D2D

GGitHub[BugFix] msprobe data collection support aclgraph (#8574)

eb27f866创建于 16 天前历史提交

文件	最后提交记录	最后更新时间
index.md	[Core][Misc] Clean up ProfileExecuteDuration (#6461) ### What this PR does / why we need it? This PR removes the custom ProfileExecuteDuration utility and its usages across the codebase. This utility was used for profiling execution duration of different stages in the inference process. It is replaced by the standard vllm.v1.utils.record_function_or_nullcontext, which integrates with PyTorch's profiler. This change simplifies the code by removing a custom implementation in favor of an upstream utility, improving maintainability. Associated documentation and tests for ProfileExecuteDuration are also removed. ### Does this PR introduce _any_ user-facing change? VLLM_ASCEND_MODEL_EXECUTE_TIME_OBSERVE env is removed now. ### How was this patch tested? CI passed. The changes are a cleanup and replacement with a standard utility. Existing tests cover the functionality. The removed feature had its own tests which are also removed. Related RFC: #5304 - vLLM version: v0.14.1 - vLLM main: https://github.com/vllm-project/vllm/commit/dc917cceb877dfd13f98c538c4c96158047d98bd Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>	3 个月前
msprobe_guide.md	[BugFix] msprobe data collection support aclgraph (#8574) ### What this PR does / why we need it? This PR fixes and clarifies msprobe dump behavior for Ascend graph mode, with two goals: 1. **Avoid dumping dummy-run data** - In model_runner_v1.py, dummy_run now finalizes debugger state with dump=False, so warmup/dummy paths do not write dump data to disk. 2. **Keep eager/graph debugger invocation compatible** - _finalize_dump_data now forwards kwargs to self.debugger.step(**kwargs). - This keeps compatibility with both: - PrecisionDebugger.step() (eager path) - AclGraphDumper.step(dump=...) (graph path) 3. **Docs alignment for graph-mode config support** - Updated msprobe_guide.md support table to reflect graph-mode constraints: - task: graph mode supports statistics - step: graph mode marked unsupported (×) - unified table markers with √/× - Removed extra explanatory paragraph and moved constraints into the table itself for clarity. --------- Signed-off-by: Tjh-UKN <2559659915@qq.com> Co-authored-by: Yizhou <136800916+yiz-liu@users.noreply.github.com>	16 天前
optimization_and_tuning.md	[Doc] correct errors in docs (#8923) ### What this PR does / why we need it? This PR corrects several errors in the documentation to ensure accuracy and prevent execution failures. Key changes include: - Fixing trailing backslashes in shell command examples that would cause syntax errors when copy-pasted. - Correcting the master node IP variable in the multi-node disaggregated prefill-decode deployment example. - Standardizing the capitalization of "vLLM-Ascend" for consistency. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Documentation updates only. --------- Signed-off-by: zouyida2052 <zouyida2002@gmail.com>	22 天前
performance_benchmark.md	[Doc] Fix documentation formatting and improve code examples (#8660) ### What this PR does / why we need it? This PR fixes various documentation issues and improves code examples throughout the project. - vLLM version: v0.19.0 - vLLM main: https://github.com/vllm-project/vllm/commit/6f786f2c506cb07f4566771fdc62e640e2c4a176 --------- Signed-off-by: MrZ20 <2609716663@qq.com>	1 个月前
service_profiling_guide.md	[Doc][Misc] Improve readability and fix typos in documentation (#8266) ### What this PR does / why we need it? This PR improves the readability of the documentation by fixing typos, correcting command extensions, and fixing broken links in the Chinese README. ### Does this PR introduce _any_ user-facing change? No. ### How was this patch tested? Documentation changes only. - vLLM version: - vLLM main: https://github.com/vllm-project/vllm/commit/v0.19.0 --------- Signed-off-by: sunshine202600 <sunshine202600@163.com>	1 个月前