@EMCprotocol I really want your project to succeed.
Take your time and have my love. 
1 Like
Have no idea about the relationship between #EMC and #ICP
Set Reminder for our AMA
Learn deeper:
Innovative technology Interpretation
Developer ecology
Ecological Development
Roadmap implementation
https://twitter.com/i/spaces/1PlKQpmvLlzxE
“We will hold another AMA on Thursday next week on the topic of SNS. Community members are welcome to participate as guests and interact with the EMC team during the AMA.”
8 Likes
Pinging frequent commenters in this thread as a FYI about first of two scheduled technical AMA twitter spaces on EMC & ICP (see above)
First one starting in 7 hours from now.
@ZackDS @Icdev2dev @MCLYang @cyberowl @jwarner
Sorry if I missed anyone. Just trying to keep information flowing by making sure you are aware in time.
3 Likes
Sorry , twitter space crashed , new link is : https://twitter.com/i/spaces/1MnxnpQVZoeGO
So much to catch up on, but has EMC applied for a grant to develop and substantiate a proof of concept that is accessible on the IC right now? If so, can the link to the canister be given?
Otherwise, independent of whether this project is a possible scam (clearly it is not), it’s not doing one of the most important things and setting a good precedent for the IC: a successful SNS sale without a usable dapp hosted on the IC no matter how compelling the whitepaper is a green light to any number of those who would take advantage. That’s not even a reason as much as one of the main points of the grant program afaict is providing the funding to do exactly the necessary work before such an endeavor and enterprise would then do an SNS sale.
My real point is that the IC is such a different way of understanding the architecture of decentralization, a project/dapp looking to SNS should really be of apodictic substance before SNSing: a white paper is not enough, in fact that period of open usability prior (of a live on chain (if proof of concept) dapp) to an SNS sale going live is a form of decentralized community action that’s at the heart what makes the IC different. Particularly if the project already has an established base, pushing through an SNS sale without having a user experience showing all those developers in its base what’s (at least) possible on the IC is likely causing the IC to miss out on significant organic growth (“establishing familiarity”) opportunity.
Not to flush your buffer or anything… it’s certainly possible for that kind of thing to happen after the SNS sale, but so far the SNS is setup for single use.
1 Like
Clear Opportunity
The idle computational power of crypto miners is an untapped resource that can be harnessed to create a robust, cost-effective, and environmentally friendly marketplace. This business model has the potential to disrupt the computational power market by providing more accessible and affordable resources to users, while also increasing the profitability of crypto mining. It’s a win-win scenario for miners, users, and the environment.
Reward > Risk
I would like to seize this moment to extend my heartfelt thanks and admiration to the team for choosing to open source the code for this project. This decision not only marks a substantial accomplishment for the project, but it also demonstrates a deep-seated dedication to transparency, collaboration, and shared innovation, which are fundamental tenets of the open-source community. This move significantly diminishes associated risks and bolsters trust in the team spearheading this project’s development. While there’s still a need to examine the code and comprehend the architecture, we now have the ability to view and understand the work at its source.
Competition
Golem Network is excited to introduce a GPU computing proof of concept (PoC) using Stable Diffusion as a use case. The PoC, which runs on the Golem Network platform, was inspired by the project developed by the community member maugnorbert and brings a new application for Golem Network: access to decentralized GPU computing power.
The PoC is a text-to-image generator inspired by the AI-based Stable Diffusion model. Users can type keywords on this tool to automatically generate images. With this PoC, Golem Network has integrated the Stable Diffusion text-to-image model with the decentralized Golem Network platform as a backend, showcasing its GPU computing possibilities.
ICP Marriage
The project is making a strategic move by building elements beyond the core protocol atop the Internet Computer (ICP). The EMC team will retain control over source changes to EMC as they possess the necessary keys. However, transactions and funds will be contained within canisters, with the SNS managing the marketplace. The critical question is whether the team plans to exclusively build on ICP. If they opt to incorporate other chains, what will be the implications for shareholders? How will their value be either increased or diluted? It’s crucial for shareholders to have confidence that their investment will appreciate over time.
Moreover, it’s essential to understand how the community can contribute to the growth of the ecosystem. Will there be opportunities to earn rewards for building user interfaces and tooling? The community’s involvement can significantly enhance the overall development and success of the project. Clarifying these points will strengthen the trust of shareholders and the wider community in the project’s growth trajectory and strategic decisions.
4 Likes
I was a silent listener in today’s AMA. In any DD, while nothing is certain, one gradually gets the feel of the project and people driving it. This then gets followed up through architecture, design and implementation that is shared.
