-
binaryFate
<tevador> I think the hashrate there is calculated from the difficulty, so it's the running average over the last 720 blocks <-- the most extremes quantiles are not part of computation though, no? So if a block is particularly fast or slow, the difficulty (and hashrate) will appear not to change as the block is "ignored"
-
binaryFate
selsta ^ the computed hashrate may not always appear to change from one block to another
-
selsta
binaryFate: yep but let’s say suddenly half the hashrate drops, would it take ~720 blocks for it to fully update?
-
sech1
675 blocks
-
sech1
75 blocks before difficulty starts to react and then 600 more to fully adjust (out of 720, top and bottom 60 are discarded)
-
sech1
75 is because there's also 15 block lag before it takes 720 blocks to calculate the difficulty
-
selsta
ok thanks I understand it now
-
tromp
tevador: where on page 88 of
github.com/zcash/zips/blob/master/protocol/protocol.pdf do you see distinct index requirement?
-
tevador
-
tromp
that's describing the gernealized birthday problem
-
tromp
the equihash conditions should all be specified following "A valid Equihash solution is t"
-
tromp
i think for a spec this is ambiguous at best
-
tromp
did you check what zcash verification code does?
-
tevador
-
hyc
so zcash has abandoned ProgPoW?
-
hyc
and they want to move off equihash still?
-
tevador
hyc: zcash has been using equihash since the beginning, it's ethereum which considered progpow
-
hyc
yes...
-
hyc
but zcash had been talking about moving off equihash for a while now
-
hyc
although then it seemed they gave up on the idea of changing, and were just going to live with ASICs
-
hyc
and ProgPoW was one of the aglos they were looking at switching to
-
tevador
AFAIK they considered changing their PoW in 2018 when the first ASICs appeared, but they decided not to do anything
-
tromp
you're right; they do enforce distinct indices. they actually do 2^17-1 comparisons to check.
-
tromp
if allowing identical indices in your equix only produces a tiny fraction of additional solution, then i see no point in forbidding them
-
tevador
tromp: 128k comparisons? I'd like to see their solution verification time...
-
tromp
correction; that's (512 choose 2) = 2^17-2^8 comparisons. still close to 128K.
-
tromp
ZCash PoW solution also dominates header size at 1344 bytes
-
tromp
how big is Monero header?
-
sech1
Empty block is around 115 bytes
-
tevador
I'm not sure why exactly they chose Equihash 200,9 - there are much better options with faster verification and smaller proofs
-
tromp
because they rushed the launch and thought other variants might take too many seconds per attempt
-
tromp
also they (actually, the Equihash authors) badly misjudged the space required to solve 200,9
-
tromp
by the time they held the miner implementation challenge, they were alrd pretty much committed to 200,9
-
tromp
the entries in that contest showed that it required far less memory than the 1GB they imagined
-
moneromooo
Empty block is more than < 100 bytes IIRC.
-
moneromooo
I see a recent one at 96 bytes (2112514).
-
sech1
Yes, it's 96 bytes or more, depending on extra nonce size.
-
UkoeHB_
that's the miner tx size; maybe 'block size' isn't taking into account the header properly
-
UkoeHB_
the header should be at least 42 bytes I think
-
moneromooo
Oh. Good point...