mirror of
https://github.com/lux-apps/papers.git
synced 2026-11-03 05:07:46 +00:00
zooqueen
12912be1b5
- canonical lux-corona-pq paper: rank-7 Module-LWE (A in R_q^{8x7}, q=48-bit), Module-SIS
- scrub residual Corona Ring-LWE/R-LWE labels across ethos-of-lux, pqsafe-il5,
lux-whitepaper, lux-quasar-composition, lp-020; Corona<->Pulsar = parameter regime,
both Module-LWE (not rank-1-vs-rank-k)
- LLM.md: drop stale 'Corona as rank-1' wobble note; align to code/Ringtail params
- rebuild affected PDFs
35 lines
1.8 KiB
TeX
35 lines
1.8 KiB
TeX
\section{Conclusion}
|
|
|
|
\label{sec:conclusion}
|
|
%% ============================================================================
|
|
|
|
QuasarSTM~3.0 launched on 2025-12-25 as the GPU-native ordered MVCC
|
|
substrate for Lux's Nova and Nebula execution modes. It got the lane
|
|
abstraction, the wave-tick scheduler, the three-tier validation
|
|
skeleton, the deterministic contention manager, and the multi-GPU
|
|
sharding stub right. It launched with placeholder cryptography, partial
|
|
EVM coverage, and the production-critical paths (Tier~1 lane-clock
|
|
fast path, Tier~3 semantic reducers, ConflictSpec, LaneClass, fiber
|
|
checkpoints, commit horizon, MVCC GC, Motor VersionBlock layout, A/B/M/F
|
|
drain services, CSMV commit server, formal CPU reference) sketched.
|
|
|
|
QuasarSTM~4.0 closes every one of those gaps. It activated on
|
|
2026-02-14, lives on the same substrate, preserves the 3.0 invariants
|
|
verbatim, and ships 175-opcode EVM coverage, real BLS12-381 / Corona
|
|
Module-LWE / Groth16 verifiers, the production validation tiers, the
|
|
production execution and commit primitives, the production drain
|
|
service set, the cross-backend determinism harness, and a formally
|
|
specified CPU reference. The wallclock cost on 1024-tx end-to-end
|
|
stress drops from $150\,\text{ms}$ to $108\,\text{ms}$ on M1 Max
|
|
despite real cryptography being four to seven times more expensive
|
|
than the HMAC-keccak placeholders, because Tier~1 lane-clock and
|
|
Tier~3 reducer commit absorb the hot writes.
|
|
|
|
There is no future work in this paper. There is research on what comes
|
|
next---multi-GPU MVCC over RDMA, kernel refinement proofs, operation-
|
|
window validation for pathological long transactions---and that
|
|
research is tracked as QuasarSTM~5.0 in a separate document. The 4.0
|
|
production spec is complete, activated, and final.
|
|
|
|
%% ============================================================================
|