Nothing that follows is legal or financial advice. It is for education and entertainment purpose only. DYOR.
Overall my impression of the AMA was quite positive. Thanks @alxgoh @zed.emc for presenting and answering questions.
I think that persuasive case was made on the business model behind EMC. It is clear that @alxgoh & @zed.emc were quite prepared with the answers from the business perspective.
From a technical perspective, one concern was essentially(i am paraphrasing): “how wedded is EMC to IC”? The answer here could have been stronger. Nevertheless the case was made that as a L2 , EMC would rollup to IC. Knowing the architecture of IC as i do, I do believe that once a commitment is made from a L2 perspective to an L1(IC) in this scenario, it would be rather difficult to switch to another L1. However the proof of the pudding will be in the analysis of the integrating code.
The reason that i found the answer to the above question not as strong was because a pivot was made to reference conversation with Dom. I believe that it would have been stronger, imo, without obvious name dropping. I believe also that the host should have translated the questions to Chinese to enable Zed to answer (a hint for the next AMA).
I look forward to the next AMA.
9 Likes
@emc, @ninadou , @alxgoh , @zed.emc
Thanks for hosting the second AMA. While I was invited to speak and ask questions at this AMA, due to a unscheduled conflict, I could not make it. I apologize.
Nothing that follows is financial or legal advice. It is meant for education and entertainment purpose only. DYOR.
Thanks @icarus @zire for participating in this AMA as interested parties outside of EMC. I believe that @zire spoke on behalf of Dfinity in this AMA. @zire please correct the above if not true.
These are my impressions of the recording that i reviewed.
I was pleased to learn that @zed.emc mentioned that he was still learning about the IC from a potential perspective for EMC. “Knowledge is proud that she knows so much, Wisdom is humble that she knows no more”.
In his session, @zed.emc covered that EMC could potentially become a distributed hugging-face with the ability to run models in the same user experience as model training. He thought it possible to have this developed in a couple of months. (I think that this is way too aggressive, but i will come back to it).
He , then, discussed an real world workflow of game development on EMC that could significantly reduce the art-development and design time, money and speed aspect of it. This is a useful use-case because it links EMC & IC to a use case outside of the “crypto” eco chamber. My impression is that how it would be really operationalized(think Unity, Blender) is something that would need a documented workflow.
Additionally he announced an Hackathon in late July; which would be awesome for AI developers. This would be more a carnival style hackathon. My impression is that this is a great idea; however significant preparation time is needed to be devoted before now and the hackathon; so that the platform is ready for the onslaught as well as enough documentation is created and translated for the developers to be effective. Having it in a Carnival style is brilliant idea!
Somewhere it was discussed that the stable diffusion model as a POC was quite successful with early adopters. While UX can be improved greatly, i think that this was good progress.
@alxgoh gave a rundown of the business aspects of EMC; bringing down the cost of GPU compute to 10-20% of AWS cost. If these numbers can be achieved, it would be a game changer; as i have mentioned before. He also acknowledged that a lot more GPU compute power needs to be brought on; for which the team is negotiating with large former ETH miners.
@zire mentioned that he was impressed with the street creds of the EMC team. He also mentioned that EMC could be one of the “unicorns” in the crypto space through a real life use case (I imagine something like game development).
@icarus came with the perspective of hosting EMC & IC nodes. Hia take on the models residing on IC resembles my philosophy of how deeper ties into IC as a layer 2 can be made. That said i think that there are significant technical and SLA related challenges to achieve this in context of IC.
Overall the meeting was quite positive from my perspective. I felt that team has become more confident in dealing with a diverse eco-system as IC. They have their vision and they are making progress on the foundational aspects.
6 Likes
Yes. EMC plan for SNS launched before end of June
3 Likes
In EMC plan for token distrubution (to be finalize soon), 50% (will remained unchage to final version) of token distribution will be the treasury holding under SNS allocation for reward contribution to EMC network and ecosystem. Generally split under 4 categories: 1. Reward for computing and nodes provider (POW + POA + POC). 2. Reward for ecosystem developers cobuilder. 3. Reward for computing market place trading (including RWA NFT trading). 4. SNS voting reward
1 Like
While EMC core developers team is focusing on the protocol optimzation and core of the computing network, we have a new Technical Partner (CTO) joining EMC in Singapore leading the architectural design of decentralize hugging face, with significant part of SNS fund-raise to be focusing on building the globalize developer and engineering team as well as global contributors network. EMC technical partner has exceptional background and experiences as CTO/co-founder of reputable AI start-up (Asia version of DI-D). Despite the aggresive plan (in term of timeline) to launch EMC version of Huggingface in order to support the coming AI heckaton, we are calling for global contributor to join EMC and co-build this platform together. @Icdev2dev we certainly looking for developer like yourself to join us. Regarding the Marriage of ICP and EMC, as disclosed in AMA, we are already working with various IC node provider in Asia to launch large GPU claster (>100xGPU) at Genesis Mainnet launch to provide stable and powerful GPU services to power the decentralize Huggingface. Icarus is very experience in operating hosting services in Oceania, and currently working very closely with me math modelling of cost/efficiency/performance analysis of EMC computing claster, our priority is still optimizing cost by recycling the power GPU from miners.
