The clock frequency of the circuit is not an indicator of its performance. I can use the generic ALU clock generator common to the group (it's easier for me to implement it as a developer) to reduce the performance of individual nodes by simply sending a PAUSE signal to them, which will cause them to skip them, but the overall clock frequency of the circuit will not change and output as an indicator to the operator console will not give him an idea of the real processes in the equipment, and the indication of loading specific ALU will allow him to make the right decision.
The same with the implementation of technologies like Turbo Burst - we feed the signal of the same frequency on the CLOCK line of the circuit, but if necessary, program the synchronization node of a specific ALU so that N cycles from M will be missed, which will reduce its effective clock frequency and circuit heating .
And besides, when designing a scheme, a developer can use much faster circuitry elements that will allow performing more operations per clock, and this has long been used in computers. At least from the appearance of the first transputer accelerators in the early 60's.
Therefore, the values of frequencies are an indicator only in the advertising booklet and it makes no sense to derive it as an indicator.