Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Bryant Eisenbach
    @fubuloubu
    we should always have a PoW testnet anyways
    Andrea Lanfranchi
    @AndreaLanfranchi
    Agree ... but test network new functionalities. Not to keep sealing algo in a perpetual testing phase. Who would want it ?
    Is ethash in perpetual testing phase ?
    Danno Ferrin
    @shemnon
    testnets are always in testing phase.
    Andrea Lanfranchi
    @AndreaLanfranchi
    Not for PoW sealing algo
    Danno Ferrin
    @shemnon
    The algo wouldn’t change, it’s the implementations it’s smoking out problems for.
    Also, to gauge how “fast” or “slow” each progpow epoch is/was/
    Just because you are monitoring something doesn’t mean it’s in testing phase.
    Andrea Lanfranchi
    @AndreaLanfranchi
    You mean ProgPoW periods right ? Epochs are constant
    Danno Ferrin
    @shemnon
    Right, periods are for progpow, epochs are for the dag. So it’s to gauge the propow periods speed and implemenations stability in translating the instructions into their GPU stack.
    Really it’s a block offset, like 8.5 mil or 9 mil. Epochs and periods will fall out of that.
    Andrea Lanfranchi
    @AndreaLanfranchi
    We have tested 1M blocks on gangnam testnet ... we have nominal variance in hashrate among periods (worse to best) of less than 8% (for 0.9.2) want to re-test it ?
    0.9.3 specs will reduce it further making variance among periods practically irrelevant
    Consider 0.9.3 will have a period of 10 blocks ... at current block-time is less than 3 minutes.
    Andrea Lanfranchi
    @AndreaLanfranchi
    Anyway ... this reasoning is only proving me that all efforts carried out by volunteers (testnet, performance tests, miners, nodes) etc are considered worth nothing cause not carried under the "supervision" of core devs or auditors. This brings to inevitable conclusion : conspirancy theories find fertile soil to bloom. By consequence don't expect to find any soon PP (better say never) on eth main net. Only possible scenario is a an "unfriendly" hf

    Really it’s a block offset, like 8.5 mil or 9 mil.

    In case of compiler issues (for wich software work-arounds may not possible) are not enough. AMD has been involved in the "bogus period" issue since Jan 2019.
    We're on July ... and if not for the #pragma unroll 2 the issue would still be there.

    Andrea Lanfranchi
    @AndreaLanfranchi

    So it’s to gauge the propow periods speed and implemenations stability in translating the instructions into their GPU stack.

    We had already done all of that ... but the "illuminati" community didn't give a shit about that. No surprise me, and all others involved, have withdrawn their work and there is no more gangnam network.

