PQ-WB-KEM: White-Box ML-KEM-768 1.0
PQ-WB-KEM: Toward a White-Box Construction of ML-KEM-768 with Arithmetic Masking for M2M Communications - Mathematics (MDPI)
Loading...
Searching...
No Matches
arm64_aggregate.py File Reference

Aggregate N independent qemu-aarch64 liboqs runs into mean +/- 95% CI. More...

Go to the source code of this file.

Namespaces

namespace  arm64_aggregate
 

Functions

 arm64_aggregate.interval (xs)
 Mean, standard deviation, and 95% CI half-width for a sample.
 
 arm64_aggregate.main ()
 Read per-run JSONs, aggregate the three operations, write the summary.
 
 arm64_aggregate.tval (n)
 Two-sided 95% t critical value for n samples.
 

Variables

dict arm64_aggregate.TVAL
 

Detailed Description

Aggregate N independent qemu-aarch64 liboqs runs into mean +/- 95% CI.

A single qemu-aarch64 run of the ML-KEM-768 benchmark is not stable: functional emulation absolute times drift run-to-run with host load and the translation cache. Reporting one run would pin a moving number. This reads the N per-run JSONs produced by build_arm64_benchmark.sh and reduces each operation's per-run mean to a mean and a 95% confidence interval, the same interval statistic wallclock_ci.py uses (Student t, N runs), so the ARM64/QEMU baseline is an interval rather than a single transcribed value. mean_ms is retained per operation so consumers that read it (overhead_calculation.py) keep working.

Note
The absolute times are functional-emulation figures dependent on host load, not real ARM silicon; only the dimensionless ratio is comparable. Called by build_arm64_benchmark.sh under WSL; not part of the Python run_all suite because it needs the aarch64 cross-toolchain and qemu-aarch64.
Author
Ugur Coruh

Definition in file arm64_aggregate.py.