AGENT SKILL

gpu-compute

Engineer parallel GPU kernel dispatches, shared memory reduction, thread divergence optimization, and atomics.

License 许可
MIT
最近更新
2026-11-09
Stars
★ 0
Forks
0

01 前言 Front Matter

name
gpu-compute
risk
low
tags
[ "graphics", "gpu-compute", "cuda", "compute-shader", "parallel" ]
author
AAS Platform
source
authoring
version
1.0.0
category
graphics
description
Engineer parallel GPU kernel dispatches, shared memory reduction, thread divergence optimization, and atomics.
disable-model-invocation
false

03 SKILL.md 正文

General-Purpose GPU Compute (CUDA, OpenCL, Compute Shaders)

Overview & Core Principles

Engineer parallel GPU kernel dispatches, shared memory reduction, thread divergence optimization, and atomics.

Maximize global memory coalescing and optimize thread block dimensions (multiples of 32/64 threads).

04 同仓库更多 Skills