{"id":1314,"date":"2019-02-21T16:19:54","date_gmt":"2019-02-21T16:19:54","guid":{"rendered":"https:\/\/2miners.com\/blog\/?p=1314"},"modified":"2020-06-30T12:19:41","modified_gmt":"2020-06-30T12:19:41","slug":"ethereum-progpow-explained","status":"publish","type":"post","link":"https:\/\/2miners.com\/blog\/ethereum-progpow-explained\/","title":{"rendered":"Ethereum ProgPoW Explained"},"content":{"rendered":"

At the beginning of February, Ethereum developers decided to delay the integration of the ProgPoW algorithm. <\/p>\n

The project members are having doubts, so they want to wait for an audit by an independent company before making a decision. Developers themselves have different opinions about the new algorithm. So what is ProgPoW? How does it work, what is so special about it and how much can it damage ASIC performance? Let\u2019s get into it.<\/p>\n

\"\"<\/p>\n

ASIC Problem<\/h2>\n

Let\u2019s start with a bit of history. Ethereum miners and other Ethash\/Dagger Hashimoto coins have been quietly doing their thing with GPUs. GPU models differed in hash rate and performance, but these differences have always been reasonable. There was no super-GPU with unrivaled advantages, so everyone was happy.<\/p>\n

This ended in April 2018, when Bitmain officially announced the Antminer E3<\/a> that boasted a hash rate of 180 mh\/s and power consumption of 800\u00a0W. After that, other companies started producing ASICs as well.<\/p>\n

\"\"<\/p>\n

One might think, what\u2019s so bad about ASICs having better mining results? There are three problems.<\/p>\n

    \n
  1. GPU rig owners suffer from drastic loss of profits and lose interest in maintaining network operations;<\/li>\n
  2. Hash rate and power are concentrated in the hands of the minority;<\/li>\n
  3. Decentralization is under threat.<\/li>\n<\/ol>\n

    Of course, it takes more than a hundred or even a thousand ASICs to control 51% of the Ethereum network<\/a>, but the outlook is alarming. This is how the concept of the fight against ASICs was born, and it leads to three essential ideas:<\/p>\n

      \n
    1. opposition to potential network centralization;<\/li>\n
    2. prevention of monopoly among mining equipment manufacturers;<\/li>\n
    3. possibility to build the community without fear of outside threats.<\/li>\n<\/ol>\n

      Also, there are miners that have spent a lot of money on their rigs<\/a> and still haven\u2019t paid them off. The large number of ASICs in the Ethereum network may prolong the already long payback period.<\/p>\n

      \"\"<\/p>\n

      The ASIC problem concerns all GPU mining algorithms. How do developers fight against it? There are different approaches. For example, many creators of Equihash projects started switching to Equihash 144.5, because there are no ASICs for it yet. BitcoinZ<\/a> programmers were the first to hardfork the network and exclude ASICs from their network. Monero<\/a> developers hardfork the network several times per year, even though it doesn\u2019t seem to help them much – in a couple of months, the ASICs appear again.<\/p>\n

      You will be forced to throw your ASIC away in a couple of months because it has become obsolete. So why spend lots of money on ASIC production?<\/i><\/strong><\/p>\n

      ProgPoW<\/a> is a much more elegant solution. Instead of excluding ASICs from the network, the algorithm would make them less effective, or bring them on par with GPUs. As a result, few people will want to spend a large sum of money for minimal advantage, which will benefit good old graphics cards. This way developers will achieve their goals, and GPU miners will no longer be in danger. Everyone\u2019s happy.<\/p>\n

      What Is ProgPoW?<\/h2>\n

      ProgPoW stands for \u2018Programmatic Proof Of Work\u2019<\/i>. It\u2019s an extension of the current Ethereum algorithm Ethash and is designed to make graphics cards more competitive and minimize centralization. These tasks are achievable thanks to two algorithm features.<\/p>\n

        \n
      1. It regularly changes the statement of problem in mining.<\/li>\n
      2. ProgPoW uses all components of graphics cards to their full extent.<\/li>\n<\/ol>\n

        The latter point is especially important because ASICs have a serious drawback. Its name speaks for itself \u2013\u00a0Application-Specific Integrated Circuit. ASICs are designed to execute only one specific task.<\/p>\n

        This is the reason why they are so good at picking combinations for blocks solution. But this is the only thing they can do \u2013 you can\u2019t use them to play video games. GPUs, on the other hand, are flexible and can be used for both mining and gaming.<\/p>\n

        Also, the code of the new algorithm is published on GitHub. Everything is transparent. Let\u2019s see who is behind the ProgPoW code.<\/p>\n

        Who Created ProgPoW?<\/h2>\n

        \"\"<\/p>\n

        The algorithm was developed by IfDefElse team. The team consists of Ms. If, Mr. Def and Mr. Else. All team members conceal their identities, except for Ms. If \u2013 her real name is Kristy-Leigh Minehan<\/a>. The team solely focuses on proof-of-work mining algorithms.<\/p>\n

        By the way, Kristy-Leigh created miners well-known\u00a0OHGODANETHLARGEMENTPILL <\/i>for Nvidia GPUs overclocking.<\/p>\n

