Souptacular commented 6 years ago

Ethereum Core Devs Meeting Constantinople Session 1 Agenda

Meeting Date/Time: Friday 31 August 2018 at 14:00 UTC

Meeting Duration 1.5 hours

Agenda

Three competing EIPs to delay the difficulty bomb and reduce/maintain the block reward: a. EIP-858 - Delay bomb and reduce block reward to 1 ETH per block. b. EIP-1234 - Delay bomb and reduce block reward to 2 ETH. c. EIP-1295 - Delay bomb, keep rewards to 3 ETH, change other factors such as POW incentive structure.
Constantinople EIPs a. EIP 145: Bitwise shifting instructions in EVM b. EIP 1218: Simpler blockhash refactoring. We are delaying this one until Istanbul. c. EIP 1014: Skinny CREATE2 d. EIP 1052: EXTCODEHASH Opcode e. EIP 1283: Net gas metering for SSTORE without dirty maps
Client Constantinople Updates
Constantinople Testing
(If we have time) Changing PoW algorithm to be ASIC resistant. There is renewed interest from miners to implement ProgPoW.

Different articles/links regarding potential issuance reduction conversation:

nico9111 commented 6 years ago

There's already ETC for that. ETH is moving on with POS

justinjja commented 6 years ago

@chfast What about a small tweak to the current Ethash algo to stop asics in the meantime? similar to what Monero did with CN to CNv1? https://github.com/monero-project/monero/pull/3253

Would be a much smaller change, and would keep miners happy until Progpow can be fully tested.

MoneroCrusher commented 6 years ago

@justinjja There will be an ASIC on day one if the change is too simple. The network is paying Bitmain & Innosilicon (which doesn't seem to bother ETH devs) several millions per day, they won't just sit around. It has to perma brick ASICs for good, which ProgPOW does.

I fail to see how the community doesn't feel like decentralization (security) is any Blockchain's biggest asset.

olalawal commented 6 years ago

@justinjja it would be better than nothing, at least if its pushed before they can tape out new asics we will be able to see what the true ASIC hash rate is.

@monerocrusher I support https://eips.ethereum.org/EIPS/eip-1355

at least as a stop gap for that reason

olalawal commented 6 years ago

@monerocrusher If there is a fork of constantinople with progpow implemented unless the Eth devs at leas brick existing asics on the offical fork so we can get the true scale of ASIC infiltration of the network , I an the group network of gpu miners I am in contact will definitely throw our support behind it.

Ill vote with my modest 3gh farm and alot of them will follow suit I imagine.

Do you have a link to this proposed ProgPow fork ?

Etc is really not an option, they DAO hack fork is not one many of us would look to as a vaiable network to support

justinjja commented 6 years ago

Nice, somehow I missed EIP1355. Will it be included in constantinople?

olalawal commented 6 years ago

@justinjja looks like no

I can’t belive someone mentioned ETC lol , yall are really reaching here.

I think GPU miners will stay on the old chain after constantinople with a ProgPow update.

There will be no place for us on th old network anyways , its a logical move

5chdn commented 6 years ago

No. Window for new EIPs closed a while back. All new proposals have to go into the subsequent Istanbul hardfork.

justinjja commented 6 years ago

You want to wait as long as possible before posting a change like this, to prevent ASIC manufactures from getting a head start.

If you really can't put a new EIP in, you can include it with EIP-1234, which is getting changed anyway.

5chdn commented 6 years ago

1234 is already changed and accepted. https://eips.ethereum.org/EIPS/eip-1234

MoneroCrusher commented 6 years ago

So even though community supports forking ASICs off you don't deem it important. So much for respecting the GPU miners that kept you secure over the years.

If there's a community driven fork to:

implement ProgPOW
merge pulling all EIPs that Ethereum does otherwise (including BR reduction to 2 ETH) in the future
reduce DAG size to 750 MB and fix it there (because DAG increase is not necessary anymore with ProgPOW)

This would provide most decentralization as people with 1 & 2 GB cards will also be able to mine & secure the network.

