13 Questions about Ethereum’s Movement to ProgPow
1. Why does Ethereum need ProgPoW?
Bitmain and Innosilicon developed 2–3x more efficient ethash ASICs, but they no longer advertise or sell them on their websites (unknown why). Linzhi announced a 6–7x ASIC coming in July, making ProgPoW a certain requirement. If Linzhi can build it, others can as well and they may create an unfair advantage or upset the ASIC-resistant mission of Ethereum.
Ethereum Network Hashrate has not changed much since the thirdening (lowering inflation from 3->2) and the difficulty bomb reset.
This means existing farms are still chugging along. The belief that miners would stop seems to have been incorrect as profitability has not changed significantly also due to a slight increase in price. If the price goes down, there is a stronger chance that ASICs can take over as unprofitable GPUs are turned off.
The amount of time it takes to make a PoW change like this, is at least a year. That is enough time to build an ASIC. There may be other PoW ideas out there, but this one appears to be the most technically effective.
Miners are the largest Ethereum userbase, with over 100k active wallets on the three largest mining pools. This is the first time in the history of Ethereum that miners and pools have come together on a vote like this. Miners add value by championing for Ethereum and spreading awareness.
Part of the proactive stance from miners is that ProgPoW levels the playing field to defeat unfair competition. Specialized tuning and ‘tricks’ (pills) used to gain performance from existing GPUs, will need to be rediscovered (if it even exists). FPGAs will need new bitstreams and may be rendered inefficient compared with GPUs.
2. Will ProgPoW stop ASICs?
Yes, for now. It will also stop FPGAs and even GPUs until new software is developed and deployed. There is no such thing as an ASIC-proof PoW and ProgPoW will not prevent ASICs from being made. It is a moot point to debate whether or not an ASIC can be built for ProgPoW. We know it can. But will it be worth it?
ProgPoW is an attack on the economic incentive to build an ASIC, not an attack on the ability to build one.
3. How will ProgPoW defeat the economic incentive of ASICs
ProgPoW makes the GPU into the perfect ASIC. It becomes economically difficult to build an ASIC that matches the price / efficiency of a commodity GPU. The cost to build a ProgPoW ASIC is estimated to be close to building a GPU from scratch.
Since we know it is still possible to build an ASIC, there is a second line of defense. Building a fully functional ASIC takes anywhere from 3–12 months or more. Ethereum has an existing ~6 month hard fork schedule. ProgPoW has a set of easily tweaked knobs (variables) that once changed will stop the next generation of ASICs.
Just threat alone of changing these knobs, will make it even more difficult for an ASIC manufacturer to produce hardware that will not be obsolete before it even gets to market.
Statements like this sound scary until you realize that Linzhi has strong economic incentives to create as much fear, uncertainty and doubt around ProgPoW as possible.
Pro-ASIC groups have a well documented history of this behavior. They are spending millions of dollars to build something that will be obsolete if ProgPoW is successful.
4. What about Proof of Stake?
ETH 2.0 and PoS is several years away from mainnet. Even then, it will live concurrently with PoW for years after that. In other words, PoW is not going away anytime soon and is therefore not a distraction from PoS.
Ethash has held strong for more than 3 years already, but the industry caught up. The threat of ASICs is real today.
If Ethereum loses the ASIC-resistance battle, when the time for PoS comes, the centralization of hash power to ASIC miners may allow them to be able to affect the transition to PoS.
5. Isn’t having only two GPU manufacturers centralized?
NVIDIA (founded April 1993, $9.7B revenue) & AMD (GPU founded 2006, $6.48B revenue) (N&A) both resell their chips to multiple OEM vendors (Asus, MSI, Zotac, Sapphire, etc.) that place them on boards. All of these vendors create a healthy and very distributed ecosystem which literally allows anyone (from hobbyists to large enterprises) to enter the game with fair competition and hardware distribution around the world.
Due to their size, they have access to large quantities of memory chips (4GB+ worth), necessary for ethash and ProgPow.
Both N&A and many of their OEMs are global publicly traded entities. We can count on the fact that they innovate, the next generation of chips will outperform their existing products. The revenue from miners is minor compared with the rest of their business.
On the other hand, there are only two existing ethash ASIC manufacturers (which no longer publicly sell their miners) and Linzhi (who has yet to release a product). All based in China. They are significantly smaller businesses and privately owned. They use the same chip fabricators. They are heavily reliant on income from miners and under financial duress in this bear market. They mine themselves, which means they compete against their customers.
Which set of companies would you want to secure your portion of a ~$14B market cap currency?
6. What is so wrong with ASICs?
Allowing the concentration of the mining service in the hands of a very few players (who have demonstrated in — recent — past to behave unfairly), exposes Ethereum to the risks being contended in the bitcoin-like hash wars.
Furthermore, ASICs are a dead end. When they are unprofitable they become e-waste. GPUs can mine multiple coins and be resold to gamers or repurposed for other tasks like Artificial Intelligence / Machine Learning / Online Gaming / Rendering.
