CUDA code performances and new hashing limit on RTX 30 families


When the market will permit it, I would like to upgrade my setup from gtx 1050ti to a new RTX 3060 ti or 3070. However, I’m somehow worried about the hashing limiting logic will be introduced (likely) on all the RTX 30 family in the next months to come. I read it would be “selective” on the Etahash and/or Dagger-Hashimoto…but how much selective?

Has somebody any further information on how it could affect other CUDA accelerated algorithms? I’m not really interested on crypto at all, but still interested to use CUDA feature uncapped.