mirror of
https://github.com/lux-apps/papers.git
synced 2026-10-30 18:15:21 +00:00
zeekay
4a753bb639
Restructured from flat .tex/.pdf files into per-paper subdirectories. Added Antje Worring as co-author on all papers. Synced PDFs to pdfs/ for download site.
40 lines
1.3 KiB
TeX
40 lines
1.3 KiB
TeX
\section{Performance}
|
|
|
|
\label{sec:perf}
|
|
%% =====================================================================
|
|
|
|
Benchmarks on Apple M4 Max (Go implementation with SIMD NTT):
|
|
|
|
\begin{table}[h]
|
|
\centering
|
|
\begin{tabular}{lrrr}
|
|
\toprule
|
|
\textbf{Algorithm} & \textbf{KeyGen} & \textbf{Sign/Encap} &
|
|
\textbf{Verify/Decap} \\
|
|
\midrule
|
|
ECDSA (secp256k1) & 0.02 ms & 0.03 ms & 0.05 ms \\
|
|
Ed25519 & 0.01 ms & 0.02 ms & 0.04 ms \\
|
|
\midrule
|
|
ML-KEM-768 & 0.04 ms & 0.05 ms & 0.04 ms \\
|
|
ML-DSA-65 & 0.10 ms & 0.35 ms & 0.12 ms \\
|
|
SLH-DSA-128f & 0.01 ms & 3.50 ms & 0.20 ms \\
|
|
\midrule
|
|
Hybrid (ECDSA + ML-DSA-65) & 0.12 ms & 0.38 ms & 0.17 ms \\
|
|
\bottomrule
|
|
\end{tabular}
|
|
\caption{Cryptographic operation latencies.}
|
|
\label{tab:latency}
|
|
\end{table}
|
|
|
|
ML-DSA verification is approximately $2.4\times$ slower than ECDSA
|
|
verification. For a blockchain that processes 4,000 transactions per
|
|
second, this adds 480 $\mu$s of verification overhead per block of
|
|
1,000 transactions---negligible compared to consensus latency.
|
|
|
|
The larger signature sizes (2,420 bytes for ML-DSA-44 vs.\ 65 bytes
|
|
for ECDSA) increase transaction data by approximately 2.3~KB. At
|
|
4,000 TPS, this is 9.2~MB/s of additional bandwidth---well within the
|
|
capacity of modern validator networks.
|
|
|
|
%% =====================================================================
|