npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@pid7/ashwa

v1.0.1

Published

Hardware accelerated routines for single substring search

Vulnerabilities

Powered byPowered by Snyk
  • 0High
  • 0Medium
  • 0Low

Links

Git

Maintainers

pid7pid7

Keywords

simdsimd128searchstringrustwasmnapi-rsnode

Readme

npm Tests License

Ashwa 🐎

Hardware accelerated routines for single substring search

💡 TIP: @pid7/ashwa runs seamlessly across Node.js, Bun, Deno, and modern browser / WebWorker environments.

ℹ️ NOTE: The optimal execution backend (native N-API SIMD bindings for Node / Bun / Deno or WebAssembly SIMD for browsers) is automatically detected and selected at runtime with zero manual configuration.

Index

Supported Platforms

| Architecture | Target Platform | Target ISA | Fallback | |:----------------|:---------------------------------------|:-------------------------------------|:------------| | x86_64 | Linux, macOS, Windows | AVX-512BW, AVX2, SSE4.2, SSSE3, SSE2 | 64-bit SWAR | | AArch64 (ARM64) | Apple Silicon, Linux ARM64 | 128-bit ARM NEON | 64-bit SWAR | | WebAssembly | Browsers, WebWorkers, Node.js (wasm32) | WASM SIMD128 (simd128) | 32-bit SWAR |

Installation

Install @pid7/ashwa using your preferred package manager:

npm install @pid7/ashwa

Usage

Node

Native N-API bindings execute synchronously with zero overhead and full hardware SIMD acceleration,

import { searchOne, searchTwo } from '@pid7/ashwa';

const haystack = new TextEncoder().encode("The quick brown fox jumps over the lazy dog");
const indexOne = searchOne(haystack, "f".charCodeAt(0));
console.log(`Found 'f' at byte index: ${indexOne}`); // 16

Browser

In browser and WebWorker runtimes, WebAssembly SIMD (simd128) is used. The WASM module automatically initializes asynchronously on the first search call, or can be pre-warmed during application startup:

import { init, searchOne, searchTwo } from '@pid7/ashwa';

// Optional pre-init
await init();

const haystack = new TextEncoder().encode("The quick brown fox jumps over the lazy dog");
const index = await searchOne(haystack, "f".charCodeAt(0));
console.log(`Found 'f' at byte index: ${index}`); // 16

API Reference

searchOne(haystack, needle)

function searchOne(
  haystack: Uint8Array,
  needle: number,
): number | null | Promise<number | null>;

Searches for the first occurrence of needle (byte value 0–255) within haystack.

  • Parameters:
    • haystack: Uint8Array / Buffer — The byte sequence to search.
    • needle: number — The target byte value (0–255) to locate.
  • Returns:
    • In Node.js / Bun / Deno: number | null (synchronous 0-based byte index, or null if not found).
    • In Browser / WebWorker: Promise<number | null> (resolves to 0-based byte index, or null if not found).

searchTwo(haystack, needle)

function searchTwo(
  haystack: Uint8Array,
  needle: Uint8Array | [number, number] | number[],
): number | null | Promise<number | null>;

Searches for the first occurrence of a two-byte needle within haystack.

  • Parameters:
    • haystack: Uint8Array / Buffer — The byte sequence to search.
    • needle: Uint8Array | [number, number] | number[] — A 2-byte sequence to locate.
  • Returns:
    • In Node.js / Bun / Deno: number | null (synchronous 0-based byte index, or null if not found).
    • In Browser / WebWorker: Promise<number | null> (resolves to 0-based byte index, or null if not found).

searchThree(haystack, needle)

function searchThree(
  haystack: Uint8Array,
  needle: Uint8Array | [number, number, number] | number[],
): number | null | Promise<number | null>;

Searches for the first occurrence of a three-byte needle within haystack.

  • Parameters:
    • haystack: Uint8Array / Buffer — The byte sequence to search.
    • needle: Uint8Array | [number, number, number] | number[] — A 3-byte sequence to locate.
  • Returns:
    • In Node.js / Bun / Deno: number | null (synchronous 0-based byte index, or null if not found).
    • In Browser / WebWorker: Promise<number | null> (resolves to 0-based byte index, or null if not found).

searchN(haystack, needle)

function searchN(
  haystack: Uint8Array,
  needle: Uint8Array | number[],
): number | null | Promise<number | null>;

Searches for the first occurrence of an arbitrary byte sequence needle within haystack.

