Hey,
my understanding goes as follows (but no guarantees):
- the middle red line (4th from top/bottom) denotes the case for RTT = 100 ms
- we see that for certain y-values/throughputs, the CUBIC value (thick turquoise line) is identical to said red line (RENO); this is where the max() operator takes the RENO value
- but from a certain point onward (they say ~50Mb/s in the solutions, but to me it looks more like 10Mb/s), the CUBIC line starts to deviate from this red line; this is where the max() operator starts to select its first argument.
This justifies it for the 100 ms case. What about the 300 ms case? Well for cases with higher RTT, i.e., the lower red lines, the critical throughput point is even lower, thus we can take the 50Mb/s (10Mb/s) as a minimum value for both of them to be "sure".
my understanding goes as follows (but no guarantees):
- the middle red line (4th from top/bottom) denotes the case for RTT = 100 ms
- we see that for certain y-values/throughputs, the CUBIC value (thick turquoise line) is identical to said red line (RENO); this is where the max() operator takes the RENO value
- but from a certain point onward (they say ~50Mb/s in the solutions, but to me it looks more like 10Mb/s), the CUBIC line starts to deviate from this red line; this is where the max() operator starts to select its first argument.
This justifies it for the 100 ms case. What about the 300 ms case? Well for cases with higher RTT, i.e., the lower red lines, the critical throughput point is even lower, thus we can take the 50Mb/s (10Mb/s) as a minimum value for both of them to be "sure".
