Add competition operators for KernelSwift operator innovation competition - #104
Open
logicore-code wants to merge 2 commits into
Open
Add competition operators for KernelSwift operator innovation competition#104logicore-code wants to merge 2 commits into
logicore-code wants to merge 2 commits into
Conversation
added 2 commits
August 6, 2026 23:45
…tion Implement four high-performance LLM inference operators: - MXFP4 W4A16 grouped expert matrix multiplication (T1) - Block-scaled FP8 matrix multiplication (T2) - Gated RMSNorm fusion operator (T3) - MLA RoPE and compressed KV Cache fusion (T4) Team: Chongli Members: Zeyu Jiang, Guangrong Jiang All operators support Hygon and Iluvatar platforms. pytest output: tests pass locally
Detailed technical report covering all four competition problems: - Optimization strategies and implementation details - Performance analysis and expected speedups - Ablation studies and design decisions Team: Chongli (Zeyu Jiang, Guangrong Jiang)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
KernelSwift 算子创新大赛参赛作品
KernelSwift 算子创新大赛
参赛队伍: 崇理
队长: 蒋泽宇 (jiangzeyu-2026)
队员: 蒋光荣
所选赛题: T1, T2, T3, T4
修改内容与动机
本PR提交了四个高性能LLM推理算子的完整实现：
T1: MXFP4 W4A16分组专家矩阵乘
T2: Block-scaled FP8矩阵乘
T3: Gated RMSNorm融合算子
T4: MLA RoPE与压缩KV Cache写入融合
方案与实现
所有算子基于九齿(NineToothed) DSL实现，遵循arrange-and-apply编程范式。
关键优化策略：
工程适用边界
测试
pytest output: All 21 tests passed
技术报告
详见
docs/COMPETITION_REPORT_CHONGLI.md参赛队伍: 崇理
成员: 蒋泽宇, 蒋光荣
日期: 2026年8月6日