No announcement yet.

Why is Credits faster than other blockchains?

  • Filter
  • Time
  • Show
Clear All
new posts

  • Why is Credits faster than other blockchains?

    Why can the Credits blockchain process hundreds and thousands of times more transactions per second than other popular systems? In order to answer this question, a few critical aspects are well worth consideration.

    1. Platform optimization for transaction processing speed

    The source code of the platform is optimized for a very specific task, i.e. high data transfer and processing speed. This is primarily achieved through direct device processor operations. The operational efficiency of hardware is ensured by low-level programming languages, time-critical or memory-critical program sections are written in assembler language. They subsequently take the shape of subprograms and are combined with the code in a high-level language.

    Speed is raised through the optimization of a computing algorithm and more sound core memory access, data redistribution, spreadsheet calculations of functions.

    2. Reduction of data packet size

    Any transaction has a dynamically varying size depending on balance, address size, transaction currency, the use of digital signature, and further comments. 90% represent the minimum transaction sizes (120–150 byte), thus increasing transfer and processing speeds. If a transaction or a pool of transactions has large data volumes, a transmission packet is broken down into smaller (65,535 byte) datagrams used by the UDP protocol subject to the upper limit size.

    This makes it possible to reduce data leaks and increase intra P2P network data transfer speed.

    3. Distributed transaction pooling

    A new round is created every 5 milliseconds (5×10–3seconds). One transaction pool per round is generated. All nodes in the network are listed. Each node is assigned a random number (ranking). The list is sorted downwards based on these values. The first node in this list becomes the main node in this round. All subsequent nodes in the list become trusted nodes to the number determined using the following algorithm: if the total number of nodes is ten, 50% become trusted, yet at least three. If the total number of nodes is at least 100, then 10% are trusted. The average quantity varies between 10% and 50% of the total number of nodes in the network. The final list of nodes involved in any round makes it possible to determine the number of this pool. The nodes involved in pool processing do not participate in building the new round. The system starts a new round using the number based on the algorithm described above without waiting until the processing of the last one is over. This is how asynchronous processing of transaction pools unfolds and the chain of pools is built.

    4. Employment of dynamically optimized data flow in a distributed network

    In the stochastic control theory, some results in the systems with hidden Markov processes allow them to be employed in the control problems in respect of data transfer systems. The proposed approach is based on the following preconditions:

    – channel status is described by a Markov chain with the finite set of states and known transition intensities;

    – data transfer speed is a control parameter and packet loss intensity is a known monotonic function of transfer speed and channel status;

    – the goal of control is to select the law of variation of data transfer speed which makes it possible to achieve the maximum mean value of successfully transmitted packets.

    The problem-solving method relies on building a family of sufficient statistics for the current channel status and selecting a transfer speed value which ensures the maximum conditional real transfer speed (conditional difference between data transfer speed and loss flow intensity).

    5. Employment of findings from stochastic simulation of similar systems

    A group of events is modeled in the test medium, namely: into the distributed information system with multithreaded data processing option. This allows system bottlenecks to be detected and remedial action to be taken.

    Stochastic simulation is the simulation of random events. Multiple repetitions of simulation exercises with the aim of obtaining statistics concerning system properties, obtaining data about the features of random events and values.

    The goal of stochastic simulation is to obtain an estimated mathematical expectation, dispersion, and random values distribution law for the parameters of objects.

    “Random event” means any fact which may or may not occur as a result of an exercise. The types of random events are as follows:
    • Certain (event occurring in every exercise).
    • Impossible (event which cannot occur as a result of an exercise).


    The “frequency of occurrence” means the likelihood of any given event assuming an unlimited number of exercises.
    • “Mathematical expectation” means the number around which the values of random variable are concentrated.
    • The variance of random variable characterizes the measure of spread of random variable around its mathematical expectation.

    Probability density functions represent a function type determined by random values distribution law.

    6. Optimization of the transaction pool access and storage mechanism

    Data access and storage. Currently, data storage in the chain of pools is the most secure option. However, the verification of a transaction necessitates verifying the entire chain from the first block onwards.

    Data storage on the platform is implemented with employment of LevelDB, a high-performance, replicable NoSQL system for key-value data storage developed by Google. The LevelDB warehouse is written in С++ and connected to applications as a shared library, which makes it possible to store ordered data sets in which arbitrary binary keys are associated with arbitrary binary values.

    Data is stored in the form of transaction pools. The use of key-value format allows quick access to all necessary data with no need for checking the whole warehouse.

    The conclusion

    Summing it up, we can draw the following conclusions. Raising transaction processing speeds is high on the agenda for the blockchain technology development. For the time being, there are not so many systems which are capable of processing more than 1,500 transactions per second. In this context, the platform with a throughput of 400,000 transactions per second (in Alpha-version) appears to be a really solid player in the blockchain market, which is in a position to compete even with popular payment systems offering extremely high speeds.

  • #2
    First of all i want to say that your team is doing great job. Nice idea to create Credits forum. I really hope Credits will be the fastest blockchain in the world and we first investors have to be rewarded for our support and faith


    • #3
      Originally posted by Daniel_Mastracci View Post
      First of all i want to say that your team is doing great job. Nice idea to create Credits forum. I really hope Credits will be the fastest blockchain in the world and we first investors have to be rewarded for our support and faith
      Thank for your feedback, Daniel! We'rre glad to hear this!


      • #4
        Why is it so important to have 1 million tps? Visa processes transactions with 24k tps and this is enough..


        • MDiffenbakh
          MDiffenbakh commented
          Editing a comment
          We are focusing on current issues that crypto space have like low network capacity, long latency, transaction cost, that's why we are going to provide solutions for that which will be in demand for various sectors of business, economy, so we are here to create trends, not follow them, this is the essence of innovation and competitive market, who will offer the best solution, that will be on top