【2026夏季训练营】Hadamard CUDA 加速与 INT4 融合 - #69
Open
a962695448-rgb wants to merge 2 commits into
Open
a962695448-rgb wants to merge 2 commits into
a962695448-rgb wants to merge 2 commits into
Conversation
保留核心实现、必要测试、五种国产平台入口、报告及摩尔可再生成的复现材料。 将 2586 份历史 results 记录改为固定归档链接,新增快速开始、证据索引与源码清单。 保留的 51 份源码和构建文件与完整归档字节一致。 CPU reference 检查通过;摩尔 6750 条计时复算及 6 项数据完整性检查通过。 本次仅做提交整理,没有修改 GPU 计算源码或重新运行 GPU 实验。
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
提交 2026 夏季训练营第三题 Hadamard 变换加速项目,代码位于 03_hadamard_tc/a962695448-rgb/。
实现
提交与既有验证
该 PR 保留核心实现、测试与验收文档。2,586 份历史 results 文件保留在固定完整档案,通过链接引用。原有 51 份源码和构建文件与该档案字节一致;本轮新增的执行上下文验证脚本在清单中单独记录。
2026-09-16 RTX 4090 路由候选实验
对提交 4049323 与一个仅减少主机端元数据/设备查询的候选做同机对照:
预先固定的保留门槛是:每轮目标配置耗时下降至少 5%,任何目标/对照配置回退不超过 2%。三轮目标耗时下降分别为 -1.1455%、-0.0448%、+0.6025%,均未达到门槛,因此保留现有实现。
固定实验档案包含原始数据、候选源码、固定协议、验证/分析脚本和取源工具。11 份原始结果回收后逐份 SHA-256 核对通过,本地复算与服务器结果一致。
这里测量带输出分配的普通 Python 接口墙钟时间,目标统计量是未加权几何平均耗时比换算的下降率;不是 kernel event、CUDA Graph 或模型端到端性能。新旧互比和重复计时不扩大独立用例数量;慢例与未采用的候选均保留。
2026-09-16 RTX 4090 D 执行上下文补测
新增测试覆盖三种接口、FP16/BF16、128/256 线程、2D/4D、维度 1/8/64/256,以及合法的 original/auto/packed/contiguous256 组合:
错误对照未被识别即整次失败,重放和错误对照不累计成更多独立配置。计算内核未变;这是单卡正确性补测,不宣称性能提升或新 A100 验证。
补充验收说明提供复现入口;固定原始证据保存逐项结果、负例、源码指纹和环境。