variadic macros not supported in OpenCL

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi,

anything I can do about this? (except running gcc prepoc first to expand everything)?
System is Windows 10, driver is 388.71.
It seems to be supported in other combinations of driver and OS.

Regards
Dirk

2

Does the opencl spec actually advertise support?

This question seems to suggest that at least through OpenCL 2.0 it is “non-standard” and when it works, it’s a function of the platform and config to support it:

[url]Variadic Macros in OpenCL: Where are they supported? - Stack Overflow

and this:

[url]opencl - Why variadic macros are not supported by the standard? - Stack Overflow

3

No - it is not standard, I know - but as the first stackoverflow thread you posted mentions support in Nvidia based platforms (Redhat Enterprise Linux 5, Nvidia CUDA SDK 4.0, OpenCL 1.0 CUDA, driver: 260.19.26), I raised the question here…

Regards
Dirk

4

you can try other opencl implementations. at least on windows, nvidia opencl implementation compiles to load of generic opencl driver which is compatible with opencl implementations from any vendor. so, you can try opencl implementation by amd, or for cpus