PARALLEL SESSION
A4L-A
ML/AI Accelerator Systems with NVM and eDRAM
TUESDAY - 9 September | 13:50 - 15:55
CHAIRS
Antoine Frappe, Takashi Maeda
SESSION PROGRAM
13:50 - 14:15
A 28nm 51.6TOPS/W 2.22Mb/mm² System-in-One-Macro Computing-in-Memory Chip Utilizing Leakage-Eliminated 2T1C and Capacitor-Over-Logic 1T1C eDRAM
Zhaori Cong, Zhihang Qian, Shengzhe Yan, Zeyu Guo, Zhuoyu Dai, Yifan He, Wenyu Sun, Chunmeng Dou, Feng Zhang, Jinshan...
14:15 - 14:40
30.7 TOPS/W Sparsity-Aware Analog-Digital Hybrid eDRAM CIM by Effective Row Activation with Simultaneous Multi-Row-Multi-Task Control
Hoichang Jeong, Seungbin Kim, Jueun Jung, Kyuho Lee
14:40 - 15:05
D3TA: 38.9TOPS/W Transformer Accelerator with Dual-Port 3T-eDRAM Digital Compute-in-Memory Using HyperAttention and Triple-Sparsity-Handling
Donghyuk Kim, In-Jun Jung, Geonwoo Ko, Seri Ham, Cuong Duong Manh, Gyeongrok Yang, Joo-Young Kim
15:05 - 15:30
Falcon: An Event-Driven Object Tracking SoC with Activation-Skip and Weight-Compression MRAM PIM and Heterogenous Q/K/V Read-Only-Once Attention Flow
Wenao Xie, Haoyang Sang, Zhamaliddin Kalzhan, Jingu Lee, Jerald Yoo, Kangho Lee, Hoi-Jun Yoo
15:30 - 15:55
28M Weights×TOPS/W/mm² PCM-Based Analog In-Memory Computing Core with 8 512×512-Weight Layers in 28nm FD-SOI CMOS
Marco Pasotti, Riccardo Zurla, Jacopo John Bertolini Agnoletto, Emanuela Calvetti, Alessio Antolini, Luigi Croce, Giu...

