Rul (.rl) Whitepaper & Empirical Evaluation — Zenodo DOI 10.5281/zenodo.23722094
Technical whitepaper, first-principles architectural breakdown, 80 MB columnar benchmark methodology, and cryptographic SHA-256 Prior Art verification for Rul (.rl).
Rul (.rl): A 128-Bit Schedule- and Proof-Carrying Binary SSA-DAG Programming Language
Traditional programming languages (C, Go, Rust, Python) are designed around human biological constraints: low-bandwidth keyboard input (~40 bits/s) and visual ASCII source reading, which impose a massive serialization, parsing, type-inference, and LLVM optimization tax on autonomous AI Agents.
Rul (.rl) eliminates textual source code as the primary representation, replacing it with a memory-mapped flat binary topological SSA-DAG where every instruction is a fixed 128-bit (16-byte) NodeFrame encoding Algorithm + ScheduleHint + ProofIdx + AffineToken/ReadLease. Because the AI Agent emits explicit schedule hints and safety witnesses alongside the algorithm, the Rul compiler performs zero type inference and zero search: a single-pass O(N) linear verifier mathematically guarantees 100% absence of memory aliasing, use-after-consume, buffer overflows, and sandbox capability violations in ~5.78 ns/node, while the 3-tier execution engine and direct x86_64 ELF64 AOT compiler fuse multi-stage columnar pipelines into single-pass register loops in ~6.4 µs.
Why Human Text Languages Are a Bottleneck for AI Agents
- • AI serializes internal tensor intent into thousands of ASCII characters (pub fn, braces, semicolons).
- • Compiler spends seconds lexing text, building ASTs, and running NP-hard borrow-checker / register-allocation searches.
- • Hardware schedule (SIMD / cache tiling / threading) is tangled inside imperative loop syntax.
- • AI emits 16-byte binary frames directly (0 ms parsing, 0% syntax errors).
- • Single forward loop verifies SSA order, memory versions, and bounds in ~5.78 ns/instruction.
- • Mutating 1 byte (Byte 2: schedule_hint) switches execution from 1-core SIMD to MultiCore + ZoneMap.
Empirical Benchmark Results (80 MB / 10M f64)
| IMPLEMENTATION | MODE | LATENCY | BANDWIDTH | VS C 8-ACC |
|---|---|---|---|---|
| [A] Imperative AoS + Vec::collect() | 1-Core AoS + Heap Allocation | 70.66 ms | 1.13 GB/s | 0.10x |
| [B] Idiomatic Rust SoA (filter().map().sum()) | 1-Core SoA Iterator Chain | 19.12 ms | 4.18 GB/s | 0.38x |
| [C] Hand-Written C/Rust 8-Accumulator SoA | 1-Core 8-Lane Unrolled Baseline | 7.32 ms | 10.92 GB/s | 1.00x (Baseline) |
| [D] Rul (.rl) Tier 2 Fused SIMD (0x11) | 1-Core L1-Tile SIMD (SimdStream) | 5.91 ms | 13.53 GB/s | 1.24x (1-Core vs 1-Core) |
| [E] Rul (.rl) Tier 2 MultiCore (0x12) | All Cores (1-Byte Schedule Mutation) | 3.19 ms | 25.04 GB/s | 2.29x (Schedule Byte) |
| [F] Rul (.rl) Tier 2 MultiCore + ZoneMap | All Cores + O(1) Cold Tile Skipping | 1.55 ms | 51.53 GB/s | 4.72x (Peak Speedup) |
Cite Rul (.rl) in Research
@software{rul_2026_spec,
author = {Rul Core Architecture Team},
title = {{Rul (.rl): A 128-Bit Schedule- and Proof-Carrying Binary SSA-DAG Programming Language and Execution Engine for Autonomous AI Agents}},
month = oct,
year = 2026,
publisher = {Zenodo},
version = {0.4.0},
doi = {10.5281/zenodo.23722094},
url = {https://doi.org/10.5281/zenodo.23722094}
}Prior Art & Cryptographic Verification
SHA-256 digests of the core specification and Rust implementation anchored via OpenTimestamps in docs/prior_art/.
| ARTIFACT PATH | ROLE | SHA-256 DIGEST |
|---|---|---|
| docs/rul_spec_v0.md | Official .rl Binary Layout & O(N) Verifier Specification | 9174ff81b429b19785646f25707034a36b4ba7b4a7964b7e04ddbc31fda9d9c6 |
| docs/WHITEPAPER_RUL_V0_1.md | Technical Whitepaper & Empirical Evaluation (Zenodo DOI) | 141ea3d9b348369b4e233746619fee443387e931c821fe2eb32b93f760571b98 |
| docs/new_language_for_AI.md | First-Principles Architectural Breakdown for AI Agents | 88043cda84b7289e9fe26f4bc3547a58b994a91b1df8774bbc206dc6dba5504d |
| src/bytecode.rs | 64B RulHeader, 16B ArenaDescriptor, ProofEntry & 128-bit NodeFrame | 3a8c780cf4181c936b303943eafa3729eba9e519cd9d55569c28f1f630f26c03 |
| src/verifier.rs | Single-Pass O(N) Linear Static Verifier (~5.78 ns/node) | feaa5ae7e94b2bede64772d1ce595286bed35ac2afc9be9b571836813a4eb11b |
| src/engine.rs | 3-Tier Execution Engine, L1-Tile Fuser, ZoneMap & White-Box Tracer | 8037e2f718dc24d372a67111d9a6c5481ade4ff55a62d1b315b35c9ccf879bcf |
| src/aot.rs | Zero-LLVM Linux x86_64 AVX2 ELF64 Compiler (~6.4 µs) | fe8eddae822fae0df6ea751aaa7919fc81675a7c169d2aadb4c3b9e6f05f27ac |
| src/mcp.rs | In-Memory stdio JSON-RPC 2.0 Model Context Protocol Server | c58faa2f0ac626412a6c8ef00ae67dabc85b863f4c0ba73c3e16a0ff909a91c6 |