#hbm
Hbm: 12 AI articles covering hbm news, analysis, and research
Articles
Vertical DRAM Stacking Expands Thermal Design Space Beyond HBMโญ8
imec researchers show vertical 3D DRAM stacking could expand thermal design space beyond HBM-on-GPU, easing thermal bottlenecks in future AI GPU memory architec...
Scaling Data Movement for Physical AI: The Interconnect andโญ9
Physical AI faces a data movement bottleneck: discover how interconnect and memory bandwidth constraints shape the scaling of robots, autonomous vehicles, and e...
Inside Samsung's Massive New Semiconductor Hub: Building Fabsโญ7
Inside Samsung's massive new semiconductor hub at the world's largest chipmaking complex, the company is building fabs to meet surging AI memory demand for high...
Breaking the Memory Bottleneck in AI and HPC SoCsโญ10
Discover how AI and HPC SoCs are overcoming the memory bottleneck with HBM4, advanced DRAM, and new architectures for higher bandwidth and efficiency.
Finding Critical Defects Before They Become Costly Failures:โญ10
Hybrid bonding and advanced packaging process control: detect critical defects like voids, misalignment, and contamination before they cause costly multi-die fa...
Make Your Interface Work For Your AI SoCโA 10-Step Program: IPโญ9
As AI SoCs hit bandwidth limits, our 10-step program guides architects through HBM4, UCIe 2.0, NoC, and chiplet IP choices to build an optimal interface.
Concurrent HBM and Host Memory Access Boosts LLM Inferenceโญ9
Georgia Tech, Nvidia, and Stanford show that concurrent HBM and host memory access boosts LLM inference throughput by hiding latency and scaling beyond HBM capa...
Reducing HBM ECC Controller Overhead for AI Inferenceโญ10
HBM ECC adds bandwidth and latency overhead for AI inference. Learn how on-die ECC and smarter controller design reduce that cost in high-bandwidth AI accelerat...
Heterogeneous Memory Chiplets Accelerate Multi-Request LLMโญ9
NUS researchers propose heterogeneous memory chiplets that combine high-bandwidth, high-capacity, and near-memory compute to accelerate multi-request LLM infere...
Samsung Chip Workers Defect to Rival SK Hynix as AI Talent Warโญ9
As AI chip competition heats up, Samsung chip workers defect to SK Hynix for better pay, HBM-dominated growth, and career advancement, intensifying the global t...