## **RNP 002: Layer 1 Network Expansion** --- |RNP # |Title |Category | Author |Created | Status |------|-------|---------|--------|--| -- | |002 |Layer 1 Network Expansion |Core-Technical|Ryan Shea|03-20-2023 | Approved and on the Roadmap | --- ### **Overview** After a successful [[vote on RNP001]](https://github.com/rendernetwork/RNPs/blob/main/RNP-001.md) introducing a Burn and Mint Equilibrium (BME) model, it is proposed that the Render Foundation invests resources to build this on the Solana blockchain. In anticipation of future network growth, there exists a need for a high throughput, low latency blockchain for network settlement and node operation. Blockchains were evaluated with the following criteria: developer community, throughput, liquidity, transaction fees, programming languages, integrating smart contracts, speed of implementation and the maturity of the project. After an exhaustive review---including input from the community---it is proposed that Solana is the best option for the future of the Render Foundation and that the BME model be implemented on Solana. ### **Category** This is a core proposal within the technical subcategory. ### **Motivation** The key advantage of the switch to a new L1 is to help scale the network to meet the demand of our technical needs and implement the BME. ### **Stakeholders** This proposal impacts all members of the Render Network community. ### **Implementation** **Researching options, community input** The [community sentiment polling](#appendix) provided a number of layer one networks to review. The results showed strong support for Solana, then followed by our current platform, Polygon. Given the Render Network is a mature project with millions of frames processed annually, it was not a surprise that these options were preferred over some of the newer layer ones and scalability solutions that have not yet proven scale, stress under load, on- and off-chain liquidity, or would require significant technical investment from the Render Foundation. The decision whether to continue with Polygon versus moving to Solana sparked a spirited debate between the pros and cons of each, with support for Solana's runtime, developer community, low transaction fees, and speed contrasting Polygon's centralization, reorg tendency, and user experience concerns. These are outlined further below. ### **Programming language, throughput, and smart contract integration** The Render Network has been built with a goal towards providing real-time, rendered, interactive holographic streams. This will require states to be updated dynamically, in real-time and on-chain. To do so, the network needs to be able to write high-performance code both on-chain and on GPUs. Rust can provide the Render Network greater speed and flexibility than Solidity, ultimately permitting the use of the same code for GPU render work and smart contracts. Further, the Solana virtual machine can process both C++ or Rust code and turn that into a smart contract. It is also important to note that once real-time, rendered, interactive holographic streams are being streamed on the network and on-chain, transaction costs will start to play an important role. Solana's blockchain has both higher transaction throughput and lower transaction fees than Polygon.\ \ ETH runs at roughly 14 TPS as you can see in this [live TPS](https://blockchair.com/ethereum/charts/transactions-per-second). Polygon averages just more than double, with roughly [29 TPS](https://polygonscan.com/), as you can see in this [live TPS here](https://dune.com/queries/1390025/2362801). Finally Solana currently averages over 140 times that of Polygon, with roughly 4,080 TPS -- here's a [link to live Solana TPS](https://explorer.solana.com/). Both Polygon and Solana claim they are built to achieve over 50,000 TPS in the long term. Moving forward, the Render Network would like to include more data on-chain for both current and future offerings, providing greater transparency for token holders and node operators alongside greater utility for artists. However, major EVM-based chains do not have the transactions per second (TPS) required as the project continues to scale. At peak moments, the Render Network could already have more frames completed within a single second than ETH's TPS, and as we look toward 10x or 100x growth along with more data written to the blockchain itself, it becomes problematic without further obfuscating the processes & information stored on-chain. Even for existing use-cases, certain network designs are not scalable due to transaction fees. For example, a 200 frame render job running on a network where data is written to the chain and payment sent to node operators as each frame completes would require 200 transactions. ETH gas fees trend [around an average of \$0.70 per tx](https://ycharts.com/indicators/ethereum_average_transaction_fee). On ETH mainnet, this would cost \$140 in fees at the current gas