| bunchanumbers said: Who was it that was working on the stacked memory? Was it AMD? |
Both AMD and NVidia where, and probably Intel, too. They use it differently, though.
NVidia wants to use HBM (High Bandwith Memory) on their Tegra Chips as on-chip memory, eleminating the need for additional memory chips. No infos about using it on future Graphics Cards yet, though.
AMD wants to use HBM in 2 different ways. First as a high bandwith Last-Level-Cache on their APUs with the size of 1GiB, and second as the main VRAM on their graphics cards (and since the first generation of HBM only goes up to 4GiB, that will be the maximum the 3xx Graphics cards will have as VRAM)
Intel is probably also checking into stacked RAM technology (possibly Hybrid Memory Cube, HMC, instead of HBM), as it is both cheaper and more efficient as embedded DRAM (eDRAM) for use in their IRIS Graphics chips.
