ascendc-evaluation

Name: ascendc-evaluation
Availability: InStock
Author: just-it

Evaluate AscendC custom operators against reference implementations for correctness and performance.

33|51|Updated Mar 12, 2026

Search Tags:#validation #evaluation #reference #benchmark #ascendc #custom_op

Ready to run

npx skills add https://github.com/Just-it/AscendOpGenAgent --skill ascendc-evaluation

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: ascendc-evaluation
Source: https://github.com/Just-it/AscendOpGenAgent/tree/main/skills/ascendc_evalution
Command: npx skills add https://github.com/Just-it/AscendOpGenAgent --skill ascendc-evaluation

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires torch, torch_npu, and includes scripts (resource) components.

What problem does it solve?

This Skill identifies and evaluates AscendC custom operators by validating correctness against reference implementations and measuring performance.

Core Features & Use Cases

Correctness verification: compare Model and ModelNew outputs for parity.
Performance benchmarking: measure latency and memory to compare reference vs custom implementations.
End-to-end workflow: environment setup, script execution, and result reporting for both Triton and AscendC backends.

Quick Start

Run the evaluation workflow to validate a custom AscendC operator against its reference implementation and obtain a performance report.

Frequently Asked Questions about ascendc-evaluation

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

FAQPage Schema

How do I validate AscendC custom operators against reference implementations?▼

To validate AscendC custom operators, run the evaluation workflow to compare Model and ModelNew outputs for parity. It checks correctness by running scripts against reference implementations in your output directory.

How do I benchmark AscendC operator performance for cross-backend comparison?▼

Benchmark AscendC operator performance by measuring latency and memory to compare reference vs custom implementations. The workflow supports cross-backend performance comparison for both Triton and AscendC backends.

Do I need torch_npu to evaluate AscendC custom operators?▼

Yes, you need torch_npu and torch dependencies to evaluate AscendC custom operators. Environment setup scripts are provided to prepare the necessary dependencies for running correctness checks and benchmarks.

What is the required directory structure for AscendC operator evaluation?▼

The required directory structure for AscendC operator evaluation is output/<op_name>/. This directory must contain both reference and custom implementations to execute correctness verification and performance benchmarking.

Can I compare Triton and AscendC backends during operator evaluation?▼

Yes, you can compare Triton and AscendC backends during operator evaluation. The workflow provides end-to-end script execution and result reporting to facilitate cross-backend performance comparison and integration testing.