rate. Polygon gas fees trend around \$0.025 per tx (this one is a bit more of a rough calculation - the [daily tx count can be found here](https://polygonscan.com/chart/tx) and the [daily total network fees here](https://polygonscan.com/chart/transactionfee)). On Polygon, this would cost \$5 in fees at the current rate, while on Solana it would cost under \$0.001, a reduction of roughly 5000 times over Polygon. Further these same 200 transactions could cost as little as \$0.00026 on Solana (19,000 less) if priority fees were not needed. While \$5 in gas fees may sound reasonable as many have been desensitized by higher prices, the Render Network is a utility where the cost of compute power used on the job could be significantly less than those fees, effectively meaning transaction fees have the ability to raise the price of compute work exponentially. In practice, adding that amount of write volume to an EVM chain would also inflate gas prices further and likely exceed those estimates while hindering user experience. As additional use-cases become available, such as holographic streaming, even moderate transaction fees could limit creators' ability to innovate. Certain applications of NFT technology - token gated ticketing of large scale streaming experiences, massively online realtime gaming programs, social media microtransaction based media, or micro asset economies for example - require minting of significant amounts of data to be encoded on the blockchain that go well beyond simple the basic tokenization of completed frames, which number in the millions. For example, a single render can include thousands of unique assets as individual nodes within its scene graph - including 3D models, textures, scripts, HDRIs, effects, custom render graph nodes, volumes, rays , sdf / meshes, and scene graph states. Many applications of the Render Network for real-time streaming experiences, gaming, generative art, and emergent forms of composable media require all of these unique node graph assets to be tracked on chain. Such a system requires multiple orders of magnitude TPS increases when compared to the TPS requirements for processing rendered frames. On-chain transaction frameworks are expensive for these emerging applications that network participants are requesting to access via the Render SDK. Not only does Solana offer significant increases in TPS throughput when compared to other L1's but [emerging Merkle Tree compression for NFTs developed by Solana Labs and Metaplex](https://www.metaplex.com/posts/expanding-digital-assets-with-compression-for-nfts) enable exponential increases in the ability to tokenize node graph assets for next generation blockchain applications based on the Render Scene Graph. Because Merkle Tree compression enables more data to be put on chain in an atomic format at orders of magnitude higher volumes than what is currently available, the network can tokenize entire scene graphs in addition to completed frames, opening up a new economy of 3D applications for network users. For example, each NFT based on a render frame or animation could be atomically decomposed into smaller NFTs that assemble a node graph where one can explore each individual piece and item that shares a parent / child. New forms of composability would then be possible where one could jump from \"NFT -\> assets -\> other NFTs using that asset\" to find out how influential an NFT is based on how many others use its original assets or explore asset / NFT relationships for whatever reason you may have - whether royalty flows, creative remixing, AI and semantic intelligence. These additional feature sets open up a host of applications that Render Network users are building, including dynamic virtual asset collections, remixable or composable NFTs, and immutable expanding large scale public NFT archives. ### **Developer community and speed of implementation** Historically the Render Network has struggled to attract and retain top Solidity developers. In a [Dec 2022 Stack Overflow Survey](https://survey.stackoverflow.co/2022/), Rust was listed as over 5X more popular than Solidity. There has been a recent emergence of protocols which incentivize people to do verifiable work similar to the verifiable GPU compute work performed on the Render Network. Solana has attracted a higher share of such projects including Helium, Hivemapper, and Audius. Helium's April 2023 migration to Solana will provide important assurance that such a move is possible, and will be the first live implementation of the BME on any blockchain (besides Helium's proprietary chain from which they are moving away). Much of Helium's Solana BME project is open sourced and could be leveraged to significantly speed up the Render Network's implementation if Solana is selected. ### **Security** On any major blockchain, the most common attack vector is the smart contract itself, and in that regard a project is vulnerable on any blockchain. When discussing the possibility of a hack, the robustness of the programming language and the skill of the developer are the primary factors to consider, and Rust excels as a programming language. When combined with the deep talent pool of experienced Rust developers, a contract on Solana is not going to be more vulnerable than a contract on any other network. The "Nakamoto coefficient" is a metric coined by Balaji Srinivasan and named after Bitcoin's creator, and is defined as the smallest number of validators who cumulatively stake 33% of the network's staked tokens. In theory, depending on the network's consensus mechanism, a collusion of these validators would be able to censor the network's transactions or halt it altogether. Thus, a higher Nakamoto coefficient represents a higher distribution of staked tokens, and a higher degree of decentralization. Solana's [July 2022 validator health report](https://solana.com/news/validator-health-report-august-2022) listed Solana's Nakamoto coefficient at 31, over 10 times Polygon's. In addition, a June 2022 [Cointelegraph report](https://cointelegraph.com/news/solana-said-to-be-more-decentralized-than-people-think-but-there-s-more) listed Solana's active validators at 1,875, over 19 times the 98 validators on Polygon. ### **Liquidity and uptime** Ethereum trumps Solana with respect to maturity, and as a result security and liquidity. That being said, both SPL and Matic tokens are listed on all current major CEX's that list RNDR. RNP001 provides the ability for emissions to be allocated as rewards for Liquidity Providers who participate in decentralized liquidity pools which enable the automated purchase of RNDR via smart contract by creators, which should further assist with liquidity. All the major chains have gone through growing pains (see [2016 ETH hard fork after the DAO hack](https://www.gemini.com/en-US/cryptopedia/the-dao-hack-makerdao#section-the-dao-hack) & the [2013 Bitcoin rollback](https://news.bitcoin.com/bitcoins-software-has-been-rolled-back-before/)), and Solana has not been immune, with a [string of outages in 2022](https://cointelegraph.com/news/network-outages-have-been-solana-s-curse-says-co-founder). While Solana's uptime has improved since becoming a 2022 focus, it is not yet at enterprise level, and a lot is expected from [Firedancer](https://decrypt.co/111486/solana-co-founder-long-term-fix-network-outages) to allow them to achieve this. While different from downtime, [ETH often has 150k+ pending transactions](https://etherscan.io/chart/pendingtx) -- users can spend more gas to get ahead, but someone who submitted a transaction with underpriced gas could experience significant delays and resulting in a user experience similar to that of downtime. Higher gas fees will also not enable the Render Network's real-time streaming goals. In order to achieve its real-time streaming objectives, the Render Network will need its blockchain to maintain an enterprise level uptime not available on either Polygon or Solana today. To mitigate, the BME model should initially be implemented without real-time dependencies, until such time as both uptime and low cost throughput are more established. To do so, the smart contract would be constructed to remove any on-chain dependencies by moving the timing of any decentralized RNDR purchase and burn process to after the job performance and approval, and allowing for it to be batched should the blockchain experience any downtime. Further, Emissions would be minted in blocks in advance of distribution. The Render Network already runs RNDR payments to Node Operators in a way that can be done asynchronously, and that off-chain record keeping should be carried over to future changes as well. ### **Migration** This proposal has focused on the question whether the new BME model should be implemented on a different blockchain. Should the community agree on the proposed implementation on Solana, consideration should be applied on whether or not a full migration of existing Polygon tokens is required. If not, being multi chain has significant benefits with respect to liquidity. Regardless, a smart contract update will be needed to facilitate. This should be a topic for future RNPs. ### **Other Considerations** Transferring the primary smart contract may come with unforeseen effects, similar to the previous expansion onto Layer 2 Polygon ("L2"). In addition, the transference of chains may make the current L2 solution either obsolete and/or incompatible with the smart contract, therefore requiring additional considerations for the change of L2 solutions, such as an update to the existing smart contract. As noted previously, this should be a topic for future RNPs. ## **Appendix** The Community Sentiment Polling results showed strong support for Solana followed by Polygon. ![](RNP-002/image1.png) Projections of cost savings from Merkle Tree NFT compression technology ![](RNP-002/image2.png)