Yiwei Yang (VickieGPT)


I'm a second-year Computer Science System Ph.D. student @BaskinEng. I'm honored to be ed by Andrew Quinn, and hacking on cutting-edge HW/SW co-design. The goal of my research is to accelerate data access and manipulation on modern heterogeneous architecture and general-purpose processors for diverse workloads across companies utilizing CXL technology. I believe in a hardware-software co-design methodology. The hardware provides specialized acceleration capabilities for common cases without sacrificing heterogeneous processing performance with a better knowledge of workloads, while the software adaptively takes advantage of the hardware by sending semantic hints, and falls back to alternative, slower paths whenever hardware acceleration is not achievable. Assured with my great passion for seeking new knowledge and commercial opportunity and situated in two metropolitans, the San Fransisco Bay area, and Shanghai, I'm proud to learn more, benefit others more, and earn more. Previously, I was an undergraduate @ShanghaiTech.
fn research_no_failure()->Result < Paper, Error > {
    let output = if cfg!(target_os = "linux") {
        .args(["-smp 6", "-numa node,cpus=0-7,memdev=mem0,nodeid=0", "-object memory-backend-ram,id=mem0,size=8G","-numa node,cpus=8-15,memdev=mem1,nodeid=1", "-object memory-backend-ram,id=mem1,size=8G","-m 16G,slots=4,maxmem=32G", "-machine q35,cxl=on", "-M cxl-fmw.0.targets.0=cxl.1,cxl-fmw.0.size=4G",
             "-device pxb-cxl,bus_nr=12,bus=pcie.0,id=cxl3.1", "-device cxl-rp,port=0,bus=cxl.1,id=root_port13,chassis=0,slot=2",
             "-device cxl-type3,bus=root_port13,memdev=cxl-mem1,lsa=cxl-lsa1,id=cxl-mem0",
             "-device cxl-type1,bus=root_port14,virtio-crypto-cxl,id=crypto0,cryptodev=cryptodev0"])
            .expect("failed to execute process")
    let paper = Paper::new(output);
        asm!("clflush" :: "r" (&paper.iter()) : "rax", "rbx", "rcx", "rdx": "volatile" );
        if (paper.is_valid()){


Current Research

Research Assistant
August 2022 - Present
  • Understanding the performance characteristics of CXL.mem systems. Data-driven far memory allocation, prefetching, and replacement policies. Emulating memory disaggregation for data center applications
  • Make Hardware Software Co-design for on CXL.cache data movement
  • Make Virtual Machine migration based on WebAssembly
  • Make cross kernel-userspace eBPF observation for distributed system

Advising Experience


Kgent: Kernel Extensions Large Language Model Agent                                                

bpftime: userspace eBPF Runtime for Uprobe, Syscall and Kernel-User Interactions

CXLMemSim: A pure software simulated CXL.mem for performance characterization

Critique of "A Parallel Framework for Constraint-Based Bayesian Network Learning via Markov Blanket Discovery" by SCC Team from ShanghaiTech University

Critique of “MemXCT: memory-centric X-ray CT reconstruction with massive parallelization” by SCC Team from ShanghaiTech University

Attack as Defense: Characterizing Adversarial Examples using Robustness

Teaching Assistant

ShanghaiTech CS131: Compiler

UCSC CSE12: Computer Systems and Assembly Language