1 |
On Sun, 2020-04-05 at 22:13 +0100, Marek Szuba wrote: |
2 |
> On 2020-04-02 15:31, Marek Szuba wrote: |
3 |
> |
4 |
> > 3. Test if downstream applications are happy with the new unified OpenCL |
5 |
> > headers required by the Khronos Group ICD loader and if they are, add |
6 |
> > dev-libs/opencl-icd-loader to virtual/opencl as an alternative to ocl-icd; |
7 |
> |
8 |
> Quick update on this: today I have attempted to rebuild and test various |
9 |
> OpenCL-dependent packages using the unified headers and |
10 |
> dev-libs/opencl-icd-loader. The good news is, they have all been |
11 |
> perfectly happy with them. The bad news is, I actually had to manually |
12 |
> symlink all contents |
13 |
> /usr/lib/OpenCL/vendors/opencl-icd-loader/include/CL/ into |
14 |
> /usr/include/CL/ for this to work because eselect-opencl contains a |
15 |
> hard-coded list of expected vendor header files. |
16 |
> |
17 |
> In other words, it will be necessary to either teach eselect-opencl how |
18 |
> to handle the unified headers or make it possible for it to let the |
19 |
> contents of /usr/include/CL be. Personally I am leaning towards the |
20 |
> latter - it doesn't even handle legacy headers that well (it installs |
21 |
> several versions into /usr/lib/OpenCL/global but in the end offers no |
22 |
> way of switching between them), and in any case even its description |
23 |
> suggests its job is to switch between implementations rather than handle |
24 |
> global headers. |
25 |
|
26 |
While at it, is there any chance to rewrite eselect-opencl so that it |
27 |
stops writing into /usr and uses environment approach like eselect- |
28 |
opengl does? |
29 |
|
30 |
-- |
31 |
Best regards, |
32 |
Michał Górny |