sub0 Asia 2024 - Shawn Tabrizi
summary
Defines standard TPS as a keep‑alive balance transfer between existing accounts and builds a simple JS meter + spammer that pre‑signs transactions to avoid the signing bottleneck. Shows substrate weight and a 2s block‑import window, demonstrates default 75% dispatch ratio vs a full block node, and measures ~1,400–1,600 STPS on one core. Changes (longer import time, shorter block time) double throughput; batching 100–1000 ops pushes apparent TPS into tens of thousands while noting weight overestimation and storage/networking trade‑offs.
AI-written summary · about
chapters (10)
- 0:00Introduction and goals
- 1:02The case against traditional TPS
- 5:09Alternative measurements
- 9:01Contextualizing metrics
- 13:06Polkadot scalability vision
- 16:25Understanding block weight
- 20:53Analyzing block imports
- 23:19Practical benchmarking demo
- 37:17Scaling and manipulation
- 43:23Optimization and conclusion
related talks
49mScaling Tezos with rollups
14mAn Off-The-Chain Execution Environment for Scalable Testing and Profiling...
15mFantom 2.0
13mThe End of Old Money
44mPerformance Testing with a "Big Data Tool"
14mEthereum’s Roadmap to 10M TPS
09mStellar's Future A Technical Perspective
23mBuilding Networks That People Actually Use