Nugget: Portable Program Snippets
Qiu, Z., Samani, M., & Lowe-Power, J., "Nugget: Portable Program Snippets." 2026 IEEE International Symposium on High-Performance Computer Architecture (HPCA). Sydney, Australia.
CV
📄 Download PDF CV
Education
- Master of Science in Computer Science, University of California, Davis, Fall 2023 – Spring 2026 (expected)
- Visiting Student in Electrical and Computer Engineering, Cornell University, Fall 2025 – Spring 2026
- Bachelor of Science in Computer Science and Engineering, University of California, Davis, Fall 2020 – Spring 2023
Work experience
- June 2023 – Present: Graduate Researcher
- DArchR Lab, UC Davis
- Advisor: Professor Jason Lowe-Power
- Nugget (LLVM IR sampling)
- Designed and implemented a cross-platform, architecture-independent sampling framework at the LLVM IR level; supports rapid interval analysis, native execution for validation, and cross-ISA simulation.
- Evaluated Nugget on real hardware platforms with diverse performance characteristics (e.g., Ampere Altra) and in the gem5 simulator to validate fidelity and portability.
- Analyzed a diverse set of workloads (SPEC CPU2017, NPB, LSMS) to demonstrate robustness across CPU, HPC, and scientific applications.
- Accelerating the Simulation of Parallel Workloads using Loop-Bounded Checkpoints
- Co-led methodology and implementation to validate LoopPoint samples on gem5; under journal review.
- gem5 contributions
- Ongoing upstream contributions since 2022 to full-system sampling support and related simulation features in gem5; author of 50+ commits.
- August 2025 – Present: Visiting Student
- Computer Systems Laboratory, Cornell University
- Advisor: Professor Christopher Batten
- Agile Robotic Hardware-Software Co-Design
- Developing a closed-loop, multi-robot evaluation framework integrating robotics simulators and architecture simulators; enables parallel per-robot simulation and decouples from a specific ISA/simulator to broaden studies (e.g., ISA extensions).
- Accurate STM32G4 Board in gem5
- Building an accurate STM32G4 MCU model in gem5 (Arm M-profile). Implemented the ART (Adaptive Real-Time) accelerator to improve Flash access latency; planning to upstream core M-profile components (e.g., NVIC).
- June 2022 – June 2023: Undergraduate Researcher
- DArchR Lab, UC Davis
- Advisor: Professor Jason Lowe-Power
- SimPoint
- Implemented SimPoint support in the gem5 stdlib and ran SPEC CPU2006 experiments using SimPoint.
- LoopPoint
- Implemented LoopPoint support in system-emulation and full-system modes in gem5; added a generalized PC-execution counter to track occurrences of specific PCs.
Publications
Talks
[CGO 2026] Nugget with LLVM Infrastructure - LLVM-CGO-2026 Technique Talk
Workshop at LLVM-CGO-2026 Workshop, Sydney, Australia
[HPCA 2026] Nugget: Portable Program Snippets - Accepted Paper Talk
Main Conference at HPCA 2026, Sydney, Australia
gem5 Bootcamp
Bootcamp at gem5 Bootcamp, UC Davis, California
[ISCA 2023] gem5 Workshop
Workshop at ISCA 2023, Orlando, FL, USA
[HPCA 2023] LoopPoint Tools: Sampled Simulation of Complex Multi-threaded Workloads using Sniper and gem5
Tutorial at HPCA 2023, Montreal, Canada
Teaching
Service and leadership
- Founder, Computer Systems Seminar at UC Davis
- Organized weekly speaker series with 12+ talks from academia and industry (ongoing).
- HPCA 2026 Artifact Evaluation Reviewer
- Helped reviewing the artifact of the accepted paper.
Course Projects
- CXL Shared Memory Filesystem Optimization
- Investigated application needs in a CXL shared-memory system and improved the FAMFS framework with more efficient allocation/deallocation and zero-copy operations to avoid redundant copies.
- Limitations of Disaggregated Memory and Innovations
- Analyzed limitations of disaggregated memory and proposed a hardware-software co-designed page management approach.
- CUDA Microbenchmarks
- Developed configurable microbenchmarks to measure shared-memory latency and memory-scaling behavior on GPUs.
- RISC-V Operating System
- Built a simple RISC-V OS on the UC Davis RISC-V Console Simulator for a course project.
Skills
- Programming Languages: C/C++, Python, Bash, CUDA, Assembly, LaTeX
- Hardware Description Languages: Chisel
- System Evaluation Tools: gem5, QEMU
- Profiling & Instrumentation: LLVM passes, DynamoRIO, Valgrind, Linux perf, PAPI
- Linux Tools: cpuset, CRIU, cgroups, Docker
- Compilers: GCC/G++, GFortran, Clang/LLVM
- Languages: English (native), Cantonese (native), Mandarin (professional)