ASIC-RAG-CHIMERA

GPU simulation of a SHA-256 hash engine inspired by Bitcoin mining ASICs, wired into a RAG pipeline. Pure software; no real ASIC hardware required.

---

What this is

ASIC-RAG-CHIMERA is a software research artifact. It consists of:

1. A GPU-accelerated SHA-256 hash engine implemented in PyTorch that simulates the kind of bulk hashing a Bitcoin-style ASIC would do. It runs on a normal CUDA GPU (or CPU fallback). It is a software simulation, not an ASIC.
2. A cryptographic RAG pipeline that indexes documents by SHA-256 tags instead of plaintext embeddings, encrypts blocks with AES-256-GCM, and verifies integrity with a Merkle tree.
3. A demonstration workflow with synthetic patient records illustrating how the pipeline could be configured for privacy-sensitive data (see ASIC-RAG-HEALTH_Validation/). The data is fabricated. This is not a clinical tool and must not be used for medical decision making.

What this is NOT

• Not real ASIC hardware. There is no silicon, no Verilog tape-out, no FPGA bitstream. The "ASIC" in the name refers to the architectural inspiration for the GPU simulator module (asic_simulator/).
• Not a medical device. The health demo uses synthetic records and is illustrative only.
• Not a Bitcoin miner. The SHA-256 engine is used for content-addressed indexing, not proof-of-work.

Installation

pip install asic-rag-chimera

Optional extras:

pip install "asic-rag-chimera[gpu]"      # Ensure PyTorch with CUDA is available
pip install "asic-rag-chimera[wandb]"    # Experiment tracking
pip install "asic-rag-chimera[dev]"      # Tests, build, twine

From source:

git clone https://github.com/Agnuxo1/ASIC-RAG-CHIMERA.git
cd ASIC-RAG-CHIMERA
pip install -e ".[dev]"

Quick start

import os
from asic_simulator import GPUHashEngine, IndexManager, KeyGenerator
from rag_system import DocumentProcessor, QueryEngine

hash_engine = GPUHashEngine()
index_manager = IndexManager()
key_generator = KeyGenerator(master_key=os.urandom(32))

processor = DocumentProcessor()
blocks = processor.create_blocks("Your document content here")

query_engine = QueryEngine(index_manager, hash_engine)
results = query_engine.search("your query", max_results=5)

Or use the integrated facade:

from asic_rag_chimera import ASICRAGSystem
system = ASICRAGSystem(storage_path="./data", master_key=os.urandom(32))
system.ingest("document.txt")
result = system.query("What is the revenue?")

Architecture

┌──────────────┐    text     ┌─────────────┐    tag hashes    ┌────────────────────┐
│  User query  │────────────▶│  LLM (GPU)  │─────────────────▶│ GPU SHA-256 engine │
└──────────────┘             └─────────────┘                  │  (asic_simulator)  │
                                    ▲                         └──────────┬─────────┘
                                    │ decrypted blocks                   │ hash lookup
                                    ▼                                    ▼
                         ┌────────────────────────────────────────────────────┐
                         │  Encrypted block storage (LMDB / AES-256-GCM)      │
                         │  Merkle tree integrity proofs                      │
                         └────────────────────────────────────────────────────┘

Running tests and coverage

pytest tests/ -v                                    # 53/53 tests pass
pytest tests/ --cov=asic_simulator --cov=rag_system --cov=asic_rag_chimera --cov-report=term --cov-report=xml

Measured line coverage on the core packages is 57% (1658 statements, 706 missed), written to coverage.xml. Previous READMEs claimed "100%" — that was never measured. The 53 tests all pass; they simply don't exercise every branch of keyword_extractor, query_engine, key_generator, etc.

Security model

Attack vector	Traditional RAG	ASIC-RAG-CHIMERA
Disk theft	Plaintext exposure	Encrypted blocks
Embedding inversion	Partial recovery	N/A (no embeddings stored)
Index enumeration	Knowledge graph exposed	Opaque SHA-256 tags
Key capture	Permanent access	30-second TTL session keys
Data tampering	Undetected	Merkle proof verification

Claims above describe the design. This is a research prototype, not an audited product.

Repository layout

asic_simulator/     GPU SHA-256 engine + tag index + key generator
rag_system/         Document processor, block storage, query engine
asic_rag_chimera.py Integrated facade (ASICRAGSystem)
tests/              53 pytest tests
benchmarks/         Microbenchmarks for hash and search latency
archive/            Historical artefacts (PDFs, HTML, duplicate dirs) — not shipped
huggingface_space/  HF Space demo app

Citation

@software{angulo_asic_rag_chimera_2026,
  author  = {Angulo de Lafuente, Francisco},
  title   = {ASIC-RAG-CHIMERA: GPU Simulation of a SHA-256 Hash Engine for Cryptographic RAG},
  year    = {2026},
  version = {1.0.0},
  doi     = {10.5281/zenodo.17872052},
  url     = {https://github.com/Agnuxo1/ASIC-RAG-CHIMERA}
}

See CITATION.cff.

Author

Francisco Angulo de Lafuente — GitHub @Agnuxo1

License

MIT — see LICENSE.

---

Related projects

Part of the @Agnuxo1 v1.0.0 open-source catalog (April 2026).

AgentBoot constellation — agents and research loops

• AgentBoot — Conversational AI agent for bare-metal hardware detection and OS install.
• autoresearch-nano — nanoGPT-based autonomous ML research loop.
• The Living Agent — 16x16 Chess-Grid autonomous research agent.
• benchclaw-integrations — Agent-framework adapters for the BenchClaw API.

CHIMERA / neuromorphic constellation — GPU-native scientific computing

• NeuroCHIMERA — GPU-native neuromorphic framework on OpenGL compute shaders.
• Holographic-Reservoir — Reservoir computing with simulated ASIC backend.
• QESN-MABe — Quantum-inspired Echo State Network on a 2D lattice (classical).
• ARC2-CHIMERA — Research PoC: OpenGL primitives for symbolic reasoning.
• Quantum-GPS — Quantum-inspired GPU navigator (classical Eikonal solver).