics spend most of their time idle.

                  1. Zagitta · 1789940656 · focus · HN ↗

                     Racing to idle is a very common power optimization technique

                     1. vlovich123 · 1789942734 · focus · HN ↗

                        Right, but HBM idle is significantly worse than LPDDR idle or even DDR idle for that matter. That matters a lot precisely because the device is idle most of the time. Your idle power draw dominates.

                        1. monster_truck · 1789967041 · focus · HN ↗

                           Comparing HBM to LPDDR is pretty silly. That's like a bath tub vs a urinal

                           1. imtringued · 1789986364 · focus · HN ↗

                              If LPDDR6 PIM ever becomes a thing then the advantage of HBM will shrink. I'm not saying LPDDR6 PIM will make HBM obsolete in the datacenter or where it is currently shining, I'm saying that large volumes have their own charm and it is more likely for LPDDR6 PIM to be in your laptop or smartphone or SBC than HBM.

                              LPDDR6 PIM would primarily help the low end and mid range accelerator market. E.g. embedded models running on SBCs can be up to 1 GiB in size with acceptable performance, small models at 8 GiB become really easy to run at reasonable speeds on a smartphone and PIM enabled laptops or mini PCs make it possible to run 32 GiB models locally without compromise.

                              Of course this also assumes that the associated accelerators (NPUs) will catch up too, but the general point is that you won't need a 5090 or a 4090 anymore.

                           2. vlovich123 · 1790001604 · focus · HN ↗

                              This is literally a thread where people are saying “but why laptops no HBM”. It’s a perfectly cromulent response that meets the question where it is. It might be mildly a more defensible question for normal plugged in desktops but that would require a massive ecosystem change since people who build those traditionally really like their DIMMs and less buying a CPU that has non-swappable RAM built in, not to mention there’s no sustainable market for the higher cost and low volume product.

                              1. monster_truck · 1790271970 · focus · HN ↗

                                 No, it really isn't. You're barking up the wrong trees for incorrect reasons

Unofficial Hacker News client; not affiliated with Y Combinator.