General-Purpose GPUs

Evaluating an adaptive LLC caching scheme for general-purpose GPUs

An overview of GP-GPU Architecture

Guide: Prof. Virendra Singh

I worked on this project as part of a Summer Project with Prof. Virendra Singh. I had been tasked with understanding and implementing an Adaptive decoupled last-level cache design for general-purpose GPUs. Since GPUs were a new topic for me at the time, I studied and reviewed the SIMT core, memory systems and the programming model related to GP-GPU architecture. I also reviewed literature about leveraging Adaptive last-level caching and implementing it on the GPGPU-Sim simulator. I also performed various benchamrk simulations on the simulator and carefully analyzed the outputs received.

     Report

References

2019

  1. Adaptive Memory-Side Last-Level GPU Caching
    Xia Zhao, Almutaz Adileh, Zhibin Yu, and 3 more authors
    In Proceedings of the 46th International Symposium on Computer Architecture, 2019