Career Profile

I completed my Ph.D. at Yonsei University, Korea, with a dissertation titled “GPU Architecture Design for Effective Computing Resource Usage” under the guidance of Professor Won Woo Ro. My expertise lies in designing computer architectures from general-purpose processing units to domain-specific accelerators, with a focus on optimizing computing resource utilization. My recent work focuses on optimizing GPUs’ computing resource usage and democratizing various domain-specific accelerators. I am currently a system architect at MangoBoost Inc., exploring the RDMA engine architecture to improve MangoBoost DPUs further.

  • Fast learner
  • Resilient personality
  • Independent and highly motivated
  • Highly reliable team player or leader
  • Professional communication, writing, and presentation skills

Professional Experiences

University of California, Riverside, USA (Advisor: Hung-Wei Tseng)

Education

PhD. in the School of Electrical and Electronic Engineering

2024
Yonsei University, Seoul, Korea (Advisor: Won Woo Ro)

Dissertation: GPU Architecture Design for Effective Computing Resource Usage

M.S. in the School of Electrical and Electronic Engineering

2021
Yonsei University, Seoul, Korea (Advisor: Won Woo Ro)

B.S. in the School of Electrical and Electronic Engineering

2019
Yonsei University, Seoul, Korea

Publications

M3XU: Achieving High-Precision and Complex Matrix Multiplication with Low-Precision MXUs
Dongho Ha*, Yunan Zhang*, Chen-Chien Kao, Christopher J. Hughes, Won Woo Ro, and Hung-Wei Tseng
The International Conference for High Performance Computing, Networking, Storage, and Analysis (SC 2024)
Generalizing Ray Tracing Accelerators for Tree Traversals on GPUs
Dongho Ha*, Lufei Liu*, Yuan Hsi Chou, Seokjin Go, Won Woo Ro, Hung-Wei Tseng, and Tor M. Aamodt
The 57th International Symposium on Microarchitecture (MICRO 2024)
Recompiling QAOA Circuits on Various Rotational Directions
Enhyeok Jang, Dongho Ha, Seungwoo Choi, Youngmin Kim, Jaewon Kwon, Yongju Lee, Sungwoo Ahn, Hyungseok Kim, Won Woo Ro
The International Conference on Parallel Architectures and Compilation Techniques (PACT 2024)
Effective Interplay between Sparsity and Quantization: From Theory to Practice
Simla Burcu Harma, Ayan Chakraborty, Elizaveta Kostenok, Danila Mishin, Dongho Ha, Babak Falsafi, Martin Jaggi, Ming Liu, Yunho Oh, Suvinay Subramanian, Amir Yazdanbakhsh
arXiv
MAD MAcce: Supporting Multiply-Add Operations for Democratizing Matrix-Multiplication Accelerators
Seunghwan Sung, Sujin Hur, Dongho Ha, Sungwoo Kim, Yunho Oh, and Won Woo Ro
The 56th International Symposium on Microarchitecture (MICRO 2023)
TensorCV: Accelerating Inference-Adjacent Computation Using Tensor Processors
Dongho Ha, Won Woo Ro, and Hung-Wei Tseng
The 2023 International Symposium on Low Power Electronics and Design (ISLPED’23)
R2D2: Removing ReDunDancy Utilizing Linearity of Address Generation in GPUs
Dongho Ha, Yunho Oh, and Won Woo Ro
The 50th International Symposium on Computer Architecture (ISCA 2023)
Investigation on NVIDIA Ampere GPU Architecture With Reverse Engineering
Sujin Hur, Seunghwan Sung, Dongho Ha, Sungwoo Kim, and Won Woo Ro
The 22th International Conference on Electronics, Information, and Communication (ICEIC 2023)
Detecting Pattern of Warp Register Value Differences in CTA using GPU Compiler
Dongho Ha and Won Woo Ro
The 19th International Conference on Electronics, Information, and Communication (ICEIC 2020)
Hardware Accelerator Systems for Artificial Intelligence and Machine Learning
Won Jeon, Gun Ko, Jiwon Lee, Hyunwuk Lee, Dongho Ha, and Won Woo Ro
Advances in Computers, Elsvier, vol. 122: Academic Press; 2020, Chapter 6
Trends of High-End Graphic Processing Unit Development
Dongho Ha, Hyunwuk Lee, Jiwon Lee, Hyun Jae Oh, Won Jeon, Yunho Oh, Won Woo Ro
Korean Information Science Society (2019)
----- On-Going Projects (only project names) -----
DMA between GPUs
On-Going Paper (Co-author)
Effective LLM Inferencing
On-Going Paper (Co-author)
Sparse/Dense NPUs
On-Going Paper (Co-author)
The next step for MAD Macce
On-Going Paper (Co-author)
General Quantum Program Optimization
On-Going Paper (Co-author)
Bit-wise Matrix Multiplication
On-Going Paper (Co-author)
Continuous Learning
On-Going Paper (Co-author, collaborating with Euiseong Seo, SKKU)
Low-Precision Training on GPUs
On-Going Paper (First author, collaborating with Babak Falsafi, EPFL)

Projects

Industry Project
- Study on Memory Sub-System Architecture for Hyper-Scale AI Training, SK Hynix, 2024
- Development of PIM Software Architecture based on Data-Flow Computing, Korea Government (Institute of Information & communications Technology Planning & Evaluation, 2024
- Analysis and Development of GPU Architecture for HPC Workloads [Patent], Samsung (SAIT), 2021-2022
- Development of Data Center Many-core NPU Architecture and Memory Interface, Samsung, 2019-2020
- Development of CPU-GPU Heterogeneous Computing Simulation Environment, SK Hynix, 2019-2020
- Development of the Identification Data Processing Technology for On-site Police Officers, Korea Government (Korea National Police Agency), 2018-2023
- Development of Multi-GPU Based High Speed Ray-Tracing Engine, Samsung, 2017-2018
Paper Peer Review
- IEEE Computer Architecture Letters (CAL) ×5
- IEEE Transactions on Emerging Topics in Computing (TETC) ×1
- ACM Transactions on Architecture and Code Optimization (TACO) ×2
Visiting Scholar
- University of California, Riverside, USA (Advisor: Hung-Wei Tseng), 2022 - 2023
Teaching Assistant
- EEE3530: Computer Architecture, Yonsei University, Seoul, Korea, 2021 Spring
- EEE4473: Embedded System Lab., Yonsei University, Seoul, Korea, 2020 Spring

blogs