Hey @alfredino,
let me address your questions!
In the case of IPC, is n comparable to the number of validators in a BFT PoS network? What are the differences?
Yes it’s the same calculation based on supermajority agreement (more than 2/3 of participants). The main difference is in security models. At PoS every validator has a stake at risk. In IC the stake is with the DAO which is assigning the node provider role to identifiable parties for rewards paid in ICP. Node providers are not anonymous, are expected to host nodes in secure certified data centers and can be made accountable for the gross negligence wrt. security.
PC claims to have a block time of less than one second and a finite time slightly longer than one second L1 comparison - Internet Computer Wiki. How many blocks are needed to have a transaction confirmed? It seems that it depends on the number of replica n.
Once a transaction was included in a block, in a happy case scenario, the finality is expected to happen within one consensus round (which is for most subnets ~1s). However, due to multiple reasons like connectivity issues or faulty nodes, the finality time can take longer. Also the number of replicas in as subnet can have an impact on that as the amount of communication overhead grows with the number of participants. Also, see this breakdown of the finality time.
- IPC performance in terms of TPS seems really good. Is there any public testing similar to the TON blockchain that recently broke the world record for TPS ?
We indeed did some tests a year ago. Check out this link.
Let me know if you have more questions!