Repository navigation
fix: guard P800 sdpa perf loading - #22
Merged
Merged
Conversation
Collaborator
Author
|
Review 完成,本机(Ascend910 真机)验证通过,LGTM ✅ 已验证:
两个小建议(不阻塞合并,可 follow-up):
CI 三门绿 + 我们平台真机回归绿,可以合并。感谢快速响应 #16 审计线索——这个"perf registry 顺序加载遮蔽"的发现是 #13/#14 同源问题的又一处实例,已记入平台绑定知识。 |
Collaborator
Author
|
最终判定:OK,可以合并 ✅ 结论依据(均为本机 Ascend910 真机实测,非推测):
边界声明:P800 真机 perf 数字(0.151ms / +1.3%)出自作者环境,本机无昆仑芯未复测——但修复不改 kernel 数值语义,该数字非门禁项。 两条不阻塞建议(reload 意图注释 / MLU 复验顺带确认)见上一条评论。合并后建议 follow-up:把这处"perf registry 顺序加载遮蔽"补进 sdpa PLATFORM.md 的绑定清单(与 #13/#14 同源问题的第三处实例)。 |
2 of 7 tasks
Collaborator
Author
|
Follow-up 已补:#23 记录 PLATFORM.md 绑定约束、解释 reload 意图,并把外来顶层 reference 遮蔽回归扩展到 P800/MLU backend。 |
2 of 7 tasks
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
自查
__init__.py(包化入口) / reference / register / example / kernel/ / test/(不新增算子;修复既有 P800 sdpa backend 与 perf 门禁)example.py p800-kunlunxin三层通过)python3 ops/test_packaging.py(全算子导入扫描 16/16)wt-YYYY-MM-DD-fix+# wt <邮箱>,commit 作者相同(小范围工程修复)说明
来自 PR #10 合入后的作者自查:多算子 perf registry 按顺序加载时,早先算子的顶层
reference模块会遮蔽ops.sdpa的同名导入,导致:修复分两层:
ops.sdpaP800/MLU backend 改为包态相对导入优先,保留单算子脚本模式 fallback。perf_run在执行本次选中用例前清理对应旧 JSON;任一用例 SKIP 时直接返回失败,防止perf_compare消费历史成功记录形成假绿。验证