    You'll have to redo everything from scratch and once again you'll enter in the same debates.
    Danno Ferrin
    @shemnon
    Two tidbits
    (1) Least Authority has publicly stated that the ProgPow audit is starting - https://medium.com/least-authority/https-medium-com-least-authority-kicking-off-our-review-of-progpow-be1368ae9a50
    (2) There was a ZCash foundation grant to investigate new PoW in general, resulting in a large focus on ProgPow, findings are here - https://github.com/ZcashFoundation/GrantProposals-2018Q2/issues/25#issuecomment-513876705
    Bryant Eisenbach
    @fubuloubu
    oh nice (re: ZF grant)
    Bryant Eisenbach
    @fubuloubu
    @shemnon I left a comment with your note that the audit completion timeframe means this cannot be considered for Istanbul anymore. I think this should be noted on the next call and discuss it's inclusion into the following fork, or a non-roadmap hard fork (possibly coinciding with the finality gadget)
    (cannot be considered due to the deadline for testing)
    This is probably upsetting to a lot of people, but I don't see it as that big of a deal since a majority of the community that didn't necessarily disagree with it's inclusion would have preferred it to be a separate fork anyways.
    So it's just a matter of defining what exactly that looks like now that we don't have the Istanbul deadline hanging over our heads
    @Souptacular should I add this to the agenda for the call after this friday (which is a special call to follow up from the last one)
    Bryant Eisenbach
    @fubuloubu
    (to summarize a different way, it seemed "number supporting ProgPoW" < "number supporting ProgPoW into Istanbul")
    ((and a small number of those that didn't support might support it alongside some changes from the finality gadget))
    Danno Ferrin
    @shemnon
    Haveing a ProgPow only fork will actually saciate some critics. Martin is also on board with the PP only fork.
    Bryant Eisenbach
    @fubuloubu
    that's what I prefer as well
    less to keep track of on deployment
    Danno Ferrin
    @shemnon
    I’m thinking once we put a bow on what is and is not intanbul I’ll spin up an EIP to cover changes from the original 1057 EIP, such as hash multiplier and period offset for a test net.
    Bryant Eisenbach
    @fubuloubu
    there's a lot of slated EVM changes to Istanbul
    Danno Ferrin
    @shemnon
    What I’m thinking will happen is that once te audit comes back a testnet will be stood up reflecting the recommenced changes in the Audit. We cannot just re-use gangham because that is 0.9.2 and 0.9.3 is the current proposal and there is no mechanism to tweak parameters mid-blockchain (yet). There will be another transition event and once the testnet appears stable with sufficient diversity in mining software a hard fork can be assigned. This is just me spitballing and not a formal commitment.
    and by a hard fork being assigned I mean picking a fork block number.
    Bryant Eisenbach
    @fubuloubu
    could we just make Ropsten that testnet?
    Danno Ferrin
    @shemnon
    Ropsten is too hard to sync and full of a bunch of zombie chains. Getting peers is a nightmare
    Ropsten needs to let the difficult bomb go off just so we can see how it looks.
    Bryant Eisenbach
    @fubuloubu
    yeah, I totally agree on that
    I think at this point though another ProgPoW-specific testnet is overkill though, except for maybe more benchmarking
    Ropsten would be important progress
    Danno Ferrin
    @shemnon
    A new testnet would be intended to replace ropsten in the long run.
    Bryant Eisenbach
    @fubuloubu
    but then we don't get the opportunity to watch the bomb?
    Danno Ferrin
    @shemnon
    nobody is stopping people from mining ropsten. The only thing that can kill a chain is the difficulty bomb.
    Bryant Eisenbach
    @fubuloubu
    unless you suggest we stand up a testnet to replace Ropsten, and then do the bomb to Ropsten, see what happens, and retire it
    Danno Ferrin
    @shemnon
    I need to do some research but IIRC ropsten has the same bomb offset as mainnet.
    Greerso
    @greerso

    no mechanism to tweak parameters mid-blockchain (yet)

    Changing ProgPoW knobs mid-blockchain possible/being explored as implied by the (yet)?

    I was wondering if it would make sense to use the Ravencoin 'nibbles' technique to randomly choose values for ProgPoW's knobs (knob nibbling?). Broadcast the values in the extra-field set of the block for miners. You wouldn't need to consider changing the values to thwart theoretical efficient fixed function hardware because the values would be changing constantly.

    Danno Ferrin
    @shemnon
    That sounds awesome for a “permanent” PoW chain (like ravencoin), but since ProgPow is meant as a bridge to Serinity I don’t know that we will ever see the value from the effort.
    Changing Params mid-chain would just involve miner (as in miners) co-ordination. However all the client verification code I’ve seen hard codes the constants. All that would be needed would be a set of variables in the client code and then mapping in when the schedule changes.
    Sonia-Chen
    @Sonia-Chen
    @shemnon nice link to solardiz zcash summary, thx!
    Jon Stevens
    @lookfirst
    Kicking a dead horse. https://github.com/ethereum-cat-herders/progpow-audit "August 16: Delivery of Initial Audit Report".... umm... looks at watch... August 20th... where is it?