2 Likes
Due to the influx of a large number of edge nodes, the block generation speed has increased, causing inefficiencies in routing nodes and validation nodes, and network latency. 
#EMC will be continuously updating the version in the near future and developing lightweight nodes to enhance operational speed. 
Please be patient and stay tune.
1 Like
Has EMC published a demo video? Or can they do so on YouTube/present on the DFINITY Global R&D?
2 Likes
Hate to say it and bring the whole Army down on me, but EMC is not IC material, YET or STILL ! As much as I love hype and see ICP moon am more concerned for the long term vision, and getting hundreds of thousands of developers and miners from Asia/China join the IC that would seem to pump the price, is not the answer. With all do respect to the official members of Dfinity who push EMC please prove me wrong. Show me code that works on the IC to make a L2 chain be able to sell computing power for AI and settle it on IC and I’ll bring mining farms from FLUX, ERGO and others that are not running on RX-470’s and single core celeron CPU’s (*exaggerated to make a point) and 4 GB of ram on a 2 Mbps connection.
2 Likes
@ZackDS DYOR, just for your reference, a standard domain server for FIL (supermicro 4124 board, AMD EYPD 7542x2, 64Gx 32, 1T M2, 3.84T U2, RTX 3090x8, 25G Ethernet, 2000W x4), this hyperscale cluster is capable of replacing AWS P3/V100 instance for training of AI model (hosted in IDC). Refurbished of piling GPU cards abandoned from Eth miner doesn’t means a plug and play, rather require rebuild of server capable to provide optimum GPU compute. All the developmen of EMC code will be open-source as and when our R&D team completed all the require validation in current testnet environment. In current architect design, any transaction that require verifiability will be running on Canister (eg. economic, marketplace), EMC will publish all the detail of Canisters that control EMC network before SNS proposal. Stay tune.
1 Like
we are working Dfinity team to get on Global R&D June (to be confirmed), we will also release the demo video over the weekend.
2 Likes
This is great news! As i understand it, it means that anytime a transaction takes place (the unit of work could be different for training versus inference), that transaction would be recorded on IC.
For example, i currently use the beta version of leonardo.ai. In this , they give me 150 tokens daily for free to use to generate images based on my prompts. This image generation makes use of some GPU somewhere (such as AWS, GCP, AZURE).They record my usage.
As I understand it, another AI company could use EMC on IC and a user (like me) could use that AI company to generate similar images. The advantage to that AI company is that they get equivalent GPU power at 10-20% of the current cost to leonardo.ai. Did i get that right?
2 Likes
You are right! The current demo that we have shared with community is like another version of Leonardoai, the SD model was trained with our EMC 8GPU cluster in T3 IDC, trained model (inferences) is deployed and run on decentralized GPU node on EMC testnet, and the UI is currently run on Canister.
We did comparison of cost of refurbished 8x3090 4U vs. on demand pricing of AWS P3 v100 instance @$2.2k/month, operational cost is estimated around 5-10%, the difference is EMC operate thru Computing Marketplace and price will be determined by supply of demand (trade as RWA NFT with ICP, cycles, EMC)
2 Likes
I’m truly amazed to witness the development of such an ambitious project on the IC platform. The potential it holds to revolutionize the integration of blockchain and AI is truly remarkable. This endeavor has the capability to be a true game changer in the industry.
I wanted to extend my best wishes to the team behind this project.
3 Likes
As of 6.11, EMC has completed the upgrade of all light nodes on the network. The new edge nodes no longer need to synchronize block heights, reducing the burden on the entire network by more than 90%. Currently, there are over 300+ edge nodes and POC is functioning normally. We will continue to monitor the stability of the network. In addition, some Ethereum miners are retrofitting their previous mining machines (upgrading hardware other than GPUs) to establish a cluster of around 100 GPUs. These design plans may become reference models for EMC’s future computing power pool.
6 Likes