Would it be fixed? It seems that initialization code should be arranged as singleton, that is not the case currently.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| nvcc (nvlink) not linking against device code library | 7 | 11586 | June 20, 2018 | |
| Linking device code | 13 | 7692 | December 8, 2014 | |
| Cmake: kernel fail to launch for the separate compilation of binary and library | 4 | 339 | August 7, 2023 | |
| Issue with cudaMemcpyToSymbol and Separable Compilation. | 10 | 1438 | February 28, 2019 | |
| Link failure with Thrust and separate compilation | 4 | 2396 | January 21, 2018 | |
| Linking multiple static cuda libs | 1 | 2195 | August 26, 2020 | |
| How to link host code with a static CUDA library after separable compilation? | 2 | 7802 | April 30, 2013 | |
| Linking Device functions from static libraries with CMake | 3 | 422 | January 7, 2025 | |
| pxtas can't find symbol, but it's definitely in the object file? | 2 | 1060 | February 5, 2016 | |
| Multiple definition of __cudaRegisterLinkedBinary_* error with relocatable utilities | 0 | 998 | September 11, 2022 |