I would be ready to point my farm to it (not disclosing exact number but its more than 20 GH/s and less than 100 GH/s). Absolutely disappointed with the IMO bad decision of ETH devs with regards to ASICs

MoneroCrusher commented 6 years ago

BTW ETH issuance reduction only will make it barely profitable for me to mine. Think about all the hobbyist miners paying 20c for their household electricity, this will definitely put them out of business. You're losing more & more miners

salanki commented 6 years ago

And gain more and more ASICs controlled by the same few companies

5chdn commented 6 years ago

I don't think the core devs are opposed to implementing progpow. But it's a very involved change and there is no way we can properly implement and test this before Constantinople. That's why it's realistically being dealt with on the way up to the Istanbul fork.

MoneroCrusher commented 6 years ago

@5chdn @ ALL ETH DEVS Did you by chance miss OhGodAGirl stating that she and her devs can easily implement ProgPOW into Constantinople and that she would gladly provide all assistance & coding free of charge? The majority is done.

Here, read it for yourself:

So much misinformation about ProgPoW here, and what it does, and how it works.

First and foremost, it's not just a tweak that invalidates ASICs. Anyone involved in the development of hardware knows that these tweaks don't actually do anything when a sequencer and a creative architecture team comes into play, and with the development frame of ASICs, you're looking at about two months before a new revision with an alt-mask comes out.

Secondly, CNv2 and all Cryptonite variants are historically flawed. On a side note, when you go to tune something for multiple variations of a hardware, you run into the issue of actually becoming less centralized-resistant. It's (part) of the reason why ProgPoW is tied to 32-bit, rather than 64-bit.

Thirdly, ProgPoW actually takes very little work to implement in Constantinople. It requires client asistance, but that's about it. The hard part - the GPU architecture - is done. With enough serious adoption, I wager that larger companies will even throw their development efforts behind it. The ask for CNv2 to be adopted is baffling to me, because anyone in the know already knows that there are masks for the damn thing. Should I just start a Medium page, with designs for all these algorithms? If you don't believe me, go believe someone who lives and breaths silicon (someone like David Vorick, for instance).

Fourth, just because code hasn't seen widespread adoption in mainnet coins, doesn't mean it is a risk to implement. That is why the code is open source - you can pretty much catch any issue before being deployed. That's called peer review, folks. Innovation requires taking a chance sometimes.

You don't want to break existing ASICs, you want to tune the algorithm to a piece of hardware (a GPU) that everyone has access to, that is used for other things (not just cryptocurrency mining), and has natural resale value. Actually, there's a ton of neat tricks you can do (like ensuring its specialized on a single card, rather than multiple to benefit Average Joe Miner).

Fifth, ProgPoW is designed to not break ASICs, but to encourage development of open-source GPU cards. Creating an "ASIC" for ProgPoW will end up just mimicking a GPU. Open-source hardware is a wonderful thing. We need more diversity in this space, rather than the big three (Intel, AMD, NVIDIA). Imagine the kind of innovation we can see in a programmable world if everyone and their aunt has access to open-source graphic cards...

The constant whack-a-mole game with ASICs is tiring and a waste of development efforts. Please, don't go down that route.

I am really tickled pink that folks are pushing for ProgPoW, but please understand the algorithm's purpose isn't just to preserve cryptocurrency mining: it's to preserve a future powered by the most energy-friendly, compact, programmable pieces of hardware out there (GPUHoarder, don't you dare start).

Again, as I have said, if there is serious adoption this time by the governance of Ethereum, I am willing to spend the (personal) money to not only staff up, but lead the development effort.

chfast commented 6 years ago

I am willing to spend the (personal) money to not only staff up, but lead the development effort.

