NCOR: An FPGA-Friendly Nonblocking Data Cache for Soft Processors with Runahead Execution

Figure 12

The impact of number of outstanding requests on IPC. Speedup is measured over the first configuration with two outstanding requests.