Experimental multi-dimensional hashing algorithm focused on hardware resistance and balanced performance.
CryptixHash v2, also known as Cryptix OX8 (Octonion 8 Dimensions), is a modern hashing algorithm designed for deterministic hashing and encryption across multiple programming languages and hardware platforms.
The algorithm focuses on resistance against specialized hardware (ASICs / FPGAs) while maintaining efficient and balanced execution on CPUs and GPUs.
- Minimize efficiency on ASICs and FPGAs
- Maintain balanced CPU ↔ GPU performance
- Ensure high entropy and unpredictability
- Introduce strong non-linear computation patterns
- Improve resistance to side-channel and low-level attacks
Unlike traditional memory-hard algorithms, Cryptix OX8 avoids excessive memory usage.
Instead of focusing on how much memory is used, the algorithm focuses on:
How memory is accessed, manipulated, and randomized
| Hardware | Memory Usage |
|---|---|
| GPU | ~200–300 MB |
| CPU | Mostly within L1–L3 cache |
- Penalizes CPUs and GPUs unnecessarily
- Breaks CPU/GPU performance balance
- Limits compatibility with older hardware
Cryptix OX8 introduces complex, dependency-heavy computation patterns to limit optimization on ASICs and FPGAs.
- Integer type switching (
u8,u16,u32,i64, etc.) - Conditional and nested branching
- Byte-level and nibble-level manipulation
- Dynamic XOR and rotation values
- Irregular iteration patterns (illiterations)
- Strong dependency on previous computation steps
- Pseudo-random memory access
- Runtime-dependent calculations
- Reduced parallelization efficiency
- Limited pipelining potential
- Increased execution unpredictability
Cryptix OX8 introduces an experimental approach based on Octonion algebra (8D).
- Non-commutative operations
- High entropy generation
- Non-linear transformations
- Resistance to reverse computation
This represents a novel direction in hashing, combining abstract algebra and physics-inspired computation.
- Expansion of Octonion-based cryptographic primitives
- Evaluation as a potential alternative to SHA-3 (research phase)
- Exploration of higher dimensions (e.g. Sedenions / 16D)
Sedenion-based approaches currently face unresolved issues such as zero-divisors.
- Rust
- Cuda (C++)
- OpenCL (C)
- Go
- C#
- WASM
- Javascript
MIT License