TokenStack: A Heterogeneous HBM-PIM Architecture and Runtime for Efficient LLM Inference

Documenti analoghi

• Not All Thoughts Need HBM: Semantics-Aware Memory Hierarchy for LLM Reasoning
  di: Yuan, Aojie, et al.
  Pubblicazione: (2026)
• Splitwise: Efficient generative LLM inference using phase splitting
  di: Patel, Pratyush, et al.
  Pubblicazione: (2023)
• Tasa: Thermal-aware 3D-Stacked Architecture Design with Bandwidth Sharing for LLM Inference
  di: He, Siyuan, et al.
  Pubblicazione: (2025)
• Lean Attention: Hardware-Aware Scalable Attention Mechanism for the Decode-Phase of Transformers
  di: Sanovar, Rya, et al.
  Pubblicazione: (2024)
• GainSight: A Unified Framework for Data Lifetime Profiling and Heterogeneous Memory Composition
  di: Li, Peijing, et al.
  Pubblicazione: (2025)