If you want to help you can benchmark ProgPoW on as many GPU models as possible.

salanki commented 6 years ago

@MoneroCrusher: From a fellow miner; it's not that easy. Even if all of the code was done, the amount of testing, review and coordination such a major change requires is not something that you can rush in a couple of weeks. It is not something specific to Ethereum, it is just how software development of production code works. The best thing we can do is to test it ourselves. Get a testnet up, run benchmarks on GPUs and do longevity tests to make sure it is stable.

MoneroCrusher commented 6 years ago

@chfast I was quoting OhGodAGirl Github on old ios device is quite terrible for me so I didn't even try to format it. Anyway here are some stock benchmarks:

Model Hashrate (MH/s) RX580 9.4 Vega56 16.6 Vega64 18.7 GTX1070Ti 13.1 GTX1080 14.9 GTX1080Ti 21.8

There's also a testnet so I can bench all the GPUs you want if it helps ProgPOW making it into Constantinople.

@salanki In 156 hours there will be a mainnet launching ProgPOW. They have a tiny dev team and managed to do it. I think 8 weeks of testing is plenty...

MoneroCrusher commented 6 years ago

Here's the testnet of ProgPOW if anyone wants to try themself (don't worry it's not gonna bite u :-))

"ProgPOW testpool up! Windows miner: https://www.dropbox.com/s/t427mpf0kzyidrg/Progpow.zip?dl=0 ....... Miner source https://github.com/BitcoinInterestOfficial/ProgPOW startup for testpool: -P stratum+tcp://mgomtozjYh6Xx5qL642bcskX2cikpqQVM9@37.16.104.240:3033"

mohsenghajar commented 6 years ago

I'm ok with forks.

On Fri, Aug 31, 2018, 1:15 PM MoneroCrusher notifications@github.com wrote:

@chfast https://github.com/chfast ProgPOW will be live on a mainnet next week in approximately 158 hours. OhGodAGirl helped them implement it, she said she could very easily offer several developers to implement it in Constantinope, for free that is. GETH is done, ethminer implementation is done, benchmarks were made.

Can you let us, the community, in on your thought process of rejecting it in Constantinople. The community won't accept a simple "no time". Please be transparent to the community what exactly there is no time for. Each piece of the puzzle that is missing to accomplish it. I'm in no way affiliated with ProgPOW or the developers behind it, I'm just a simple miner.

But I know: if you reduce BR without honoring miners and decentralization, there will be a fork. I'm not personally doing one but I've read about numerous people willing to support it, and I really hope it won't end that way, as it would probably damange crypto and the market if it were to happen. There's also always the option to delay Constantinople until you are sure a majority of the community will support it. Supporters of BR issuance reduction only all believe BR reduction will make ETH "moon", let me tell you: it won't.

Do you know what surpresses the price? Bitmain and other manufacturers mining ETH on a big scale in secret, putting constant market pressure on the books because they don't care about ETH. They'll sell it to buy BTC/BCH, the only thing that is more equalized is a big chaotic community of miners, each deciding themselves what they do with their coins, and not the investors behind them, which don't exist in the first place, mostly.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ethereum/pm/issues/55#issuecomment-417732448, or mute the thread https://github.com/notifications/unsubscribe-auth/ANH0wj5iLRV3yRjMV5G3yjq9poKpNAyLks5uWW8igaJpZM4WQsxW .

olalawal commented 6 years ago

@monerocrusher you should stop wasting your breath here I think the ship has sailed on an Asic Resistance update for Ethereum.

I think we should spend our energy on Bci since they are focusing on progPow , that way we could roll alot of those changes into a potential fork on the mainnet Constantinople rollout.

Im positive a different chain will be supported by gpu miners since no sane miner will mine on the main chain at a loss

Let asics have the mainet like the took over the zcash mainnet.

Im also spinning up a znomp pool for bci to test progpow this weekend in anticipation of a successful fork.

olalawal commented 6 years ago

found this post on reddit its summs the current situation up perfectly :

Exactly, decentralized mining leads to widespread adoption, centralization in mining leada to stagnation which is what we are seeing now. Now of course if dapps start to gain usage or scaling solutions take off you will see a boost.

