It seems all host function launched by cudaLaunchHostFunc within different stream is executed in single thread sequentially. I don’t find any runtime API to configure CUDA to use a thread pool. Anyway to do this?

Hello @SparkHu , Any luck with this ?

Anyway to tell Cuda run host function launched in different stream in multiple threads(such as a threadpool)

Robert_Crovella July 11, 2023, 8:44pm 3

Topic		Replies	Views
cudaLaunchHostFunc API example CUDA Programming and Performance	31	7343	February 8, 2025
Does cudaLaunchHostFunc block work added to all streams? CUDA Programming and Performance	18	1811	April 27, 2021
cudaLaunchHostFunc blocking work on Linux CUDA Programming and Performance cuda	2	788	September 22, 2022
Executing kernel from different host threads CUDA Programming and Performance	1	1838	September 1, 2011
cuLaunchHostFunc Questions CUDA Programming and Performance cuda , kernel , nsight	2	749	March 1, 2024
Using multiple streams with multiple host threads takes longer? stream CUDA Programming and Performance	3	993	February 10, 2021
CUFFT doesn't support multiple host threads each with a separate stream? Profiling Linux Targets cuda	2	106	December 9, 2024
CUDA per-thread and cudnn behaviour CUDA Programming and Performance	1	1351	September 15, 2017
Do CUDA graph host nodes execute on more than one thread? CUDA Programming and Performance	1	509	October 22, 2020
culaunchHostFunc overhead latency usage + CPU->GPU signaling CUDA Programming and Performance	6	357	April 1, 2025