Share

Apple researchers describe Minions, a system that aggregates idle GPUs on macOS devices to run large-scale ML inference, according to machinelearning.apple.com, which says the approach tackles cost, elasticity and wait-time limits of centralized GPU infrastructure.

What else the source says

• Centralized GPU infrastructure has high cost, limited elasticity, long wait times.
• macOS devices carry powerful GPUs that sit largely idle in normal use.
• Minions pools idle Apple Silicon GPUs across macOS devices into a virtual cloud.
• System is built to run large-scale machine learning inference workloads.

Why it matters

Apple may aim to cut inference costs via idle device pooling.

Source. machinelearning.apple.com

SOTO home