7. What happens if Ethereum is dominated by ASICs?
The security of the network could be at risk. There can only be one major GPU-centric coin and it needs to remain Ethereum. If Ethereum loses its GPU dominance, it could risk being attacked or hard forked in the future.
If enough Ethereum miners decided to switch to a smaller hashrate GPU coin, they could perform a 51% attack. This attack has already happened to Ethereum Classic.
If Ethereum was a brand new coin today, for security reasons, it would probably be in their best interest to go with an ASIC design or a migration path over time to ASICs. Grin is a recent example of this.
8. I heard that ProgPow uses more power and hashes slower?
The design goal of ProgPoW is to have the algorithm’s requirements match what is available on commodity GPUs: If the algorithm were to be implemented on a custom ASIC there should be little opportunity for efficiency gains compared to a commodity GPU.
In other words, ProgPoW takes advantage of GPU-specific features, which requires more energy and generates fewer hashes per second.
However, miners are not concerned with speed or hashes/second, they are concerned with profitability through efficiency or hashes/watt.
The energy increase is offset by the reduction of Ethereum network difficulty, which is based on the total amount of Ethreum network hashrate.
In short, kicking existing ASICs and FPGAs off of the network and lowering the hashrate will automatically reduce the difficulty. Therefore, profitability will remain about the same.
9. What software changes need to be made?
In the context of ProgPoW, there are three layers that make up Ethereum: miners, pools and clients. All will require updates to their code during a hard fork.
Funding is needed for code development and that is proving difficult to raise because the Ethereum CoreDevs need to publicly state that they approve ProgPoW for a specific future hard fork, on the condition of successful analysis.
So far, this approval has been difficult to get primarily due to confusion and miscommunication in the community around ProgPoW. It is the intention of this post to help clarify the details.
10. How will we know ProgPoW is safe?
ProgPoW is a new version of ethash, with knobs. It needs independent analysis and testing, just like ethash received. A 3rd party must be funded to perform at least the following tests:
- Validate comparable NVIDIA vs. AMD to make sure there isn’t a bias. This has already been proven in the community with benchmarks, so it does not need to be a focus of the testing. But doing it at the code level is important too.
- Cryptographic security review of the code.
- Simulate (in a lab setting) the hard fork process to ensure the switch is smooth.
- Simulate twisting the knobs a second time (with another hard fork in the lab) to make sure that works.
Results need to be published and any issues resolved. Funding for this task is an open issue due to the lack of support from the CoreDevs. The paradoxical situation is that funding this is an additional strong signal for the CoreDevs.
11. When do we need ProgPoW?
There have been various reports of anywhere from 2–60% of the network being ASICs. My gut feeling is that the current ASICs on the market are not impacting the total hashrate significantly because they only offer a 2–3x advantage over GPU. When Linzhi releases in July, it will be 6–7x and therefore much more important that ProgPoW is ready to go in the next hard fork.
The largest issue is the amount of time it takes to make a PoW change. ProgPoW has been in development for almost a year already. Doing the analysis and scheduling it for a hardfork is another period of time.
Just having the threat of ProgPoW is likely slowing down ethash ASIC development and could be part of the reason why we haven’t seen updated models from Innosilicon or Bitmain.
12. What are the criteria for success?
Take away any unbalanced advantages on the network. Make all hardware compete based on price and merit. An $800 GPU should perform within 20% of any other equally priced hardware regardless of type or vendor.
13. How can I contribute?
Two grants need to get funded in order for ProgPoW to happen. CoreDevs won’t accept ProgPow until it passes analysis. In order to get analysis, there needs to be funding.
It is necessary to have an open source miner implementation. There was one, but it is now closed source because the developer working on it does not want to continue to work for free (or give his work away to paid miner apps like Claymore). Rightfully so.
So, we need to fund Andrea Lanfranchi. He is asking for about $80k and most of the work is already done. He is a top notch developer who has been working for months now on the ProgPoW implementation of ethminer, performance enhancements and an upgrade to the stratum mining protocol.
He is active in the community and patiently answers all questions. It is important to the community that there is an open source version of a ProgPoW miner. Until April 19th, 2019 this is also a matching fund!
Andrea recently published an update on things in his own words.
The second fund is … well. We don’t know yet. It was going to be a fund for analysis by a company called WhiteBlock, but as this story develops, it seems that is no longer the case. Please hold for an update.
Miners can contribute by testing ProgPoW on their hardware and reporting their results online. They can also spread the word about the need for funding (please share this post!).
It is healthy to have disclosures in articles like this. I HODL-long a very small amount of Ethereum and I am a cryptocurrency miner since 2014. I support whatever maintains the security and GPU-centric mission of Ethereum. I feel that is ProgPoW today, but if in the future if something else comes along, I will evaluate it equally as well.
If you have questions or concerns with anything I have said, please leave a comment! I consider this post a work in progress and will update it as necessary.