WaferScaleAI Orchestrator
8.7
A system designed to facilitate the deployment of trillion-parameter AI models across multiple Cerebras WSE-3T wafer-scale chips. The software would manage model splitting, weight distribution, and coordinate processing across multiple wafers to achieve high-throughput inference (1000-2000+ tokens per second).
280h
mvp estimate
8.7
viability grade
0
views
technology stack
C#
Java
Difficult
inspired by
Cerebras deploys trillion-parameter AI models across wafers