AIThe Registerabout 2 hours ago
Qualcomm stacks DRAM on XPUs for AI inference
8 min read
Qualcomm announced high-bandwidth compute (HBC), stacking DRAM on XPUs for unified compute and memory. The AI250-series Dragonfly systems claim 768 GB memory and 133 TB/s effective bandwidth per card. The technology launches next year, aiming to improve inference economics against Nvidia GPUs.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyQualcomm
