BTN 12/28 — Looking back at the BME Release & Proof of Coverage
Founder of the Render Network, Jules Urbach and Director at the Render Network Foundation, Trevor Harries-Jones, were recently featured guests on the leading DePIN podcast, Proof of Coverage, discussing the latest milestones for the Render Network, including the recent implementation of the Burn Mint Equilibrium on Solana.
With the show’s hosts, Connor Lovely and Sami Kassab, Jules and Trevor explored the Render Network’s growth from 3D art to the emerging fields of decentralized AI/ML GPU compute, spatial media, and mobile 3D.
This post highlights some of the key moments of the Proof of Coverage podcast and ends with a recap of the Render Network’s recent Twitter Space on the first phase of the BME implementation.
The Rise of Decentralized GPU Computing
Jules kicked off the podcast discussing the genesis of the Render Network over a decade ago when he began working on distributed tokenized GPU computing — patenting the idea that became Render back in 2010 — to build infrastructure for next generation interactive and immersive 3D media.
The conversation moved on to explore the evolution of 3D rendering and the rise of new immersive technologies like holographic displays and light fields — looking at how spatial media can transform the media ecosystem. Jules describes the power of new devices like the Apple Vision Pro as providing an immersive overlay in one’s physical space, offering a unique approach to personal computing and interacting with content.
“There’s nothing quite like [the Apple Vision Pro], you have never had this, you’ve never had this perfect position tracking, perfect hand tracking where you could explore worlds and edit them and, you know, collaborate on those things with this kind of hardware.”
Jules predicts that it will take a decade for the Vision Pro to reach mass adoption, similar to the iPhone’s evolution. The device could transform how content is consumed, making traditional experiences like watching movies or collaborating in workspaces more immersive and interactive. He also envisions a future where Light Field displays, which are currently under development, eliminate the need for any headgear or goggles. These displays will integrate into everyday environments like homes, museums, and concerts, providing immersive experiences, and will be aided by the growth of decentralized GPU graphics infrastructure.
The Decentralized GPU Compute Landscape:
After Jules’s discussion of next generation media technologies, the podcast shifted to exploring AI, examining how the Render Network is approaching the field with its GPU network.
Trevor discussed the spectrum of the GPU market that moves from specialized GPUs like NVIDIA’s H200, H100, or A100 used for AI training to more common consumer GPU models — which, while less suited for training, hold tremendous potential for rendering, inference, and fine tuning when aggregated through decentralized networks.
“There’s a significant chunk in the hundreds of millions that can be used for rendering and can be used for inference and some other aspects of AI. And so, first, it’s a pretty broad market out there when you look at the potential available GPUs for at least inference and rendering. So that step one is that it’s barely been tapped.”
Trevor highlighted that while demand in GPUs for AI compute has grown tremendously, 3D rendering still offers the best monetization of GPU cycles for node operators. He laid out how the network’s focus on the motion graphics artist community gives it a unique integrated toolset — enabling it to provide a unique aggregation of supply and demand, different from other GPU offerings.
“We’ve got a demonstrated model of performing, being able to successfully run jobs on those nodes security-wise, making sure they run securely, making sure they’re paid out over time. So, you know, we’re in a pole position just by the fact that we’ve got five years of operation and successfully operating. And I think also because we’ve focused on that motion graphics artist community, a community that really sees the value in the integrated tool set, in leveraging their assets in a way that’s somewhat different from the other GPUs that are out there…the final point that’s probably one of the biggest drivers is today, render jobs are probably the premium job in terms of paying you for your resources on the GPU.”
The Render Network and Generative AI
The conversation progressed to exploring the network’s Compute Client initiative designed to supporting emerging AI GPU compute workloads, discussing progress the network has made partnering with a range of AI compute aggregation networks over the past 6-months.
Trevor outlined the rationale behind Render’s strategy of partnering with third-party clients like Io.Net, Beam, and FedML for AI and machine learning compute workloads. He described the unique advantages of each network in areas ranging from Ray orchestration to scalable serverless AI APIs and finally, federated deep learning — emphasizing how these partnerships enable the network to support a broad range of specialized AI / ML applications beyond what one single provider or network can offer.
Jules follow up with a discussion of the intersection of generative AI, 3D media and blockchain, describing how AI will augment 3D workflows in areas like modeling and texturing — with text-to-3D models and shaders. He also described the importance of Proof of Render, providing on-chain provenance for 3D media that enables artists to authenticate 3D artworks in an age of AI with on-chain hashes that show granular creation metadata.
Jules went on to describe the overlap of infrastructure between 3D rendering, gaming and consumer AI. He noted the power of consumer GPUs to make AI / ML training more widely available, pointing out that HPC GPUs like the A100s or H100s are not always necessary for a wide range of AI tasks. For many applications like Stable Diffusion, RunwayML, and other generative AI workflows, consumer GPUs like the RTX 4090 are sufficient, especially for tasks like fine-tuning and inference.
What’s interesting is that a ton of AI jobs don’t need an A100 or an H100. In fact, if you’re looking at anything like something like stable diffusion, or things like RunwayML, or those kinds of things, even the training datasets aren’t pushing you towards that very high end. So for that, a 4090 is perfectly fine.
The conversation ended with Jules’s discussion of Apple’s M3 chipset and how its rendering optimizations open the potential for mobile 3D creation, and even leveraging iPhone, iPad and Macbook GPUs on Render.
“What Apple has is something pretty remarkable. So the fact that if you want to run a job that requires a lot of memory, the fact that there’s going to be millions of these MacBook Pros and just even Mac Studios that are out there, those are absolutely interesting for us as vendor nodes… certainly for processing jobs that are memory-constrained, those are great.. with the scale of the iPhone”
“And so I do think that most Apple users are going to want to use the Render Network because Apple, the way it’s set up, has one pretty good GPU now on these systems. But if you want to have more rendering power, unlike on a PC, you can’t expand that.. but the Render Network, the reason it works well is if you have 1,000 frames or 2,000 frames under one minute render, 30 frames a second, and there’s 2,000 GPUs out there..that kind of power is amazing. There’s a lot of increased latent power that we’re seeing in the Apple ecosystem, we’ve been very carefully thinking about that strategy.”
Twitter Spaces Celebrating the BME Release
To celebrate the launch of the Burn Mint Equilibrium (BME) Emissions on Solana with the Render community, Head of Communications for the Render Network Foundation Andrew Hyde joined Trevor to walk through the release and discuss future plans for 2024.
Trevor emphasized that this is just the first step in a much larger journey. The upgrade to Solana (announced at Breakpoint 2023) is not only a technical upgrade but also a path to new possibilities for the network now that it is on Solana.
“This is a big milestone, it’s as big as the upgrade to Solana (announced at Breakpoint 2023). It will unlock the direction of where the community can push us. There’s been a lot on how we can leverage putting things on chain from royalties, provenance, and more.”
The Burn Mint Equilibrium (BME) live on Solana provides a more frictionless pricing and usage model, removing a lot of complexity for end users while also enabling the network to support a wider range of applications. With Solana’s high TPS, the network can explore new on-chain capabilities like granular royalties, digital rights management, and provenance tools — Proof of Render — as well as greater levels of automation needed for compute clients which are central to the network’s growth.
Below are key notes on the near term impacts of the BME:
- Impact for Artists: The release of BME Emissions means artists can now process their transactions entirely on-chain, often without even realizing it. This integration enhances the user experience. Artists can continue their work as usual on the network.
- Impact on Node Operators: Node Operators will have to upgrade from Ethereum’s RNDR to Solana’s RENDER. This transition involves updating and adding Solana-compliant wallets into their systems, which is a necessary step to enable smooth payment processing at the conclusion of each epoch.
- Allocation: RNP-001 introduced a graph detailing monthly allocations. 47,000 RENDER tokens are now distributed weekly to Node Operators, averaging about 6,000 per day.
- Emissions Schedule and Operations: The emission schedule under BME is more fixed in comparison to total emissions, but it still reflects the relative percentage of work completed in each epoch. The Network will continue to provide upgrade tools for legacy RNDR through an operational bridge that enables users to convert RNDR to RENDER. Moving forward, participation in the Network will focus on RENDER SOL. Currently, over 20 million RNDR tokens have already been converted to RENDER in this upgrade process.
The launch of the BME on Solana was a huge milestone for the network, but one of just many milestones throughout the year. The next BTN will look back at the highlights of 2023, and look towards 2024.
Join us in the Rendering Revolution at:
Website: https://render.x.io
Twitter: https://twitter.com/rendernetwork
Knowledge Base: https://know.rendernetwork.com/
Discord: https://discord.gg/rendernetwork
Telegram: https://t.me/rendernetwork
Render Network Foundation: https://renderfoundation.com/
