Matrix AI Network: Masternode descriptives, selection and ROI.
OUTDATED REWARD MODEL: PLEASE SEE https://medium.com/@Price_Steven/matrix-ai-network-masternode-rewards-and-roi-eb6ef32c06fa — WILL COMPILE BOTH SOON
The following article is a compendium of sorts for the community to refer to regarding every masternode-related question. The goal of this article is purely educational; therefore I will limit myself strictly to sources from the MATRIX team. I will first explain which kind of masternodes there are, the requirements for holding a masternode, which rewards are in place and finally hardware requirements. If I missed anything, please let me know.
Author’s note: this article is a work-in-progress and will be updated each time new information becomes available.
What is the node structure of MATRIX?
First we need to clarify something about the node structure in place for MATRIX as some terms are used interchangeably within both the MATRIX whitepaper and github, this may lead to confusion. This article aims to end that confusion. Without going into too much detail there is a clear distinction between a masternode and a supernode. The community seems to refer to the supernodes as masternodes. It is important to understand that both have separate rewards systems in place. Supernodes are selected from all available masternodes (also referred to as delegates) and will enjoy better rewards.
Schematically this looks as follows:
There are currently two types of masternodes one could aim for: a mining masternode (PoW) and a validator masternode (PoS).
Other nodes exist as well and will be briefly explained in the article below. From an investment point of view however one needs to mainly worry themselves about the masternodes.
A mining masternode requires a deposit of 10k MAN and requires your hardware to have the processing power to run 1,000 transactions per second (TPS). Miners will earn a base reward in addition to their mining reward. These nodes are the agents for transaction processing.
A validator masternode requires a deposit of 100k MAN and requires your hardware to have the processing power to run 2,000 TPS. Validators essentially audit whether the delegate masternodes are working honestly.
Note that the MATRIX network also runs two other types of nodes called AI-service nodes and Agent nodes. Both require no deposit and simply serve to provide power output and the distribution of processing power respectively. Finally, there is something called a Dark Guardian node — which is deployed by the MATRIX foundation itself. The role of Dark Guardian is a voluntary contribution, however upon reporting fraudulent activity the Dark Guardian can be rewarded by the foundation. The related codes of this Dark Guardian node will be published later.
Update (22–8–2018): When a third party applies for the computing power of the MATRIX blockchain network then the proceeds of the applicable fee go to the nodes that provide the computing power. In this case that are the agent nodes. It thus appears you don’t necessarily need to own a masternode to receive proceeds and your contribution to the network is thus incentivized in other ways as well.
Rewards for holding a masternode
The main reward for holding a masternode is being eligible for supernode selection, this requires a sufficient deposit of MAN and the appropiate setup to handle the amount of TPS required for being a miner/validator.
Only supernodes can act as miners/validators!
As an incentive, masternodes that are eligible for supernode selection shall enjoy a 3% annual yield — as far as i understand this reward is thus only distributed when your hardware is sufficient for supernode selection. Also, when the computing power of selected supernodes are insufficient they will be supported by lower-tier masternodes in the same group. The underlying nodes providing help will then share a corresponding reward.
- Miner Masternodes that aren’t selected as Miner Supernodes can still earn rewards by providing computing power to external parties. It thus acts in the same way as an agent node, rewards will be based on the fee paid by a third party for providing the necessary computing power. Validators receive a reward in the form of GAS in this type of transaction.
Ok, so you managed to get 10k MAN/100k MAN respectively and have the hardware to support the required tps. Congratulations, you are eligible to be selected as a supernode. Now, how does this selection take place, what are the odds of being selected and which rewards can i enjoy then?
Supernode selection process
To ensure the network remains decentralized, supernodes will be randomly selected from all available masternodes in every given period, as of now this is chosen to be approximately 1 hour. Random selection of these supernodes ensures that the probability of being chosen as a delegate supernode remains sufficiently low, thereby avoiding centralization of the network. A supernode is either randomly selected, or selected based on a combination of: computing power, communication bandwidth, network connectivity and other metrics.
The amount of MAN staked and uptime seems to be most important in the selection process (Q&A Matrix — july 13, 2018). This seems to hint that the more MAN you hold, the higher your odds (unconfirmed).
Supernodes consistently exist of newly selected delegated masternodes via a randomized selection process. No central entity can thus control the network. A total of 32 supernodes are selected for mining/validating each selection sycle. 21 supernodes are selected for mining, 11 are selected for validating.
This design provides a steady and standard time interval for leasing computing power. It thus ensures consistency of processing transactions across the entire blockchain. Considering potential instabilities that could arise during the handover between validators and miners, there will be a shift of 16 blocks between the selection cycle and validation cycle.
What are the odds of selection?
For the mining masternodes MATRIX has calculated the odds of being selected as a delegate supernode. Statistically speaking, any given miner masternode is likely to be selected as a supernode once every two weeks, even if there are 10,000 masternodes.
For validator nodes the odds of selection have not been calculated or shared yet. What MATRIX did share is the total number of validator nodes. The maximum number of validator supernodes per round is 11. However, there are up to three layers of supporting validator nodes that are available: 22 in the first layer, 44 for the second layer and 88 for the third layer. This means that there is a total of 165 available validator supernodes. I assume the first 11 validators nodes enjoy the most rewards, while the rest act as a supporting layer should the network need it so.
If you want to know the exact algorithm for selection please see the link here.
What is the distribution model and total supply of tokens reserved for miners and validators?
Before I mention the actual rewards in place for having a supernode it helps to know the total amount of tokens reserved for miners and validators alike. Rewards depend on the following distribution model:
40% of all the available tokens (1 billion) are reserved for mining and supporting operations. This means 400 million tokens. These 400 million tokens will be paid out in the following percentages:
· +- 25% will be distributed to miners (total is 100 million tokens).