  • Parameters:
    • haystack: Uint8Array / Buffer — The byte sequence to search.
    • needle: Uint8Array | number[] — An arbitrary byte sequence to locate.
  • Returns:
    • In Node.js / Bun / Deno: number | null (synchronous 0-based byte index, or null if not found).
    • In Browser / WebWorker: Promise<number | null> (resolves to 0-based byte index, or null if not found).

init(moduleOrPath?)

function init(moduleOrPath?: any): Promise<void>;

Asynchronously pre-initializes the WebAssembly module backend for Browser / WebWorker environments. In native Node.js environments (isNative === true), this is a no-op that resolves immediately.

  • Parameters:
    • moduleOrPath (optional): WebAssembly.Module | Response | ArrayBuffer | string — Custom WASM source or URL.

initSync(bytesOrModule?)

function initSync(bytesOrModule?: any): void;

Synchronously initializes the WebAssembly module backend with pre-compiled bytes or a WebAssembly.Module. In native Node.js environments, this is a no-op.

  • Parameters:
    • bytesOrModule (optional): ArrayBuffer | Uint8Array | WebAssembly.Module — Pre-loaded WASM binary or module.

isNative

const isNative: boolean;

Boolean flag indicating whether @pid7/ashwa is executing via native N-API bindings (true) or WebAssembly SIMD (false).

Benchmarks

Benchmarks are evaluated across dedicated AWS EC2 hardware environments on Node.js v22.23.2,

  • x86_64 (x64)
    • Instance: Intel(R) Xeon(R) Platinum 8488C (8C/16T)
    • ISA: AVX-512BW · WASM SIMD128
    • Cache: L1d: 384 KiB · L1i: 256 KiB · L2: 16 MiB · L3: 105 MiB
    • STREAM Triad (Read + Write): 25.76 GiB/s
    • Max Single-Core Read Bandwidth: ~12.5 GiB/s
  • AArch64 (arm64)
    • Instance: AWS Graviton3 ARM Neoverse-V1 (16C/16T)
    • ISA: NEON
    • Cache: L1d: 1 MiB · L1i: 1 MiB · L2: 16 MiB · L3: 32 MiB
    • STREAM Triad (Read + Write): 75.48 GiB/s
    • Max Single-Core Read Bandwidth: ~25.5 GiB/s

All benchmarks execute while pinned to a single physical core (mostly to 0th core). For out-of-cache RAM-bound payloads (>256 MiB), every search routine fully saturates the hardwares maximum single-core memory bandwidth ceiling (~12 GiB/s on x64, ~25 GiB/s on arm64). Execution is 100% memory-bandwidth bound rather than compute-bound searching data as fast as the physical memory bus can deliver bytes into the CPU registers.

searchOne

Native Node.js (V8 N-API)

| Level | Payload | Latency (x64) | Latency (arm64) | Throughput (x64) | Throughput (arm64) | |:-----------|:----------|:--------------|:----------------|:-----------------|:-------------------| | L1 Cache | 32 KiB | 259.88 ns | 730.67 ns | 117.43 GiB/s | 41.77 GiB/s | | L2 Cache | 512 KiB | 6.90 µs | 11.15 µs | 70.76 GiB/s | 43.78 GiB/s | | L3 Cache | 16 MiB | 488.06 µs | 376.89 µs | 32.01 GiB/s | 41.46 GiB/s | | RAM | 256 MiB | 20.19 ms | 9.84 ms | 12.38 GiB/s | 25.40 GiB/s | | RAM | 512 MiB | 41.79 ms | 19.60 ms | 11.96 GiB/s | 25.52 GiB/s | | RAM | 1 GiB | 83.68 ms | 39.11 ms | 11.95 GiB/s | 25.57 GiB/s |

WebAssembly (WASM SIMD128)

| Level | Payload | Latency (x64) | Throughput (x64) | |:-----------|:----------|:--------------|:-----------------| | L1 Cache | 32 KiB | 1.28 µs | 23.88 GiB/s | | L2 Cache | 512 KiB | 20.77 µs | 23.51 GiB/s | | L3 Cache | 16 MiB | 1.53 ms | 10.22 GiB/s | | RAM | 256 MiB | 48.19 ms | 5.19 GiB/s | | RAM | 512 MiB | 96.64 ms | 5.17 GiB/s | | RAM | 1 GiB | 192.52 ms | 5.19 GiB/s |

searchTwo

Native Node.js (V8 N-API)

| Level | Payload | Latency (x64) | Latency (arm64) | Throughput (x64) | Throughput (arm64) | |:-----------|:----------|:--------------|:----------------|:-----------------|:-------------------| | L1 Cache | 32 KiB | 563.60 ns | 1.57 µs | 54.15 GiB/s | 19.38 GiB/s | | L2 Cache | 512 KiB | 12.14 µs | 23.93 µs | 40.22 GiB/s | 20.40 GiB/s | | L3 Cache | 16 MiB | 558.19 µs | 771.07 µs | 27.99 GiB/s | 20.26 GiB/s | | RAM | 256 MiB | 20.05 ms | 13.03 ms | 12.47 GiB/s | 19.18 GiB/s | | RAM | 512 MiB | 41.59 ms | 26.07 ms | 12.02 GiB/s | 19.18 GiB/s | | RAM | 1 GiB | 83.70 ms | 52.08 ms | 11.95 GiB/s | 19.20 GiB/s |

