Hands-On Networking Thorsten Herfet, Andreas Schmidt, Frank Waßmuth

News

MMCP Performance

To clarify the passing criteria for the MMCP project, here are the results of our reference implementation. To pass the project, the residual error ratios must be at least as low as the reference implementation's. Runtimes depend on your hardware, so there might be differences, but must be below the upper bounds mentioned on the task sheet. This also holds for the redundancy you add.

Residual / Errors = Ratio     | Measured   / Ideal Redundancy | Runtime | Pattern
       0 /    210 =     0.00% |   1.00e+00 /         4.86e-06 |    151s | Benevolent_Pattern_1.txt
       0 /    190 =     0.00% |   1.00e+00 /         4.40e-06 |    144s | Benevolent_Pattern_2.txt
       0 /    190 =     0.00% |   1.00e+00 /         4.40e-06 |    142s | Benevolent_Pattern_3.txt
     118 /  21634 =     0.55% |   1.00e+00 /         5.01e-04 |    149s | Adversary_Pattern_1.txt
     135 /  21932 =     0.62% |   1.00e+00 /         5.08e-04 |    166s | Adversary_Pattern_2.txt
     116 /  21851 =     0.53% |   1.00e+00 /         5.06e-04 |    139s | Adversary_Pattern_3.txt
     186 /    442 =    42.08% |   1.00e+00 /         1.02e-05 |    142s | Burst_Pattern_1.txt
     252 /    502 =    50.20% |   1.00e+00 /         1.16e-05 |    150s | Burst_Pattern_2.txt
     366 /    654 =    55.96% |   1.00e+00 /         1.52e-05 |    149s | Burst_Pattern_3.txt

Update: We have changed our reference implementation slightly, to make your life a bit easier and account for data-dependency.