Oracles

June 03, 2022


Read more

Worse than 2008?

May 27, 2022


Read more

Credit

May 23, 2022


Read more

Predictable Policy

May 13, 2022


Read more

digital
daily

digital daily: One by One, Two by Two

One by One, Two by Two: A decentralized network is robust because of redundancy. Many computers performing the same calculations, all cross-checking results. Complex blockchain transactions often depend on getting the order of operations right, so to prevent errors, most blockchains execute all transactions serially. One by one is safe. However, most of the transactions running through a blockchain are totally independent. If you can isolate the independent ones, you can let them pass through at high speed, order be damned. This is where blockchain meets parallel processing, a technology that has quietly powered a computing revolution. Parallel computing went mainstream in 2000 with Pentium 4's "Hyperthreading," and the technology became standard in 2005 with Pentium D's dual-core chip. It was a game changer. Today, serial processing is outdated for any demanding application, and most improvements in computer power come from restriking the balance between speed and parallelism. This year, Aptos, the blockchain built by former Meta engineers, released a paper and an article describing their efforts to bring parallel computing onchain. Inspired by this, Polygon recently announced that they will introduce parallel transaction processing on their network, approximately doubling throughput. New protocols targeting parallel execution are also emerging. Some are dedicated to Ethereum, while others remain Ethereum-compatible, but seek wider applicability across base-layer blockchains. The race is on. Going fast will enable new use cases in payments, gaming, trading, and more. The technological limitations on blockchains are being removed. Transactions are marching two-by-two and more. Hoorah! Hoorah!

BACK