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

Figure 9

The average queue latency changement for different classes of memory requests that have the same priority.