Hello,
I didn’t want to create a bug report for that, so I start discussion here.
Intel has recently created “intel-opencl” which allows Intel GPU to use advance OpenCL on Linux (at least) - it’s open source.
Note: It works great with LuxRender (10x faster than CPU on my computer).
I checked in source code to verify if it can work with Cycles. But the platform name is not complete.
First, I changed that line in opencl_split.cpp:
else if (platform_name == "Intel(R) OpenCL" || platform_name == "Intel(R) OpenCL HD Graphics") {
And I use “force_all” boolean to force my card. Of course it fails as explained in the comment “it segfaults on linux currently”
The only thing I want to say is that platform_name is not “Intel ® OpenCL” only. My line is the one to used to check both platform name.
Should I create a bug report ? or a patch ?
It doesn’t fix anything… only the platform name string is fixed