AI Hardware
AI Hardware: From Chip to System
Here are the results of our research on AI hardware:
1) Neural Processing Unit (NPU)
BitBlade: Variable Bit-Precision Hardware
Abstract-
Related Publications
Sungju Ryu, Hyungjun Kim, Wooseok Yi, Eunhwan Kim, Yulhwa Kim, Taesu Kim, Jae-Joon Kim, "BitBlade: Energy-Efficient Variable Bit-Precision Hardware Accelerator for Quantized Neural Networks," IEEE Journal of Solid-State Circuits (JSSC), Accepted for publication (Early access).
Sungju Ryu, Hyungjun Kim, Wooseok Yi, Jongeun Koo, Eunhwan Kim, Yulhwa Kim, Taesu Kim, Jae-Joon Kim, "A 44.1TOPS/W Precision-Scalable Accelerator for Quantized Neural Networks in 28nm CMOS," IEEE Custom Integrated Circuits Conference (CICC), Mar. 2020.
Sungju Ryu, Hyungjun Kim, Wooseok Yi, Jae-Joon Kim, “BitBlade: Area and Energy-Efficient Precision-Scalable Neural Network Accelerator with Bitwise Summation,” ACM/IEEE Design Automation Conference (DAC), Jun. 2019. (BK21+ Computer Science분야 우수국제학술대회)
Mobileware: Hardware for Depthwise Separable Convolution
Abstract-
Related Publications
Sungju Ryu, Youngtaek Oh, Jae-Joon Kim, "Mobileware: A High-Performance MobileNet Accelerator with Channel Stationary Dataflow," IEEE/ACM International Conference on Computer-Aided Design (ICCAD), Nov. 2021. (BK21+ Computer Science분야 우수국제학술대회)
SPRITE: Sparsity-Aware NPU
Abstract-
Related Publications
Sungju Ryu, Youngtaek Oh, Taesu Kim, Daehyun Ahn, Jae-Joon Kim, "SPRITE: Sparsity-Aware Neural Processing Unit with Constant Probability of Index-Matching," Design Automation and Test in Europe (DATE), Feb. 2021. (BK21+ Computer Science분야 우수국제학술대회)
2) Processing-in-Memory (PIM)
Near-Memory Computing on 3D HBM-like Memory
Abstract-
Related Publications
Naebeom Park, Sungju Ryu, Jaeha Kung, Jae-Joon Kim, "High-Throughput Near-Memory Processing on CNNs with 3D HBM-like Memory," ACM Transactions on Design Automation of Electronic Systems (TODAES), Jun. 2021.
Fast Online Learning on 6T SRAM Array
Abstract-
Related Publications
Jongeun Koo, Jinseok Kim, Sungju Ryu, Chulsoo Kim, Jae-Joon Kim, “Area-Efficient Transposable 6T SRAM for Fast Online Learning in Neuromorphic Processors,” IEEE Custom Integrated Circuits Conference (CICC), Apr. 2019.
3) Datacenter AI
Datacenter AI Hardware
In Progress..
Related Publications
