Enables a way to overcome the problem where some OpenMP* offload SPIR-V* devices produce incorrect code when a target object is larger than 4GB.
Linux: | -fopenmp-target-buffers=keyword |
Windows: | /Qopenmp-target-buffers:keyword |
keyword |
Possible values are:
|
default |
If you do not specify this option, the compiler may produce incorrect code on some OpenMP* offload SPIR-V* devices when a target object is larger than 4GB. |
This option enables a way to overcome the problem where some OpenMP* offload SPIR-V* devices produce incorrect code when a target object is larger than 4GB (4294959104 bytes).
However, note that when -fopenmp-target-buffers=4GB (or /Qopenmp-target-buffers:4GB) is specified on Intel® GPUs, there may be a decrease in performance.
To use this option, you must also specify option -fopenmp-targets (Linux*) or /Qopenmp-targets (Windows*).
This option may have no effect for some OpenMP* offload SPIR-V* devices, and for OpenMP* offload targets different from SPIR*.
Visual Studio: DPC++ > Language > Specify buffer size for OpenMP offload kernel access limitations (DPC++)
Visual Studio: C/C++ > Language [Intel C++] > Specify buffer size for OpenMP offload kernel access limitations (C++)
Eclipse: Intel® oneAPI DPC++ Compiler > Language > Specify buffer size for OpenMP offload kernel access limitations (DPC++)
Eclipse: Intel C++ Compiler > Language > Specify buffer size for OpenMP offload kernel access limitations (C++)
None