skills.rest
VMAFx
VMAFxOFFICIAL
Explore all skills by this author →
WebsiteGitHub

add-simd-path

Scaffolds SIMD implementations for features with runtime dispatch and bit-exact tests.

3|1|Updated May 28, 2026
Search Tags:#dispatch#neon#simd#intrinsics#avx2#feature scaffolding#bit-exact
One-click install
npx skills add https://github.com/VMAFx/vmafx --skill add-simd-path
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: add-simd-path
Source: https://github.com/VMAFx/vmafx/tree/main/.claude/skills/add-simd-path
Command: npx skills add https://github.com/VMAFx/vmafx --skill add-simd-path

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill removes the manual boilerplate involved in adding a new SIMD implementation for an existing feature, while preserving correctness through a bit-exact validation workflow.

Core Features & Use Cases

  • SIMD Scaffold Generation: Creates the intrinsics source and header for a new ISA-specific implementation.
  • Dispatch Integration: Wires the new path into runtime feature selection so the optimized code is actually used.
  • Bit-Exact Verification: Adds a comparison test that checks the SIMD result against the scalar reference implementation.
  • Use Case: A developer can use it to add an AVX2 or NEON path for a feature like SSIM or VIF without manually assembling the file structure, dispatch hooks, and test harness.

Quick Start

Ask the skill to scaffold a new AVX2 SIMD path for an existing feature and include the matching bit-exact test.

Frequently Asked Questions about add-simd-path

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add an AVX2 SIMD path to an existing feature extractor without breaking scalar compatibility?▼

Scaffolding a SIMD path generates the intrinsics source and header, wires runtime dispatch integration, and adds a bit-exact comparison test against the scalar reference implementation to verify correctness.

What does bit-exact comparison testing mean for SIMD intrinsics?▼

Bit-exact comparison testing validates that vectorized SIMD output matches the scalar reference implementation exactly, ensuring optimized paths introduce no numerical drift across kernel-spec and tail-handling scenarios.

Can I generate NEON SIMD paths for ARM64 feature extractors?▼

Yes, you can scaffold NEON SIMD paths for ARM64 feature extractors, alongside x86 targets like AVX2, AVX-512, and AVX-512ICL, complete with intrinsics files and runtime dispatch.

What is the best way to integrate SIMD runtime dispatch into a feature extractor?▼

The best way to integrate SIMD runtime dispatch is by scaffolding ISA-specific intrinsics files and automatically wiring the new path into the runtime feature selection logic for the feature extractor.

Does scaffolding a SIMD path include support for kernel-spec and tail-handling?▼

Yes, scaffolding a SIMD path includes bit-exact comparison tests configured with kernel-spec and tail-handling support to manage data boundaries during vectorized processing.

Related Skills

Explore More Software Engineering Skills→
wimbleroboticswimblerobotics
18

anomaly-detection-vision

Compare camera frames to a stored baseline to detect scene changes.

Community
Advanced
gaojesse999gaojesse999
1

ceva-dsp-programming

Guide CEVA DSP kernel development with intrinsics, SIMD, and mode-bit configuration.

Community
Advanced
rust-worksrust-works
51

simd-optimization

Optimize SIMD code paths for x86_64 AVX2/AVX-512 and ARM NEON.

Official
Advanced
qualcommqualcomm
2

arm64-inlineasm-to-intrinsics

Convert ARM64 NEON inline assembly to C/C++ intrinsics with verification.

Official
Advanced