Tag
Micron is exploring near-GPU NAND flash technology to enable running larger large language models, potentially improving AI performance and scalability.
Micron states that High Bandwidth Memory (HBM) requires three times more wafer area compared to DDR5 memory.
NVIDIA expands NVLink Fusion with NVHBM, a custom high-bandwidth memory technology that improves memory performance and efficiency for AI workloads, with AWS's Annapurna Labs collaborating as an early partner.
At Hot Chips 2026, Samsung presented opportunities to utilize unused area on HBM base dies after moving to a logic node, with phases including integrating memory controllers to optimize PHY area and power.
A talk at Hot Chips 2026 explores High Bandwidth Flash (HBF) technology and its potential applications in machine learning, focusing on using vLLM to store model weights and KV cache in HBF for improved capacity.
A discussion highlights the significant power delivery problem in tHBM technology, requiring routing of thousands of Amps, and critiques Samsung's zHBM for potential melting issues due to stacking memory on hot GPUs, as revealed by HBM creator Kim Jung-ho.