Education

Australian National University (2021-2025 (expected), Canberra, ACT, Australia)

Doctor of Philosophy in Computer Science
Designing a distributed near-memory accelerator inside a RISC-V SoC for garbage collection.
Developed state-of-the art performance analysis and optimization tooling and methodology.
Supervision of undergraduate students.
Received Australian Government Research Training Program (AGRTP) International Fee Offset and Stipend Scholarship
Advisers: Steve Blackburn (Google DeepMind; ANU), Mike Bond (OSU), Martin Maas (Google DeepMind)

Australian National University (2017-2020, Canberra, ACT, Australia)

Bachelor of Philosophy-Science (Honours)
Thesis: Concurrent Copying Garbage Collection with Hardware Transactional Memory
First Class Honours, University Medal
Major: Computer Science, Minor: Mathematics
Advisers: Steve Blackburn, Mike Bond (OSU)
Received ANU Honours Scholarship

Research and Industry Experience

Australian National University (2017-, Canberra, ACT, Australia)

Research Assistant, Programming Languages and Systems Lab
I am a core developer of MMTk—a high-performance memory management framework written in Rust, with customers including Shopify (for Ruby) and RelationalAI (for Julia).
I write and optimize low-level performance critical code in C++/Assembly, such as OpenJDK JIT compiler for ARM and RISC-V.
I lead multiple microarchitectural optimization projects, which uses latest CPU features, including PEBS load latency, and Last Branch Record (LBR).
I contribute my changes to upstream open-source projects, including JikesRVM, OpenJDK, and Linux perf.
I previously worked on the Mu Micro Virtual Machine project, including a Mu backend for the RPython JIT compiler.
I manage a cluster of machines for high-fidelity performance measurements of software systems.

Google Research (2022, Pyrmont, NSW, Australia)

Student Researcher, Brain Systems and Programming Languages Research Team
Mentored by Qian Ge, and Martin Maas. I also work with Hans Boehm, and Lokesh Gidra from the Android Team.
I developed performance analysis tools for Android in Perfetto and eBPF. I worked towards quantifying the impact of garbage collection on the energy and memory consumption, and responsiveness of apps, including Google Maps, and YouTube, on Pixel devices.

Twitter (2020-2021, Canberra, ACT, Australia)

Reliability Engineer Intern, Infrastructure Optimization and Performance (IOP) team
Mentored by Xi Yang and Yao Yue
I analyzed and optimized the OpenJDK compiler and GC configurations across the fleet to improve the latency and reduce the resource usage in core services, including ads prediction and ads selection.

Microsoft Research (2020, Canberra, ACT, Australia)

Contractor, Parasail Team, Research in Software Engineering (RiSE) group
Supervised by Todd Mytkowicz
Continuing my internship work

Microsoft Research (2019-2020, Redmond, WA, USA)

Research Intern, Parasail Team, Research in Software Engineering (RiSE) group
Mentored by Todd Mytkowicz
I built SCCL (Synthesized Collective Communication Library) to exploit GPU topologies in the datacenter and synthesizing optimal implementations of MPI-style collective operations in C++ and CUDA for distributed ML training. SCCL led to up to 2.2× speedup over GPU vendor libraries, and won the PPoPP’21 Best Paper Award.

Publications

Peer-Reviewed Conference Papers

Rethinking Java Performance Analysis. Stephen M. Blackburn, Zixian Cai, Rui Chen, Xi Yang, John Zhang, John Zigman. Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2025. pdf doi artifact

Improving Garbage Collection Observability with Performance Tracing. Claire Huang, Stephen M. Blackburn, Zixian Cai. International Conference on Managed Programming Languages & Runtimes (MPLR), 2023. pdf doi code

Distilling the Real Cost of Production Garbage Collectors. Zixian Cai, Stephen M. Blackburn, Michael D. Bond, Martin Maas. International Symposium on Performance Analysis of Systems and Software (ISPASS), 2022. pdf doi code video

Understanding and Utilizing Hardware Transactional Memory Capacity. Zixian Cai, Stephen M. Blackburn, Michael D. Bond. International Symposium on Memory Management (ISMM), 2021. pdf doi code video

Synthesizing Optimal Collective Algorithms. Zixian Cai, Zhengyang Liu, Saeed Maleki, Madan Musuvathi, Todd Mytkowicz, Jacob Nelson, Olli Saarikivi. Principles and Practice of Parallel Programming (PPoPP), 2021. Best Paper Award. pdf doi code (SCCL) code (MSCCL) video

Technical Reports

Activity Recognition in Videos with Segmented Streams. Zixian Cai. ANU SCNC2103, 2019. Supervised by Stephen Gould. pdf doi

Verification of Concurrent Data Structures with TLA. Zixian Cai. ANU SCNC2102, 2018. Supervised by Michael Norrish. pdf doi