Its strange how you guys and the devs seem to have forgotten how we got here

earthloong commented 6 years ago

@MoneroCrusher I suspect that you should have a large number of graphics card mines. Your point of view is to maintain the interests of a large number of graphics card mining machines or some graphics card manufacturers, and not the future of ethereum, a mature community or public chain platform should be much more A variety of miners and mining equipment to maintain an overall balance.

MoneroCrusher commented 6 years ago

@earthloong Now, do you prefer giving Bitmain 40-50% of the network, or 0.0075% to 0.035% of the network to me and dozens of thousands other people like me? Do you see the "future in ETH" as youve mentioned by ASIC mining monopolies? Do you think Bitmain gives a single shit about ETH?

Check their investor's deck, they seem to have almost no ETH, do you know what that means? Yes, they're mining ETH, worth 3 million per day (up to 50% of emission) and are constantly putting them on the market to buy their BTC & BCH, driving ETH price down. Would you rather give all those ETH to Bitmain, a single entity, that have a huge political agenda to promote their shitcoin and generally hurting the market, or would you rather give it to thousands of people like me, who see the large future potential and that are more likely to hold, effectively reducing "ask book" pressure? Thousands of miners & I have aligned ourselves (consciously or by chance) with decentralization by buying hardware that isn't specially made for ETH mining, Bitmain & Innosilicon are the cancers of this space, they are good at what they do but have very scammy marketing schemes, they'll use their miners to mine in secret and once their second gen is developed they'll sell you their old version, giving you the illusion that you bought awesome hardware, which will be obsolete very fast. Bitmain's internal ASIC is the E9, now they're selling E3, their old product. They do that to offroll risk and to appear as if they're a hardware selling company, when in fact they're also a mining company. You could blow them out of the water with a fork to ProgPOW (brought for free to the crypto space by OhGodAGirl, who has been developing miners in this space before ETH was even born), and give the power back to thousands of people, making the network inherently more decentralized, therefore more secure & ultimately more valuable.

tl;dr Investors could care less about which hashing algo is used in ETH, they won't notice a change in the overlying protocol. They only change they'd notice is rising prices because thousands of miners is better for price than a single entity selling 50% of daily emission.

MoneroCrusher commented 6 years ago

@olalawal I think it's very important to spread awareness that centralization, which as of the moment will happen at Constantinople fork is extremely harmful to ETH (and other CCs). Everybody's thinking resucing issuance will make ETH moon, this is very wrong, all what will happen is a further network shift to ASIC monopolies, boosting them far beyond 50% of network composition. And since they are the ones putting on daily sell pressure, it's quite easy to imagine that ETH might decline even further, as effectively Constantinople in current state will fork off all "miner holders", aka hobbyst miners & smaller farms.

justinjja commented 6 years ago

I do think it is worth forking to ProgPow or EIP1355, but 40-50% sounds like a huge stretch. It's not like Cryptonight where asics can beat a gpu by 100x, Ethash can only be optimized by about 2x. Honestly I would bet that the percentage of Asics on Ethereum is lower than Monero right now.

MoneroCrusher commented 6 years ago

@justinjja I don't think it's a stretch at all. Check my previous approximation based on GPU sales:

"How can you say ASICs are only 10% of the network? I can also say it's 60% of the network. Both are opinions. Actually let me show you some arguments supporting my claim for ASICs making up ~40% of the network: According to your own article you say $1.7 billion (in GPUs) were invested from September to March. Yes, it would have cost $1.7B in GPUs if the hashrate came solely from GPUs, but it didn't.