· +- 20% will be distributed to validators (total is 80 million tokens).
· +- 50% will be distributed to suppliers of computing power for AI calculations using the MCMC algorithm (total is 200 million tokens).
· The remaining 5% is reserved for the MATRIX community and distributed according to community votes (total is 20 million tokens).
Note that this is the distribution model as of now, MATRIX will check consensus among the community via survey. A distribution proposal will be shared every 6 months, if consensus fails the above mentioned distribution model will remain valid.
What are the actual supernode rewards?
Rewards for miners work in the following way:
- The available total payout is equal to the total number of [mined] tokens x 40% x 25% (= 10% of the mined tokens).
- 1/8th of this available payout is equally distributed between the miner supernodes.
- The remaining 7/8ths are distributed between the selected miner Supernodes based on a Proof of Work algorithm.
This algorithm ranks the setup you host your supernode on based on the available computing power. Your setup will then compete with others and will be ranked across:
1) Service times
3) And TPS.
Holding more AI and CPU computing power naturally then means higher rewards. The actual reward will probably be low in the beginning as there may not be a high volume of tasks at mainnet launch yet. MATRIX has introduced a class of computing tasks that provide a public service at first, these will have no rewards but honor value. Once real tasks are assigned and complete, the system will automatically assign a considerable number of rewards.
- Miner Masternodes that aren’t selected as Miner Supernodes can still earn rewards by providing computing power to external parties.
Rewards for validators work in the following way:
- As you all know by now, a total of 165 validator masternodes are automatically selected based on total deposits.
- The total payout available to validator nodes equals the total number of [mined] tokens x 40% x 20% (= 8 % of the mined tokens).
- 1/8th of the total available payout is distributed to these 165 validator masternodes.
- The remaining 7/8ths is distributed equally among the 11 selected Validator Supernodes.
As of now only the rewards model is known. Any information pertaining to actual mining rewards is something that is currently unknown and will have to be established by the team near mainnet launch. Calculating ROI is therefore not possible at the time of writing.
Summarized, rewards for the miner supernodes and validator supernodes are not the same, naturally validators seem to earn more as they need to share less of the allocated rewards with other node holders — the distribution is also not dependent on hardware setup in this case.
While the testnet launch is in September 2018, note that rewards will not be distributed then. This period will only be used — as the name implies — for testing. Rewards are distributed when the platform launches at the full mainnet release in december.
Will I be able to run a masternode?
As previously noted, your setup must be able to provide 1,000 or 2,000 TPS throughput respectively. Besides that, your setup must support Remote Procedure Call (RPC) — for those that don’t know this is basically a protocol that allows to run a piece of code in another computer. No big deal, the actual setup of a masternode is beyond the scope of this article for now. Just know that this is not something you need to worry about now.
To achieve 1KTPS, MATRIX used the DELLR730 server setup. Now most of you might still wonder: “will I be able to run a masternode on my setup at home? “
The following configuration was deemed sufficient to meet entry-level mining requirements, with the exception of its single quad-core processor — which might not be sufficient to reach the desired TPS according to MATRIX. Instead they suggest an alternative core.
In the mining setup of MATRIX, components they used include:
An Intel® Xeon® processor E5–2600 v4 supporting up to 22 Cores.
RAM: 64G, DDR4, up to 2400MT/s
Storage Controller: SSD (1x)
Hard Drive: 1 x 1TB
Form Factor: 2U Rack
GPU accelerator: 2
Matrix dedicated their cores as follows:
1 core for network processing;
1 core for the remote procedure calls;
1 core for encryption, decryption and validation;
1 core for transaction processing and moving memory.
All other cores were used for PoW computing. Having a quadcore processor therefore seems like the bare minimum. An 8-core setup might be better in this case.
Please note that MATRIX is also working on their own AI hardware mining tech. It might be possible that this is more efficient and powerful than traditional tech — and thus generate better rewards based on the distribution algo. Stay tuned for more info on that!
Masternode Frequently Asked Questions
- Does holding more MAN increase the odds of selection?
The Q&A in this link seems to suggest that the amount of MAN staked does play a role in supernode selection, to what degree however remains unclear.
- Does holding MAN on more devices increase the odds of selection? (e.g. 20k MAN total with 10K MAN on each device).
No, the selection algo of MATRIX does not seem to take this into account.
- Can multiple separate devices be used to mine the same node?
Matrix does not support multi-device mining on a single node.
- What happens when the supply reserved for supernodes has been distributed? How will supernode holders be rewarded then?
When the total rewards for miners are exhausted, transaction fees and commissions paid for computing services will be used as the main rewards for mining. All masternodes will be eligible to undertake CPU computing tasks or AI computing tasks to receive rewards. These rewards will be assigned through a task distribution network, validation network or external agents.
This article is written mainly as a community service and will be the first of many related to the Matrix AI Network project, to ensure quality content I would very much welcome every form of feedback. Please feel free to drop a comment below. If there are any questions I will try to answer these as best I can. Also feel free to contact me if there are any information gaps I could help close in the future, maybe I will write my next article about that. Thanks in advance!
- Steven, Dutch Community Manager Matrix AI Network
Edit: Small fixes and clarified some things regarding the amount of masternodes and random selection.
Edit 30–06–2018: included specs sheet and changed duration of masternode selection.
Edit 01–07–2018: included info on annual staking rewards.
Edit 08–07–2018: updated per Q&A of 06–07–2018
Edit 22–07–2018: Restructed and rewrote part of the article to clear up confusion between masternodes and supernodes. Added FAQ section and more info selection and rewards system.