mirror of
https://github.com/lux-apps/papers.git
synced 2026-10-30 18:15:21 +00:00
zeekay
eaf2da6db5
New section 08: the end-to-end result behind the microbenchmarks ā a real 5-node BFT network producing + finalizing C-Chain EVM blocks under load, byte-identical across all 5, with ML-DSA-65 proposer identity + ML-KEM-768 handshake, no automining. Documents the 4-gate block-production dependency chain (NodeID scheme match, ML-DSA proposer identity, engine consume-on-error, cadence floor), the measured throughput table (44ā108 TPS as WindowDuration 5sā1s), the throughput law (TPS = tx_per_block / cadence, cadence-bound not serialization-bound), the sub-second frontier (slot-handoff pin), and the per-chain cadence policy (1ms co-located DEX vs 100ms-1s public). Compiles clean (13pp).
68 lines
2.2 KiB
TeX
68 lines
2.2 KiB
TeX
\documentclass[11pt,a4paper]{article}
|
|
\usepackage{../shared/luxcover}
|
|
\usepackage[utf8]{inputenc}
|
|
\usepackage[T1]{fontenc}
|
|
\usepackage{amsmath,amssymb,amsfonts}
|
|
\usepackage{graphicx}
|
|
\usepackage{hyperref}
|
|
\usepackage{booktabs}
|
|
\usepackage{enumitem}
|
|
\usepackage{xcolor}
|
|
\usepackage{siunitx}
|
|
\usepackage{listings}
|
|
|
|
\hypersetup{colorlinks=true,linkcolor=black,citecolor=blue,urlcolor=blue}
|
|
|
|
\lstset{
|
|
basicstyle=\ttfamily\scriptsize,
|
|
breaklines=true,
|
|
frame=single,
|
|
showstringspaces=false,
|
|
}
|
|
|
|
\title{The Lux ZAP Stack:\\
|
|
A Decomplected Ten-Layer Blockchain Architecture}
|
|
\author{Lux Industries, Inc. \\
|
|
\texttt{[email protected]}}
|
|
\date{2026-06-03}
|
|
|
|
\begin{document}
|
|
\luxcoverpage
|
|
|
|
\begin{abstract}
|
|
We present the Lux ZAP Stack, a blockchain architecture in which a
|
|
single binary buffer is the wire format, the in-memory representation,
|
|
the on-disk record, and the snapshot stream. There is no
|
|
\texttt{codec.Manager}, no reflection-driven marshal, no encode/decode
|
|
hop between mempool and state. Each of ten layers does exactly one
|
|
job; each consumes its predecessor as bytes and produces its successor
|
|
as bytes. The architecture is the umbrella for LPs 022, 023, 182, 184,
|
|
185, 186, 201, 205, 208, 209, 210, 211, 217, and 218. Measured
|
|
head-to-head against the reflection codec \texttt{avalanchego} master
|
|
still runs, ZAP parses a transaction $73\times$ faster, validates a
|
|
\num{1000}-tx block $70\times$ faster, and --- because a validated block
|
|
stays as bytes with light views rather than a materialized Go object
|
|
graph --- holds a live block in $114\times$ less resident memory
|
|
(\SI{17}{\kilo\byte} vs \SI{1990}{\kilo\byte}). This paper expands
|
|
LP-200 (\texttt{lps/LP-200-zap-stack.md}) with the layer model, the ten
|
|
decomplected concerns, the measurement methodology, and the explicit
|
|
composition table tying the umbrella to its sibling LPs.
|
|
\end{abstract}
|
|
|
|
\tableofcontents
|
|
\newpage
|
|
|
|
\input{sections/01-introduction}
|
|
\input{sections/02-ten-layer-model}
|
|
\input{sections/03-decomplected-concerns}
|
|
\input{sections/04-performance}
|
|
\input{sections/05-composition}
|
|
\input{sections/06-security}
|
|
\input{sections/08-live-consensus}
|
|
\input{sections/07-future-work}
|
|
|
|
\bibliographystyle{plain}
|
|
\bibliography{biblio}
|
|
|
|
\end{document}
|