go-randomx/README.md

RandomX (Golang Implementation)

RandomX is a proof-of-work (PoW) algorithm that is optimized for general-purpose CPUs. RandomX uses random code execution (hence the name) together with several memory-hard techniques to minimize the efficiency advantage of specialized hardware.

---

Fork from git.dero.io/DERO_Foundation/RandomX. Also related, their Analysis of RandomX writeup.

Original code failed RandomX testcases and was implemented using big.Float.

---

This package implements RandomX without CGO, using only Golang code, native float64 ops, some assembly, but with optional soft float purego implementation.

All test cases pass properly.

Supports Full mode and Light mode.

For the C++ implementation and design of RandomX, see github.com/tevador/RandomX

Feature	386	amd64	arm	arm64	mips	mips64	riscv64	wasm
purego	✅	✅	✅	✅	✅	✅	✅	✅
Hardware Float Operations	✅	✅	❌	✅	❌	❌	❌	❌
Hardware AES Operations	❌	✅	❌	❌	❌	❌	❌	❌
Native Superscalar Execution	✅	✅	✅	✅	✅	✅	✅	✅
Superscalar JIT Execution	❌	✅*	❌	❌	❌	❌	❌	❌
Native VM Execution	✅	✅	❌	✅	❌	❌	❌	❌
VM JIT Execution	❌	✅*	❌	❌	❌	❌	❌	❌

A pure Golang implementation can be used on platforms without hard float support or via the purego build flag manually.

Any platform with no hard float support or when enabled manually will use soft float, using softfloat64. This will be very slow.

Native hard float can be added with supporting rounding mode under asm.

JIT only supported under Unix systems (Linux, *BSD, macOS), and can be hard-disabled via the disable_jit build flag, or at runtime.