On many PoS coins the processor load rises on small transaction sizes. If this is not the case with blackcoin, it's ok then. If the load is an issue, then the fragment size could be ten times larger, between 5000 and 10000 BLK.
[/quote]

So, it is true that the processor load rises on small transactions. But as far as I know, the staking process is single threaded. But it is a very good question, and if you want you can try different stakecombine or stakesplit configurations. There is the possibility, but it is currently hardcoded in the code.
You would need to alter the code and compile it yourself (or send an PR on Gitlab and we can compile it for you)

the code is here
https://gitlab.com/blackcoin/blackcoin-more/-/blob/master/src/wallet/wallet.cpp#L50

and from here on
https://gitlab.com/blackcoin/blackcoin-more/-/blob/master/src/wallet/wallet.cpp#L801

We are currently trying to combine inputs, so it will always contain an x amount of dust txs. (lets say 100 inputs <= 0.004)

Thanks!
Michel
 
[/quote]

This combine could help, but it can only be useful, if the fees/costs for it are less than the fees/costs for the attacker.

I testet to combine by hand, but the fees for it are much higher than the input.
[/quote]

A Coinstake transaction doesn't have fees, but it only happens when you win the stake

example:
https://chainz.cryptoid.info/blk/tx.dws?8753783.htm