Chainge Technology Salon — Chain / AI Blockchain + Artificial Intelligence
Chainge Technology Salon #3 — Yu Lang Talks About Bytom, Tensority and Blockchain’s AI Edge
AI edge computing and blockchain are a perfect match. That was one of the central messages from Yu Lang at the 11th Technology Salon “Chain” — Chain/ AI. The event was held at Shanghai Jiaoda Huigu Training Centre on 16 September 2018. It was organized by Blockchain EXE × 8BTC together with TomatoWallet, Bytom and Cortex.
“Chainge” is an offline event started by 8 BTC — the most influential independent platform for bitcoin, blockchain, and cryptocurrency news in China — to focus on blockchain’s hotspot discussion and communication between industries and companies. By hosting offline meetups, hackathons, blockchain summits, the organisers hope to help industry experts create high quality content, collaboration and promote the dissemination of knowledge and innovation in the field of blockchain.
Yu Lang has a Master’s degree in computer science from Xidian University. He was a senior system engineer at Alibaba Alipay from 2012 to 2014 and he joined the founding team of 8BTC in February 2014.
During his presentation, he talked about how blockchain gives AI edge computing boundless influence. He used examples and insights from the Chinese company Bytom, where Lang works as the CTO.
What is Bytom?
Bytom is a specialized public chain protocol that targets the assets domain. The purpose of Bytom is to bridge BitAssests (0/1 number) and Atomic Assets (real assets). It targets user organizations that wish to let their own assets to become assets flow. It has around a billion markets around the world.
Architecture of Bytom
- Asset Unique → uniqueness and ownership of digital assets, breaking the boundary between byte and atom.
- UTXO Model → based on the UTXO model, showing flexibility and controllability.
- Smart Contract → unique design based financial assets for smart contract
- AI-Hash POW → the consensus algorithm “Tensority” that has broken the boundary between AI and blockchain.
Tensority: A consensus algorithm based on tensor calculus
- Why POW is needed
- Bytom is a protocol, basic requirement of public chain is “permissionless”, and POW matches the conditions
- Force barrier is the best way to build a public chain ecology
- The protocol can update iterations on itself, separate from single point break down
- Why the need for a new calculation
- Traditional bitcoin’s consensus algorithm required SHA256 results. This calculation has no validity (validity is needed)
- Bitcoin mining machine iteration is fast, leading to a lot of waste in hardware (reuse the hardware)
- The full node community of public chain is an important indicator. There is a need to attract many users to join. convenient operating conditions)
How Tensority breaks down the barrier
Tensor calculus, specially in matrix multiplication , is a core part of deep learning. The architecture of Tensority is built on 256x256x256 matrix multiplication. The final result received will be used to perform final calculations with SHA256. It has been proved in collaboration with Southwest Jiaotong University.
Tensority’s consensus mechanism is suitable for AI hard drive speedup. AI consensus calculation is the business model. Vectordash & Snark AI lend out their GPU calculation to deep learning users. Tensority have its own special chip AI-SIC where TPU and GPU coexist. It is adjustable based on the requirements and can select between two different tasks: Mining BTM and AI acceleration.
Bytom needs to have these two tasks because they think that most of AI calculation equipment will become Bytom’s node. Especially now after the appearance of FaceID, it has attracted many semiconductor manufacturers. ARM predicted that in year 2020 they will sell 100 billion chips that are suitable for all smart equipment. The chip will be used for AI Deep Learning in the future. Edge calculation will be used in everything that humans use. Blockchain will be functioning in every Bytom node.
As the number of AI chips grows, the number of Bytom users will grow and the higher Bytom’s calculation power will become.
The open source for Bytom code has already been published. Matpool is the mining pool for Bytom and it is now online to realize the first step of cloud AI.