About

About

Currently, I am a System Software Engineer at ByteDance Infrastructure, focusing on Java workload optimization and toubleshooting.

Prior to joining ByteDance, I spent two years at Alibaba Infrastructure Service. I hold a Computer Architecture Master degree from Shandong University, where I was advised by Lei Ju.

Project

ByteDance Infrastructure | Compiler and Library Group (2020.7~Now)

• Maintained OpenJDK Long Term Support Version at Bytedance
  • Self build OpenJDK binary at ByteDance, Provide base docker images, jdk deployment at Datacenter.
  • Optimized BigData workload and build observability tools with performance engineering team.
  • Be good at troubleShooting Memory Leak and Coredump.
• DataCenter Workload Optimization
  • Transparent Hugepages Apply in CodeCache and LibJVM.
  • Optimized Compress library and Parse library in BigData Business.

Alibaba Infra Servive | Hybrid Computing Compiler Group (2018.7 ~ 2020.7)

• Wasm AOT Compiler Framework for Ant Blockchain Platform (2019.9-2020.6)

  • Convert wasm bytecode to LLVM IR and Add T-Head CSKY Extension to the RISCV ISA
  • 一种程序编译方法、设备以及计算机可读介质 2021-09-07 已申请
  • 资源计算方法、装置、电子设备及可读存储介质 2025-02-28 已授权

• Coordinate software and hardware Optimization for CNN & DNN Accelerator implemented in FPGA (2018.7-2019.10)

  • Implement CNN Operation like Conv/DeConv and DNN Operation like matmul/relu
  • Proposed a spliting HD pictures algorithm for Limited on-chip memory
  • Graph Optimization like Concat/Slice/Fusion

ShanDong Univ. Embedded and System Labs

• Energy Efficient Object Detection for Edge Computing
  • As project lead won 3rd place in DAC 2018 System Design Contest with 荣登山东大学融媒体中心新闻
  • Implemented half-precision calculation on GPU
  • Network Pruning and Reduced the down sampling rate

Publication

• NVM in GPGPU Memory Hierarchy

  • Shared Last-level Cache Management for GPGPUs with Hybrid Main Memory (Design, Automation and Test in Europe Conference and Exhibition (DATE) 2017 Best Paper Award Nominations) Guan Wang, Xiaojun Cai, Lei Ju, Chuanqi Zang, Mengying Zhao and Zhiping Jia.
  • Shared Last-Level Cache Management and Memory Scheduling for GPGPUs with Hybrid Main Memory (ACM Trans. Embedd. Comput. Syst. (Volume 17 Issue 4, August 2018)) Guan Wang,Chuanqi Zang, Lei Ju, Mengying Zhao, Xiaojun Cai, and Zhiping Jia

• Cache Coherence Research in GPGPU

  • 基于编译器辅助的GPGPU缓存一致性研究 (Master Thesis 2018)
  • Proposed a static program analysis which enable GPU kernels to conservatively load global data in the private L1 cache which are guaranteed to have no coherence issue.