Power Consumption of Instruction Encodings on Cortex-M4. Zixian Cai. ANU COMP2300 ASE, 2018. Supervised by Ben Swift. pdf doi

Have Your Cake and Eat It? Productive Parallel Programming via Chapel’s High-level Constructs. Zixian Cai. ANU SCNC2101, 2018. Supervised by Josh Milthorpe. pdf doi

Theses

Concurrent Copying Garbage Collection with Hardware Transactional Memory. Zixian Cai. Bachelor of Philosophy (Honours) - Science, Australian National University, 2020. Advised by Steve Blackburn and Mike Bond (OSU). pdf doi

Awards

PPoPP Best Paper Award (2021)

ANU University Medal (2020)

ANU Chancellor’s Letter of Commendation (2017, 2018, 2019, 2020)

ANU Burton and Garran Hall Distinguished Scholar Award (2017, 2018, 2019)

Teaching Experience

ANU COMP3710, Computer Microarchitecture

in 2021 Semester 2
- Teaching Assistant
- Course Convener: Shoaib Akram

ANU COMP3703, Software Security

in 2021 Semester 1
- Teaching Assistant
- Course Convener: Alwen Tiu

ANU COMP3710, Compiler Design and Implementation

in 2020 Semester 1
- Teaching Assistant
- Course Convener: Tony Hosking
- Lecturer: Ben L. Titzer
- Received ANU CECS 2020 Remote Teaching and Student Experience Awards Notable Mention for exceptional adaptation to remote delivery of course material

Students

Claire Huang. High Performance Garbage Collection: A Microarchitectural Perspective. ANU Honours thesis (2024 Semester 1-2025 Semester 1). Co-supervised with Steve Blackburn. pdf

Alex Mirrlees-Black. Pointer Identification Using Run-Time Taint Tracking. ANU undergraduate project course (2024 Semester 1). Co-supervised with Steve Blackburn, and Stephen Kell (King’s College London).

Zheyuan Chen (University of California, Santa Cruz). Adding ARMv8 support to MMTk/OpenJDK. Open Source Promotion Plan (2023 Q3). Mentor.

Claire Huang. Debugging GC Performance Problems with eBPF Tools. ANU undergraduate project course (2022 Semester 2). Primary supervisor, co-supervised with Steve Blackburn.

Service

Conferences

Programming Language Design and Implementation (PLDI), 2024. Organizing Committee, Video Co-Chair.

Programming Language Design and Implementation (PLDI), 2023. Organizing Committee, Video Co-Chair.

Programming Language Design and Implementation (PLDI), 2023. Artifact Evaluation Committee.

Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2023. Artifact Evaluation Committee.

Systems, Programming, Languages, and Applications: Software for Humanity (SPLASH), 2022. Organizing Committee, Video Co-Chair.

Programming Languages Design and Implementation (PLDI), 2022. Artifact Evaluation Committee.

Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2022. Artifact Evaluation Committee.

Systems, Programming, Languages, and Applications: Software for Humanity (SPLASH), 2021. Streaming Operator.

Monitoring the video streaming and providing technical support during mirrored sessions

Systems, Programming, Languages, and Applications: Software for Humanity (SPLASH), 2020. Student Volunteer.

Helping with the streaming of the first edition of virtual SPLASH
See Alex Potanin’s blog post on how we made SPLASH accessible to people in different timezones

University

ANU School of Computing HDR Student Representative, 2024-2025.

ANU PhB Student Mentoring Program, 2019. Mentor.

Talks

Trashy Inflation: How Garbage Collectors Are Getting Worse, and How We Can Fix It. Zixian Cai. Sydney Area Programming Languages INterest Group (SAPLING), 2023. video pdf

Distilling the Real Cost of Production Garbage Collectors. Zixian Cai. International Symposium on Performance Analysis of Systems and Software (ISPASS), 2022. video

Understanding and Utilizing Hardware Transactional Memory Capacity. Zixian Cai. International Symposium on Memory Management (ISMM), 2021. video

Activity Recognition in Videos with Segmented Streams. Zixian Cai. ANU PhB Symposium, 2019 for SCNC2103.

Portable Memory Management Toolkit in Rust. Zixian Cai, Brenda Wang, Pavel Zakopaylo. Sydney Area Programming Languages INterest Group (SAPLING), 2018.

Have Your Cake and Eat It? Productive Parallel Programming via Chapel’s High-level Constructs. Zixian Cai. ANU PhB Symposium, 2018 for SCNC2101.

Micro Virtual Machines. Zixian Cai. ANU PhB Mini Symposium, 2018.

Mu for Dynamic Languages: Retargeting the RPython JIT to a Micro Virtual Machine. Zixian Cai. Sydney Area Programming Languages INterest Group (SAPLING), 2017. pdf

Table of Contents