By the time block height reaches 1,420,800 (estimated to happen at 4PM on October 25th (GMT+8)), Matrix AI Network is about to receive a major update.

The update will bring to the Mainnet AI mining and a new block creation method, alongside changes to the mining process, the election algorithm, the reward distribution method and a penalty policy for inactivating miners. The previous method for storage will also get optimized with this new update.

**Mining Process**

Block creation previously occurred with POS and POW running in serial execution. This lead to periods of time when miners were latent and compute was underutilized. This update introduces AI mining while increasing the productivity and utilization of compute.

During each broadcast period the 99 blocks other than the broadcast block will be divided into sets of 33, with each set representing 3 consecutive blocks. In each set the first block will be the AI block, the second will be a POS block, and the third will be a POW block.

A miner begins POW mining once the information from the mining request is received. POW mining is comprised of X11 compute and SM3 compute. X11 and SM3 mining are executed in parallel. X11 algorithm involves the following Hash algorithms: blake, bmw, groest, skein, jhash, keccak, luffa, cubed, shavite, simd and echo.

If the results of the X11 compute process satisfy the minimum difficulty requirements, the compute results can be sent to the validator and can continue to do POW mining.

**Miner Nodes Election Algorithm and Difficulty Adjustment Algorithm**

**Algorithm**In order to increase the probability of a miner being elected as Masternode in a given time frame without altering the overall probability of getting elected, Matrix AI Network introduces a round-robin algorithm for this update. The fundamental rationale is getting a snapshot of staked nodes during each election cycle. The mining masternodes for the next election cycle will be chosen from this random snapshot, then the elected nodes will be removed from the snapshot until there are no longer qualified nodes in the snapshot and the cycle is completed.

**The mechanism will work according to the following principles:**1. Regardless of the number of staked nodes, 32 is the minimum number of masternodes.

2. After reaching 1024 staked nodes, for every additional 64 staked nodes, there will be an additional 2 mining masternodes.

**Difficulty adjustment**The anticipated mining compute will change quickly, due to the unique Matrix characteristic masternodes participating in POW in mining network election cycles. The newly adjusted difficulty algorithm is suited for the following dramatic changes in compute:

1. Difficulty for each election cycle will have a base difficulty

2. The first N POW blocks in rapid succession will increase the difficulty

**Mining time adjustment**Due to changes in the mining process and the election algorithm, the time it takes to generate a block is expected to be different as well. Therefore, mining time is adjusted accordingly in the updated Mainnet.

1. Currently each POW mining process lasts for the creation of 3 blocks, the mainnet hopes to achieve a blocktime of 12 seconds, creating 3 blocks in 36 seconds. Due to additional time delays for the validation and sending blocks, there is an additional 1 second, so the actual POW mining time is 33 seconds.

2. In order to prevent faults in powerful nodes, and cause problems for the data chain, if elected nodes time out during mining, mining results will be taken from “mining nodes in the Matrix foundation” (the nodes from the foundation do not participate in the distribution of rewards). If a miner times out on the previous POW block, the difficulty algorithm will be adjusted downward. With this update the mining time out window will be set at 100 seconds.

3. When difficulty allows miners to create blocks quickly in close succession even outpacing the POS process, the difficult index will increase. According to the situation of the mainnet, the POS process takes approximately 46 seconds, the configuration for the fastest blocktime will be 6 seconds. When the POW mining time is lower than 3 times the lowest blocktime, or 18 seconds, the difficult will quickly increase. If POW mining time is higher than 3 times the lowest blocktime, the system will use an iterate process to linearly adjust the difficulty level in relation to blocktime.

**NOTE**: To prevent elected nodes from later dropping out and impacting the compute pool, nodes that have dropped out or have compute penalties will be automatically filtered out.

**New Block Reward Mechanism**

Changes in block reward mechanism will affect miner nodes, validator nodes, transaction fee and interest.

**Miner node rewards**An individual miner participation reward is equal to the total miner participation count / miner masternode count.

A fixed 10% of block rewards will be distributed for AI rewards every third block. At the current rate 1.44 MAN (compute method: 10% of the individual block mining rewards times 3).

The change in mining reward distribution and distribution cycle is as follows:

**Validator node rewards**The distribution for validator fixed block participation rewards will not change. A difference in the update is the distribution rate is that formerly the participation rewards were issued once per block, and this time participation rewards accumulate in the state tree and are distributed at the beginning of the next election cycle.

**Transaction fee**There are changes affecting how participation reward is calculated. See below for details.

**Interest rewards**The current mainnet interest rewards are calculated during each block and saved into the state tree. In order to optimize storage, in the update interest rewards will be calculated every 100 blocks. The interest algorithm will be unchanged.

**Compute Penalty**

At present a number of miners elected on the mainnet, have only undergone staking, but have not activated their nodes and were still receiving mining rewards without supporting the development of Matrix mainnet. The update will introduce a penalty policy.

**Compute blacklist**Once a node is elected to be a mining node, for each POW cycle they will need to simultaneously compute a minimum low difficulty POW result, which is reported to the validator. The validator will publish the minimum compute results to the blockhead. Before miners are elected in each election cycle, there will be an assessment of the minimum compute requirement results, and node that didn’t qualify will be added to a blacklist.

Implementing penalties:

Blacklisted nodes will

1. be denied mining rewards from the current cycle.

2. lose eligibility to participate in the subsequent mining cycle

**Storage Optimization**

The present the storage space for the data chain is fairly large, giving nodes supporting operation of the data chain a considerable storage burden. After optimization, blocks will be reduced to about 10% of the previous amount.

Optimization methods

1. Increased storage capacity for multi-currency, and only storing changes in the balance, and reporting the prior value if there is no change.

2. Mining participation and rewards will only be recorded once per cycle.

3. Validator participation and rewards will only be recorded once per cycle.

4. Staking information will go from being calculated and stored once a block to once per broadcast cycle.

Besides the changes listed above, there are other improvements and optimization, details about which can be found by downloading *Matrix Mainnet Update Notice* from our official website.

**Update the gman File**

Download the required files from our official website before trying to update the gman file.

If you haven’t deployed your Masternode, please download *Matrix Masternode Deployment Guide and Files* from our official website. If you need to update your Masternode, please download *Matrix Masternode Update Guide and Files*.

**NOTE**1) You will find guides for both windows and Mac users after you decompress the downloaded file, use the correct one for reference.

2) In order for your Miner or Validator nodes to work smoothly, you need to keep them up-to-date. Please go to our official website and download the latest version of the gman.zip file.

**Deadline**By the time when block height reaches 1,420,800, Matrix AI Network will no longer support the previous version of gman. If you choose to update before this time, the normal function of your node will not be affected.

**Node rewards**If you fail to update your gman file before block height reaches 1,420,800, you may suffer losses due to your node being rejected from Matrix blockchain network.

To help the community know more about this update, our team has planned a live stream this week. If you’ve miss the live stream, please check our official YouTube channel for the recorded video.

Should you have any problem in the updating process, please send an enquiry to public@matrix.io.

Thank you for your support!

Matrix AI Network leverages the latest AI technology to deliver on the promise of blockchain.

FOLLOW MATRIX:**Website** | **Github** | **Twitter** | **Reddit** | **Facebook** | **Youtube** | **Discord****Telegram** (Official) | **Telegram** (Masternodes)**Owen Tao** (CEO) | **Steve Deng** (Chief AI Scientist)