Sangwoo Park

I am an Assistant Professor in the Department of Digital Media at Seoul Women's University. Before joining Seoul Women's University, I worked at Samsung Electronics, where I conducted research on AI accelerators, RISC-V CPU design, and RISC-V vector engines. I received my Ph.D. in Engineering from Korea University under the supervision of Prof. Taeweon Suh.

I am looking for prospective PhD and MSc students interested in computer architecture and AI systems research. Interested students are welcome to contact me at psw0113@swu.ac.kr.

International Publications

1. “Performance Characterization and Optimization of LLM Inference on Tenstorrent AI Accelerators,”Jangho Lim, Dongin Shin, Uichan Kim, Jinhyeok Choi, Kyung Min Kim, Sangwon Shin, Gunjae Koo, Taeweon Suh, and Sangwoo Park^†International European Conference on Parallel and Distributed Computing (Euro-Par), August 2026. Accepted.AIArchitectureRISC-V
2. “Three Birds, One Stone: Fast, Accurate-aware and Cost-Efficient Accelerator for Ternary LLM,”Wonseok Jung, Junseok Kang, Sangwon Shin, Hongjun Um, Jangho Lim, Gunjae Koo, Yongjun Park, Taeweon Suh, and Sangwoo Park^†ACM International Conference on Supercomputing (ICS), July 2026. Accepted.AIArchitectureNPU
3. “I²SR: Immediate Interrupt Service Routine on RISC-V MCU to Control mmWave RF Transceivers,”Jimin Lee, Sangwoo Park^*, Junho Huh, Sanghyo Jeong, Inhwan Kim, and Jae Min KimDesign, Automation and Test in Europe Conference (DATE), March 2024.ArchitectureRISC-V
4. “Continual Learning with Speculative Backpropagation and Activation History,”Sangwoo Park^* and Taeweon SuhIEEE Access, Vol. 10, April 2022.AIArchitecture
5. “Speculative Backpropagation for CNN Parallel Training,”Sangwoo Park^* and Taeweon SuhIEEE Access, Vol. 8, November 2020.AIArchitecture
6. “Speculative Backpropagation in Deep Neural Network for Parallel Training,”Sangwoo Park^* and Taeweon SuhDesign Automation Conference (DAC), WIP Session, July 2020.AIArchitecture

Domestic Publications

1. “Cycle-accurate NPU Simulator and Performance Evaluation According to Data Access Strategies,”Guyun Kwon, Sangwoo Park, and Taeweon SuhIEMEK Journal of Embedded Systems and Applications, Vol. 17, No. 4, pp. 217–228, August 2022.AIArchitectureNPU
2. “Design and Evaluation of 32-Bit RISC-V Processor Using FPGA,”Sungyeong Jang, Sangwoo Park, Guyun Kwon, and Taeweon SuhKIPS Transactions on Computer and Communication Systems, Vol. 11, No. 1, pp. 1–8, January 2022.ArchitectureRISC-V
3. “FPGA-based Artificial Neural Network Accelerator Optimization Using Approximate Computing,”Sangwoo Park^*, Hanyee Kim, and Taeweon SuhKorea Information Processing Society Spring Conference, Vol. 26, No. 1, May 2019.AIArchitecture
4. “Optimization of Artificial Neural Network Inference by ReLU Function Prediction,”Sangwoo Park^*, Hanyee Kim, and Taeweon SuhKorea Information Processing Society Fall Conference, Vol. 25, No. 2, November 2018. Best Paper Award.AI

1. “Vector Extension Rotation Operation Architecture,”Sangwoo Park^‡, Sangwon Seo, Keun-Hwi Koo, Jae Eun Shim, Ky-Yeop Lim, and Kang Ji MinSamsung patent, details to be updated.ArchitectureRISC-VNPU
2. “Vector Extension LUT-based Non-linear Function Approximation,”Sangwoo Park^‡, Sangwon Seo, Keun-Hwi Koo, Jae Eun Shim, Ky-Yeop Lim, and Kang Ji MinSamsung patent, details to be updated.ArchitectureRISC-V
3. “ReLU-based Gradient Compression for Large-scale Parallel Training,”Sangwoo Park^‡, Jae Min Kim, and Junho HuhSamsung patent, details to be updated.AI
4. “GeLU-based Gradient Compression for Large-scale LLM Parallel Training,”Sangwoo Park^‡, Jae Min Kim, and Junho HuhSamsung patent, details to be updated.AI
5. “Quantization Method and Apparatus for Artificial Neural Network,”Sangwoo Park^‡ and Taeweon SuhSamsung patent, U.S. Patent Application Publication No. US 2025/0068895 A1, published February 27, 2025; U.S. Application No. 18/811,039, filed August 21, 2024; Korean Application No. 10-2023-0111513, filed August 24, 2023.AIArchitecture
6. “Device and Method for Lifelong Learning Based on Speculative Backpropagation and Activation History,”Sangwoo Park^‡ and Taeweon SuhKorean Patent Application No. 10-2022-0090968, filed July 22, 2022; Korean Patent Publication No. 10-2024-0013438, published January 30, 2024.AI