# Should I worry about measure error when auto-tuning multiple tasks simultaneously?

**URL:** <https://discuss.tvm.apache.org/t/should-i-worry-about-measure-error-when-auto-tuning-multiple-tasks-simultaneously/17381>\
**Category:** Uncategorized\
**Created:** [June 22, 2024, 6:53am UTC](https://discuss.tvm.apache.org/t/should-i-worry-about-measure-error-when-auto-tuning-multiple-tasks-simultaneously/17381 "2024-06-22T06:53:47Z")\
**Posts on this page:** 1\
**Showing post:** 5

<div class="post-metadata">

**Author:** ![tsu-bin](https://discuss.tvm.apache.org/user_avatar/discuss.tvm.apache.org/tsu-bin/32/5004_2.png) [@tsu-bin](https://discuss.tvm.apache.org/u/tsu-bin)\
**Post date:** [July 3, 2024, 3:58am UTC](https://discuss.tvm.apache.org/t/should-i-worry-about-measure-error-when-auto-tuning-multiple-tasks-simultaneously/17381/5 "2024-07-03T03:58:40Z")

</div>

Concurrent launching parallel cuda-graph or cuda kernel can definitely cause latency become larger, so results in inaccurate measuring. From my observation, Ansor tuning process is almost cpu bond (please correct me if I’m wrong), the gpu measuring span is only small portion, the most of time is spend on sampling valid schedule from sketch, and this is done by python multiprocessing, by default setting it will utilise all cpu cores, so I doubt is there any benefit to run multiple Ansor process concurrently.

---

_[View the full topic](https://discuss.tvm.apache.org/t/should-i-worry-about-measure-error-when-auto-tuning-multiple-tasks-simultaneously/17381)._