The main factor is depend upon answers to these questions.
How much throughput of one bp was on the testnet of bitmex?
Because if one bp throughput is low that cause decrement in overall network CPU and users will have less available CPU.
For more information :- https://support.dexeos.io/hc/en-us/articles/360009054093-EOSIO-Why-CPU-bandwidth-varies-
How many bp were there for load balancing?
If thousand txs validated by one bp then bp will suffer all stress by validating all transaction. if there are as many as bp then stress can be distrubuted among these block producer by parallel transaction execution.
Were there any light node to validate only block header to reduce the stress for bps?
In "light" mode all incoming blocks headers will be fully validated; transactions in those validated blocks will be trusted.
From my point of view, if EOS mainnet can acheive some throughput that doesn't mean all EOS sister chain will produce the same throughput.