Potuz
This was widely discussed in ACD at the time and thoroughly agreed upon. The only reason this wasn't merged then is that we couldn't make public some aspects of pulled...
> It modifies the rules on how to vote for a block. This is what a bribery protocol to e.g. censor certain transactions permanently would do, too. Edit: ah I...
> having things like this behind flags dubious at best and should be considered a temporary solution to address urgent problems and not a permanent state. Just to clarify this,...
Closing in favor of #11981
> The interesting bit that we can potentially reuse is the `hash_64b_blocks` function defined here: > https://github.com/potuz/mammon/blob/main/ssz/sha256.asm That assembly is only for SHA-ni capable CPUs, if there's interest in a...
FWIW, I posted SSE, AVX, AVX2 and SHA modifications to Intel's assembly. If you raise an issue with your SHA provider you may want to link to my repo if...
> > more from @potuz : https://hackmd.io/@potuz/BJyrx9DOF > > For hardcoded padding, is it something like precomputing sha256(0x0000...0000) like done here for hash to curve? > > https://github.com/supranational/blst/blob/0fdf02e9d220e970d574163bfaf8591c38923ca2/src/hash_to_field.c#L22-L35 nope, the...
We'll be using https://github.com/prysmaticlabs/hashtree at prysm, it's still too early to adopt that library, but I'd be very happy to see some benchmarks from Nimbus if you test it.
Here's another weird case: 1 Block A, SYNCING, it's head 2. Block A
I would ask some more technical questions > The first step would be to design a delayed sync mechanism that allows the node to only sync up to a certain...