Liou Jhe-Yu (劉哲宇)

Hi, my name is Liou, Jhe-Yu. You can call me either LJ or Elvis. My research interesting are 3D computer graphics and GPU architecture. Besides, I also create and maintain this CASLAB wiki site on a Synology NAS for our lab use.

• PhD student, School of Computing, Informatics, and Decision Systems Engineering. Arizona State University (2015.8 ~ present)
• M.S., Institute of Computer and Communication Engineering. National Cheng Kung University(NCKU) (2007.8 ~ 2009.8)

• B.S., Department of Electrical Engineering, NCKU (2003.8 ~ 2007.7)

• Kaohsiung Municipal Kaohsiung Senior High School

• R.A., Computer Architecture and System Laboratroy, Institute of CCE, NCKU. (since 2012.8)

• Administrative Assistant, Computer Architecture and System Laboratroy, Institute of CCE, NCKU. (since 2012.8)
  • If you are also the AA (or in near future) and want to know the whole working flow, please see here.

• Digital IC designer., ALi Corporation (2010.10 ~ 2012.7)
  • ALi Corporation is mainly focusing on set-top box developing. during working in this company, I was responsible to integrade ARM Mali GPU into our own SoC. I was also introduing about 3D graphic and GPU rendering algorithm to my team and helping them to know the connection between 3D application and Set-top Box.
  • Basiclly, integrading a GPU IP into SoC is not requiring much knowledge about 3D graphic or GPU architecture but RTL coding(plus FPGA verification, Code coverage, even with some SoC backend procedure, such as placment&routing), and has less challenge for me. Therefore I quit after this integration job was done and went back to university seeking a chance to study abroad.

• Second Lieutenant in Army, Republic of China(Taiwan). (2009.10 ~ 2010.9)
  • Compulsory military service
  • Under this period, I took the leisure time to read Real-Time Rendering.

• Jhe-Yu Liou and Chung-Ho Chen,
  Re-visit Blocking Texture Cache Design for Modern GPU,
  in the 11th International SoC Design Conference (ISOCC), Nov. 3-6, 2014, Jeju, Korea.
  • This paper is also generated by my OpenGLES 2.0 GPU simulator.

• Session speech, From SIMD to SIMT,
  Workshop on Heterogeneous System Architectures and Platforms(Chinese), Aug 1-3, 2014, Tainan, Taiwan.

• Kai-Han Tseng, Sheng-Chi You, Jhe-Yu Liou, and Tsung-Yi Ho,
  A top-down synthesis methodology for flow-based microfluidic biochips considering valve-switching minimization.
  In Proceedings of the 2013 ACM international symposium on International symposium on physical design (ISPD '13). Mar. 26, 2013

• Class speech, 3D Computer Graphic GPU Architecture Intro,
  invited from National Sun-Yat-Sen University, Dec. 10 and 17, 2012, Kaohsiung, Taiwan.

• Master thesis, Design, Analysis, and Implementation of a Rasterization Engine based on Tile-Based Rendering Architecture in 3D Graphics, Jul. 2009

• Support HSA for the Unified Shader GPU architecture, Industrial Technology Research Institute(ITRI) (since 2014.1)
  • This continuous project trys to upgrade the shader core by manipulating HSA intermediate language (HSAIL) instead of Nvidia graphic programming language series 4 (NVGP4) from last project. My responsibilities are creating the Binary Instruction format and managing the SIMT architecture issue, such as branch divengence, register and program counter sharing.

Invalid Link

POM example rendered by my gpu simulator

• Development of Unified Shader Architecture GPU, ITRI (2013.1 ~ 2013.12)
  • This project contained a programmable GPU based on OpenGL ES 2.0. The most stuff, including benchmarks, are investigated or created by myself with few are rewritten from the previous project.
  • You can find the introduction in my OpenGLES 2.0 GPU simulator and the source code in my git repository. There is also a project gallery inside the repository.

• Design and Implementation of a Multiview supported GPU, Himax Technology (2007.9 ~ 2009.4)
  • During studying my master degree, Shekenny, Cashman, and I completed this project together. Because our advisor wasn't familiar with 3D graphic at all, we worked hard by our own to survey a bunch of material. Although the final performance result may be not strong as we expected, it was still a great journey for me which I really satisfied with.
  • This project contained both a C model simulator and a synthesizable RTL model, and all based on OpenGL ES 1.1. We let the RTL model worked under QEMU and Coware so that our OpenGL runtime could simulate with it concurrently.
  • This is where I started learning and understanding 3D graphic. The part of this project is also my master thesis.

I was a member of the badminton school team when I studied in NCKU. It means I was participated in about 6 years. And, sadly to say, I didn't get any competitions record which I always tried hard to, but only an offcial paper as a school team service proof.

Besides, I was also the badminton club leader when I worked in ALi, and was responsible to teach my colleagues how to play badminton.

The motivation to research GPU is from my enthusiasm of 3D video game. My first 3D game is Tomb Raider III which supported both software-render(by CPU) and hardware render (by GPU). Because of this game, I bought my first piece of graphic card, Voodoo 2 by 3Dfx Interactive, and enjoy the significant 3D effect compared to CPU rendered. After that, I realized the GPU is the essential componet to 3D game and tried to get the top GPU cards each time I bought my new computer.

Although the the most 3D games presented as FPS the TPS like DOOM 3 or Crysis have the stunning 3D effect, I also love the simulation game or RPG with good story such as Planescape:Torment. This game by far had the most impressive story about philosophy of live and dead which had no other story could even compared with in my experience. The most famous motto “Time is not your enemy. Forever is” is still haunted in my brain. I strongly recommend this game.