Determine the root cause of repeated-crash behavior during rapid consecutive mints on higher-RAM devices

Determine the root cause of the crash error observed on higher-RAM smartphones when running the FibRace application and performing many consecutive proof mints (usually more than one hundred) using the Cairo M prover. Ascertain whether improper termination of the spawned Rust proof-generation thread or insufficient inter-proof delay is responsible for out-of-memory–type failures during rapid, repeated on-device proof generation.

Background

During the FibRace campaign, the authors observed crash errors on devices with higher RAM when users minted many proofs consecutively, typically after more than a hundred iterations. Although they hypothesize that incomplete termination of a spawned Rust thread between proofs could cause OOM-type crashes—potentially exacerbated by short delays between verification/settlement and the next proof—they did not conclusively identify the root cause within the campaign timeframe.

This uncertainty affects the reliability of high-frequency, back-to-back on-device proving workflows and points to potential issues in thread lifecycle management or inter-proof timing when using the Cairo M prover in a mobile app context.