According to Jon Peddie research ( source: https://www.jonpeddie.com/press-releases/gpu-market-declined-seasonally-in-q4-cryptocurrency-provides-smaller-offset) a total of 3 million GPUs, worth $776m were sold to miners in 2017 by both manufacturers. In Q1 2018 GPU sales went down by -10% compared to Q4 2017 (source: https://www.jonpeddie.com/press-releases/gpu-market-increased-year-to-year-by-3.4/). And according to this article (https://www.ccn.com/declining-gpu-sales-to-cryptocurrency-miners-healthy-amd-ceo/), AMD and Nvidia both confirm sales of $165m and $289m to miners, or $454m for Q1 2018. Amd reported sales of GPUs to miners in Q2 2018 to be $65m (https://www.coindesk.com/amd-sees-q2-drop-in-gpu-sales-to-crypto-miners/).

So now we can do some math: If $454m worth of GPUs were sold to miners in Q1 2018, and it was down 10% from Q4 2017, then during Q4 2017 approximately $500m worth of GPUs were sold to crypto miners or a total of $954m worth of GPUs in Q4/17-Q1/18. If we take a RX 570 as a benchmark (doing 30 Mh/s) and say it sold for $300 (that's a low price for a Q4 2017 RX 570), then the hashrate should have increased by 95 TH/s. From the 1st of October to the 1st of April 2018 (Q4-Q1) the network hashrate rose by over 160TH/s.

Now what about the rest of 2017? According to our numbers, $276m worth of GPUs should have sold during that timeframe to miners (Q1 virtually nothing, because crypto wasn't hyped back then, and then it slowly rose during Q2 and at the latest by Q3 people were crazy for GPUs and crypto).

At the start of 2017 the network hashrate for Ethereum was 5.96 TH/s (https://bitinfocharts.com/de/comparison/ethereum-hashrate.html). Therefore by the end of Q1 2018, we should have had a total network hashrate of $776m+$454m/$250 per RX 570 = 4'920'000 RX 570 or 147.6 TH/s. We had 110 TH/s more than that by end of Q1 2018.

Where did that come from? Yes, probably Bitmain and other ASIC manufacturers (as was also evident in Bitmain secretly mining Monero and owning over 80% of the hashrate). So the network hashrate of Ethereum as of this moment is more likely to be 35-45% ASICs, according to my estimations. Open for debate."

Here I even completely forgot about other networks (it was late in the night)! Since a big amount of GPUs went to mine other networks like Monero, Zcash, ETC and others, then even 50% is quite a conservative estimate. The only way to to find out is through ProgPOW fork, and everyone will have an "Oh" moment, like when Monero forked ASICs off in April and it was revealed Bitmain owned 80% of nethash. Bitmain even lied about it and said they didn't secretly mine Monero (but they conveniently put their miners up for sale 2-3 days after fork announcement had been made...right?! lmao) and some people even bought them for $13k a piece.

earthloong commented 6 years ago

@MoneroCrusher I don't support bitmain's mine behavior, but for ethereum or other public blockchain platforms, the consensus mechanism should be widely used to allow as many mining equipment as possible, not one (such as graphics cards). Moreover, the participants are more specialized (such as asic), but also have ordinary or discrete (such as video cards, mobile phones, etc.), which is more in line with the autonomy and operational logic of the public blockchain platform. In addition, as far as the present, asic does not pose a threat to Etherum (there is no direct evidence), we can use btc to compare, btc has such a large value (market value and security), the important reason is that it uses Pow algorithm and powerful asic computing support. The bitmain you mentioned, I know only the E3 mining machine, and basically the failed product. As for the so-called E9, I don't know where you are from? (Is there any evidence?), hasn't the other ethereum mining machines yet appeared on the market? So where does the so-called asic threat come from? Is it worrying?

mohsenghajar commented 6 years ago

I am not saying the narrative is wrong but the study has a lot of holes. For example, the possibility of hash rate coming into Ethereum from other networks was not considered. A lot of people I know went from playing fallout to mining Ethereum. Another, too much reliance on GPU manufacturers data and its quality. They are not obligated to try hard and give accurate miner sales number (they can't know for sure). So they're just rough estimates.