        How ProgPoW Works?<\/h2>\n

        First, let\u2019s see where the name comes from. ProgPoW means that each inner loop of memory access, that is, one iteration of problem-solving process, is a randomly generated program based on the block number. The program is designed for efficient operation of commodity GPUs and uses a greater portion of the GPU functionality.<\/p>\n

        Random sequence of problems makes it impossible to create fixed-function pipelines, and that\u2019s what ASICs manufacturers do. In other words, ProgPoW requires flexibility, and ASICs can\u2019t be flexible.<\/p>\n

        ProgPoW regularly changes the statement of problem in mining. GPUs are able to adapt quickly, thus narrowing the gap in performance compared to ASICs.<\/i><\/p>\n

        ProgPoW uses almost all components of graphics cards to their full extent. The exception is graphics pipelines and floating-point operations. The creators had to leave them out in order to make the algorithm compatible with equipment made by different manufacturers.<\/p>\n

        By the way, the IfDefElse team collaborated with Nvidia and AMD engineers to ensure impartiality. Both companies like the idea, because protection from ASICs makes graphics cards more relevant. And relevance means sales \u2013\u00a0right now AMD and Nvidia are not doing very well in this area.<\/p>\n

        Major Innovations of ProgPoW<\/h2>\n

        Let\u2019s take a look at five innovations of ProgPoW that distinguish it from Dagger Hashimoto. Thanks to them, the algorithm relies on computing power, bandwidth capacity and memory amount.<\/p>\n

          \n
        1. The size of the keccak is reduced from 64-bit keccak-f1600 to 32-bit keccak-f800. The latter is optimized for 32-bit platforms, which is great for graphics cards architecture.<\/li>\n
        2. Larger mix state. That is the larger number of processor registers that can store intermediate values. They are much faster than random access memory because they are inside the processor itself. ASICs have small chips and a smaller number of registers compared with graphics coprocessors. As a result, ASICs are forced to use RAM, which is much slower and deprives them of any potential advantages.<\/li>\n
        3. Addition of a random computing sequence to the main loop. This feature makes it impossible to create an ASIC with a fixed-function pipeline that could have increased operation speed or decreased power consumption.<\/li>\n
        4. Support of small-sized random cache reads with minimum delay. Cached reading forces ASICs to follow the rules of GPU memory hierarchy. This also limits their capacity and performance.<\/li>\n
        5. An increase of DRAM from 128 bytes to 256 bytes. This is good for GPUs. For the sake of better performance, specialized ASICs can\u2019t optimize memory controller for a small number of references.<\/li>\n<\/ol>\n

          ProgPoW Hashrate<\/h2>\n

          We tested the algorithm in real life on six graphics cards from two manufacturers. Since ProgPoW uses GPUs almost to the fullest extent, the results are similar to GPU performance when it runs games or graphic design apps.<\/p>\n

          Here are GPU hash rates on ProgPoW:<\/p>\n

            \n
          1. \n
            AMD Radeon RX 580 \u2013 11.56 MH\/s;<\/pre>\n<\/li>\n
          2. \n
            AMD Radeon Vega 56 \u2013 16.6 MH\/s;<\/pre>\n<\/li>\n
          3. \n
            AMD Radeon Vega 64 \u2013 18.7 MH\/s;<\/pre>\n<\/li>\n
          4. \n
            Nvidia GeForce GTX 1070 Ti \u2013 13.1 MH\/s;<\/pre>\n<\/li>\n
          5. \n
            Nvidia GeForce GTX 1080 \u2013 14.9 MH\/s;<\/pre>\n<\/li>\n
          6. \n
            Nvidia GeForce GTX 1080 Ti \u2013 21.8 MH\/s.<\/pre>\n<\/li>\n<\/ol>\n
            And now let\u2019s compare the performance of ASICs on ProgPoW:<\/p>\n
              \n
            1. \n
              GPU with GDDR on Ethash \u2013 x1;<\/pre>\n<\/li>\n
            2. \n
              GPU with GDDR on ProgPoW \u2013 x1;<\/pre>\n<\/li>\n
            3. \n
              ASIC with GDDR on Ethash \u2013 x2;<\/pre>\n<\/li>\n
            4. \n
              ASIC with GDDR on ProgPoW \u2013 x1.2;<\/pre>\n<\/li>\n
            5. \n
              ASIC with HBM on Ethash \u2013 x4;<\/pre>\n<\/li>\n
            6. \n
              ASIC with HBM on ProgPoW \u2013 x1.5.<\/pre>\n<\/li>\n<\/ol>\n
              As a result, the ASIC advantage decreases by as much as 2.6 times. GPU rig owners would not have to worry about their future and potential centralization of cryptocurrency projects.<\/p>\n
              Opinions about ProgPoW<\/h2>\n
              The ProgPoW integration has been in question since August 2018. ETH Core developers first raised the issue on January 4th. All in all, the majority of leading programmers are in favor of the idea but want more time for testing and implementation.<\/p>\n
              Criticism by the Community<\/h3>\n
              Opponents of the ProgPow think that the integration of the algorithm would mean the following:<\/p>\n