By using Convex Optimization combined with EZKL, we can now prove trading strategies are optimal and enable new financial products on-chain that don't compromise on security.
Here's why this matters for the future of finance:
Traditional DeFi has a limitation:
The Ethereum Virtual Machine (EVM) can't handle complex calculations.
This means AMMs like Uniswap have been restricted to using simple formulas that don't always capture market complexity.
But EZKL changes this:
Nov 28, 2023 • 12 tweets • 4 min read
1/⚙️ZKML with Accelerated Hardware
We've integrated the Icicle GPU acceleration library into EZKL, in collaboration with the Ingonyama team.
While we've written about the specifics, we also discuss broader strategies for hardware acceleration with ZKML systems.
2/📋Preliminary Results
Against a baseline CPU run for aggregate circuits, we observe a ~98% reduction in MSM times and ~35% reduction in total aggregate proof time. 🚀