Crate gpu_scatter_gather 

GPU Scatter-Gather Wordlist Generator

The world’s fastest wordlist generator using GPU acceleration.

Core Innovation

Instead of sequential odometer iteration (like maskprocessor), this uses direct index-to-word mapping via mixed-radix arithmetic on the GPU:

Index → Mixed-Radix Decomposition → Word

This enables:

• Massively parallel generation (every thread works independently)
• O(1) random access to any position in the keyspace
• Perfect GPU utilization with coalesced memory access
• 500M-1B+ words/second throughput (3-7x faster than maskprocessor)
• Multi-GPU support for linear scaling (v1.1.0+)

Example Usage

use gpu_scatter_gather::{WordlistGenerator, Charset};

let mut generator = WordlistGenerator::builder()
    .charset(1, Charset::from("abc"))
    .charset(2, Charset::from("123"))
    .mask(&[1, 2, 1, 2])  // Pattern: ?1?2?1?2
    .build()?;

// Generate wordlist
for word in generator.iter() {
    println!("{}", String::from_utf8_lossy(&word));
}

Re-exports

pub use charset::Charset;

pub use keyspace::calculate_keyspace;

pub use keyspace::index_to_word;

pub use mask::Mask;

Modules

bindings

Output bindings for different consumption patterns

charset

Charset management for wordlist generation

ffi

C Foreign Function Interface (FFI) layer

gpu

GPU acceleration module using CUDA

keyspace

Keyspace calculation and index-to-word conversion

mask

Mask pattern parsing and management

multigpu

Multi-GPU context management

transpose

SIMD-optimized matrix transpose for column-major to row-major conversion

Structs

WordlistGenerator

Main wordlist generator

WordlistGeneratorBuilder

Main wordlist generator builder

WordlistIterator

Iterator over wordlist entries