Post
Topic
Board Development & Technical Discussion
Re: O(1) Block Propagation, IBLT
by
TierNolan
on 12/11/2015, 14:19:31 UTC
There is a limit to how small the table can be.  For it to work, at least one cell must have +/- 1 transaction piece stored in it.

This means that the table size is proportional to the difference between the memory pool at the receiver and the memory pool for the miner who created the block.

If all nodes create the same block but with 1% difference then the table size is proportional to the block size. 

For IBLTs to operate with O(1) efficiency, there needs to be some method to boost efficiency of keeping all miners' memory pools aligned. 

If 1MB blocks tended to have 50kB differences, is it reasonable to assume that 100MB blocks would also be aligned to within 50kB?