A 52.03TOPS/W DCIM-Based Accelerator with FlashAttention and Sparsity-Aware Alignment for LLMs

Bo Liu 0019, Xingyu Xu 0008, Yang Zhang, Xilong Kang, Qingwen Wei, Zihan Zou, Jun Yang 0006, Hao Cai 0001, Xin Si. A 52.03TOPS/W DCIM-Based Accelerator with FlashAttention and Sparsity-Aware Alignment for LLMs. In IEEE Custom Integrated Circuits Conference, CICC 2025, Boston, MA, USA, April 13-17, 2025. pages 1-3, IEEE, 2025. [doi]

References

No references recorded for this publication.

Cited by

No citations of this publication recorded.