MLRTHash v1 — Technology

Block Header

96-byte structure

Total: exactly 96 bytes. The mining loop varies Nonce (offset 80, 8 bytes); Height is included in the hash so reorgs at different heights produce different work.

Hash Function

Double SHA-3-256

MLRTHash(header) = SHA3-256(SHA3-256(serialize(header)))

Both inner and outer hashes are the standard NIST-finalised SHA-3-256 (Keccak-f[1600] sponge with rate=1088 bits, capacity=512 bits, 24 rounds). The output is interpreted as a 256-bit big-endian unsigned integer for difficulty comparison.

A header is a valid block header iff:

MLRTHash(header) ≤ CompactToBig(header.Bits)

Consensus

Difficulty & retargeting

• Initial difficulty: 1 (bits = 0x207fffff)
• Retarget interval: every 2016 blocks (~67 hours at target)
• Target block time: 120 seconds (2 minutes)
• Adjustment formula: newTarget = oldTarget × actualTime / targetTime, clamped to [1/4, 4] per cycle
• Median-time-past rule: every block timestamp must exceed the median of the prior 11 blocks

Reference Implementation

Go (canonical)

package consensus

import (
    "github.com/computervirtualservices/malairte/internal/crypto"
    "github.com/computervirtualservices/malairte/internal/primitives"
)

func MLRTHash(header *primitives.BlockHeader) [32]byte {
    return crypto.DoubleSHA3256(header.Serialize())
}

func HashMeetsDifficulty(hash [32]byte, bits uint32) bool {
    target := CompactToBig(bits)
    hashInt := new(big.Int).SetBytes(hash[:])
    return hashInt.Cmp(target) <= 0
}

CUDA (GPU)

Verification

Test vectors

Reference outputs produced by the canonical Go implementation and verified byte-for-byte by the CUDA implementation. If your implementation produces these three hashes, it's correct.

Parameters

Consensus constants