skills.rest
xchang1121
xchang1121COMMUNITY
Explore all skills by this author →
GitHub

triton-ascend-reduce

Optimize multi-axis reduce operators on Ascend with two-stage Triton kernels.

6|1|Updated Apr 19, 2026
Search Tags:#optimization#normalization#kernel#triton#reduce#ascend#multi-axis
One-click install
npx skills add https://github.com/xchang1121/AutoResearch-CC-hook --skill triton-ascend-reduce
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: triton-ascend-reduce
Source: https://github.com/xchang1121/AutoResearch-CC-hook/tree/main/skills/triton-ascend/guides/triton-ascend-reduce
Command: npx skills add https://github.com/xchang1121/AutoResearch-CC-hook --skill triton-ascend-reduce

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Optimize reduce-type and composite operators that involve reductions, including multi-axis reductions and normalization scenarios, to improve performance on Ascend using Triton.

Core Features & Use Cases

  • Supports non-final-dimension reductions with correct multi-dimensional indexing to avoid costly reshapes.
  • Describes two-stage reduction workflows for complex operators like normalization (layernorm, rmsnorm, groupnorm, batchnorm) and statistical computations (variance, std).
  • Provides practical guidance for implementing efficient Triton kernels on Ascend hardware, including axis-aware tiling and atomic reductions.

Quick Start

Start by identifying a non-final axis reduction and implement a two-stage Triton kernel on Ascend to validate performance gains.

Frequently Asked Questions about triton-ascend-reduce

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

FAQPage Schema
How do I optimize multi-axis reductions in Triton when the reduced axis is not the last dimension?▼

To optimize multi-axis reductions across non-final dimensions, apply correct multi-dimensional indexing within your Triton kernel to avoid costly reshapes and improve execution efficiency.

What is the best way to implement two-stage reduction workflows for normalization on Ascend hardware?▼

Implementing two-stage reduction workflows for normalization involves splitting operations like layernorm or rmsnorm into separate passes, calculating statistics and normalizing outputs to maximize hardware utilization.

Does the triton_ascend DSL support Atlas A2 and A3 hardware for reduce-type operations?▼

Yes, the triton_ascend DSL supports Atlas A2 and A3 hardware, providing specific backend optimizations and axis-aware tiling for efficient reduce-type operations.

How do I handle statistical computations like variance and standard deviation efficiently using Triton kernels?▼

Statistical computations like variance and standard deviation require multi-stage reduction strategies, calculating intermediate means before finalizing the statistics within the Triton kernel to ensure accurate results.

When should I use axis-aware tiling and atomic reductions for Triton kernel optimization?▼

Use axis-aware tiling and atomic reductions when implementing complex multi-dimensional reductions on Ascend hardware, ensuring data is processed efficiently across multiple axes without performance degradation.

Why does reducing non-final dimensions cause performance issues in standard Triton kernels?▼

Reducing non-final dimensions causes performance issues because standard kernels often require costly reshapes to align data, a problem solved by using correct multi-dimensional indexing and axis-aware tiling.

Related Skills

Explore More Software Engineering Skills→
xchang1121xchang1121
6

triton-ascend-example-double-kernel

Execute two Triton Ascend kernels sequentially in forward for staged computation.

Community
Intermediate
moirai-internalmoirai-internal

diffusion-kernel

Integrate optimized CUDA and Triton kernels for diffusion model inference.

Official
Advanced
mindspore-aimindspore-ai
258

triton-ascend-case-reduction-amin-atomic

Optimize atomic amin reduction on Ascend by mapping the reduction axis across multiple cores.

Official
Advanced
xchang1121xchang1121
6

triton-ascend-example-layernorm

Implement a two-stage LayerNorm reduction kernel with Triton Ascend.

Community
Intermediate