CaLRS: A Critical-Aware Shared LLC Request Scheduling Algorithm on GPGPU

Figure 6

Priority distribution of GPU memory requests.