[P0] `devshards`: Limit amount of inferences #977
enhancement devshards
As described in https://github.com/gonka-ai/gonka/issues/914#issuecomment-4090483233, we want to limit the amount of inferences that can be done in a devshard, as a way of limiting the amount of damage a hijacked devshard can cause on the network.
For now, let's say each devshard can run up to 2k inferences.
Upon settlement, the protocol should verify "Missed inferences + Invalidated inferences" does not exceed 2k, for the whole group.
💬 Comments (1)
🔄 Auto-synced from Issue #977 every hour.
At 0.2.13 nonce limit was introduced. Inference count could be limited by nonces limit and there is no need in extra parameter. Handling on devshard side is implemented at https://github.com/gonka-ai/gonka/pull/1258