On Sat, Sep 1, 2018, 1:10 AM MoneroCrusher notifications@github.com wrote:

@justinjja https://github.com/justinjja I don't think it's a stretch at all. Check my previous approximation based on GPU sales:

"How can you say ASICs are only 10% of the network? I can also say it's 60% of the network. Both are opinions. Actually let me show you some arguments supporting my claim for ASICs making up ~40% of the network: According to your own article you say $1.7 billion (in GPUs) were invested from September to March. Yes, it would have cost $1.7B in GPUs if the hashrate came solely from GPUs, but it didn't.

According to Jon Peddie research ( source: https://www.jonpeddie.com/press-releases/gpu-market-declined-seasonally-in-q4-cryptocurrency-provides-smaller-offset) a total of 3 million GPUs, worth $776m were sold to miners in 2017 by both manufacturers. In Q1 2018 GPU sales went down by -10% compared to Q4 2017 (source: https://www.jonpeddie.com/press-releases/gpu-market-increased-year-to-year-by-3.4/ ). And according to this article ( https://www.ccn.com/declining-gpu-sales-to-cryptocurrency-miners-healthy-amd-ceo/), AMD and Nvidia both confirm sales of $165m and $289m to miners, or $454m for Q1 2018. Amd reported sales of GPUs to miners in Q2 2018 to be $65m ( https://www.coindesk.com/amd-sees-q2-drop-in-gpu-sales-to-crypto-miners/).

So now we can do some math: If $454m worth of GPUs were sold to miners in Q1 2018, and it was down 10% from Q4 2017, then during Q4 2017 approximately $500m worth of GPUs were sold to crypto miners or a total of $954m worth of GPUs in Q4/17-Q1/18. If we take a RX 570 as a benchmark (doing 30 Mh/s) and say it sold for $300 (that's a low price for a Q4 2017 RX 570), then the hashrate should have increased by 95 TH/s. From the 1st of October to the 1st of April 2018 (Q4-Q1) the network hashrate rose by over 160TH/s.

Now what about the rest of 2017? According to our numbers, $276m worth of GPUs should have sold during that timeframe to miners (Q1 virtually nothing, because crypto wasn't hyped back then, and then it slowly rose during Q2 and at the latest by Q3 people were crazy for GPUs and crypto).

At the start of 2017 the network hashrate for Ethereum was 5.96 TH/s ( https://bitinfocharts.com/de/comparison/ethereum-hashrate.html). Therefore by the end of Q1 2018, we should have had a total network hashrate of $776m+$454m/$250 per RX 570 = 4'920'000 RX 570 or 147.6 TH/s. We had 110 TH/s more than that by end of Q1 2018.

Where did that come from? Yes, probably Bitmain and other ASIC manufacturers (as was also evident in Bitmain secretly mining Monero and owning over 80% of the hashrate). So the network hashrate of Ethereum as of this moment is more likely to be 35-45% ASICs, according to my estimations. Open for debate."

Here I even completely forgot about other networks (it was late in the night)! Since a big amount of GPUs went to mine other networks like Monero, Zcash, ETC and others, then even 50% is quite a conservative estimate. The only way to to find out is through ProgPOW fork, and everyone will have an "Oh" moment, like when Monero forked ASICs off in April and it was revealed Bitmain owned 80% of nethash. Bitmain even lied about it and said they didn't secretly mine Monero (but they conveniently put their miners up for sale 2-3 days after fork announcement had been made...right?! lmao) and some people even bought them for $13k a piece.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ethereum/pm/issues/55#issuecomment-417833751, or mute the thread https://github.com/notifications/unsubscribe-auth/ANH0wvk9IU2-CR-m5ZLN_VS2E8HAcj43ks5uWhbQgaJpZM4WQsxW .

lrettig commented 6 years ago

Closing as I think nothing is outstanding here, @Souptacular reopen if I've missed something.

ethereum / pm