Hey, Nathan - good to see you again! 
Por que no los dos?
Higher density is always good if affordable/feasible - and then herding them with Hashtopolis can still totally happen on top of that. I'd personally go for the 8-GPU option, since it also allows you to carry out non-HtP jobs easily.
Either way, it's still definitely worth getting Hashtopolis into the mix. Some attacks are more easily "hashtopolized" than others (most attacks are possible, but it gets trickier). If it's mostly a fast hash like NTLM, then even some simple workload assignment - still using Hashtopolis, just throwing some nodes at a stack of mask attacks, while doing wordlist+rules on other nodes, etc. - can compensate for that. But even just being able to pause and resume jobs, and move systems around the workloads, etc. is definitely worth the setup effort.
And yeah, hashtopolis is quite stable - and performant
https://twitter.com/TychoTithonus/status...7208970240
Por que no los dos?
Higher density is always good if affordable/feasible - and then herding them with Hashtopolis can still totally happen on top of that. I'd personally go for the 8-GPU option, since it also allows you to carry out non-HtP jobs easily.Either way, it's still definitely worth getting Hashtopolis into the mix. Some attacks are more easily "hashtopolized" than others (most attacks are possible, but it gets trickier). If it's mostly a fast hash like NTLM, then even some simple workload assignment - still using Hashtopolis, just throwing some nodes at a stack of mask attacks, while doing wordlist+rules on other nodes, etc. - can compensate for that. But even just being able to pause and resume jobs, and move systems around the workloads, etc. is definitely worth the setup effort.
And yeah, hashtopolis is quite stable - and performant
https://twitter.com/TychoTithonus/status...7208970240
~