WebAssembly (WASM SIMD128)

| Level | Payload | Latency (x64) | Throughput (x64) | |:-----------|:----------|:--------------|:-----------------| | L1 Cache | 32 KiB | 1.87 µs | 16.31 GiB/s | | L2 Cache | 512 KiB | 28.72 µs | 17.00 GiB/s | | L3 Cache | 16 MiB | 1.80 ms | 8.69 GiB/s | | RAM | 256 MiB | 53.44 ms | 4.68 GiB/s | | RAM | 512 MiB | 106.71 ms | 4.69 GiB/s | | RAM | 1 GiB | 213.08 ms | 4.69 GiB/s |

searchThree

Native Node.js (V8 N-API)

| Level | Payload | Latency (x64) | Latency (arm64) | Throughput (x64) | Throughput (arm64) | |:-----------|:----------|:--------------|:----------------|:-----------------|:-------------------| | L1 Cache | 32 KiB | 600.64 ns | 2.26 µs | 50.81 GiB/s | 13.50 GiB/s | | L2 Cache | 512 KiB | 9.56 µs | 34.96 µs | 51.08 GiB/s | 13.97 GiB/s | | L3 Cache | 16 MiB | 487.42 µs | 1.13 ms | 32.06 GiB/s | 13.86 GiB/s | | RAM | 256 MiB | 18.66 ms | 17.96 ms | 13.40 GiB/s | 13.92 GiB/s | | RAM | 512 MiB | 44.00 ms | 35.90 ms | 11.36 GiB/s | 13.93 GiB/s | | RAM | 1 GiB | 87.98 ms | 71.86 ms | 11.37 GiB/s | 13.92 GiB/s |

WebAssembly (WASM SIMD128)

| Level | Payload | Latency (x64) | Throughput (x64) | |:-----------|:----------|:--------------|:-----------------| | L1 Cache | 32 KiB | 2.30 µs | 13.26 GiB/s | | L2 Cache | 512 KiB | 34.11 µs | 14.31 GiB/s | | L3 Cache | 16 MiB | 1.82 ms | 8.58 GiB/s | | RAM | 256 MiB | 54.99 ms | 4.55 GiB/s | | RAM | 512 MiB | 114.54 ms | 4.37 GiB/s | | RAM | 1 GiB | 227.27 ms | 4.40 GiB/s |

searchN (where len(N) = 8)

Native Node.js (V8 N-API)

| Level | Payload | Latency (x64) | Latency (arm64) | Throughput (x64) | Throughput (arm64) | |:-----------|:----------|:--------------|:----------------|:-----------------|:-------------------| | L1 Cache | 32 KiB | 940.14 ns | 2.62 µs | 32.46 GiB/s | 11.63 GiB/s | | L2 Cache | 512 KiB | 15.53 µs | 38.68 µs | 31.44 GiB/s | 12.62 GiB/s | | L3 Cache | 16 MiB | 590.67 µs | 1.23 ms | 26.45 GiB/s | 12.69 GiB/s | | RAM | 256 MiB | 21.28 ms | 19.64 ms | 11.75 GiB/s | 12.73 GiB/s | | RAM | 512 MiB | 44.52 ms | 39.02 ms | 11.23 GiB/s | 12.82 GiB/s | | RAM | 1 GiB | 89.46 ms | 78.20 ms | 11.18 GiB/s | 12.79 GiB/s |

WebAssembly (WASM SIMD128)

| Level | Payload | Latency (x64) | Throughput (x64) | |:-----------|:----------|:--------------|:-----------------| | L1 Cache | 32 KiB | 2.64 µs | 11.54 GiB/s | | L2 Cache | 512 KiB | 39.46 µs | 12.37 GiB/s | | L3 Cache | 16 MiB | 2.00 ms | 7.82 GiB/s | | RAM | 256 MiB | 56.92 ms | 4.39 GiB/s | | RAM | 512 MiB | 114.60 ms | 4.36 GiB/s | | RAM | 1 GiB | 228.40 ms | 4.38 GiB/s |