Beignet offers a far more flexiable and complete interface, so choose it by default if available. --- On 23/03/2019 12:27, Mark Thompson wrote: > On 22/03/2019 01:40, Ruiling Song wrote: >> ffmpeg | branch: master | Ruiling Song <ruiling.s...@intel.com> | Fri Nov 23 >> 13:39:12 2018 +0800| [61cb505d18b8a335bd118d88c05b9daf40eb5f9b] | committer: >> Ruiling Song >> >> lavu/opencl: replace va_ext.h with standard name >> >> Khronos OpenCL header (https://github.com/KhronosGroup/OpenCL-Headers) >> uses cl_va_api_media_sharing_intel.h. And Intel's official OpenCL driver >> for Intel GPU (https://github.com/intel/compute-runtime) was compiled >> against Khronos OpenCL header. So it's better to align with Khronos. >> >> Signed-off-by: Ruiling Song <ruiling.s...@intel.com> >> >>> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=61cb505d18b8a335bd118d88c05b9daf40eb5f9b >> --- >> >> configure | 2 +- >> libavutil/hwcontext_opencl.c | 2 +- >> 2 files changed, 2 insertions(+), 2 deletions(-) >> >> diff --git a/configure b/configure >> index a817479559..331393f8d5 100755 >> --- a/configure >> +++ b/configure >> @@ -6472,7 +6472,7 @@ fi >> >> if enabled_all opencl vaapi ; then >> enabled opencl_drm_beignet && enable opencl_vaapi_beignet >> - check_type "CL/cl.h CL/va_ext.h" >> "clCreateFromVA_APIMediaSurfaceINTEL_fn" && >> + check_type "CL/cl.h CL/cl_va_api_media_sharing_intel.h" >> "clCreateFromVA_APIMediaSurfaceINTEL_fn" && >> enable opencl_vaapi_intel_media >> fi >> >> diff --git a/libavutil/hwcontext_opencl.c b/libavutil/hwcontext_opencl.c >> index d3df6221c4..b116c5b708 100644 >> --- a/libavutil/hwcontext_opencl.c >> +++ b/libavutil/hwcontext_opencl.c >> @@ -50,7 +50,7 @@ >> #include <mfx/mfxstructures.h> >> #endif >> #include <va/va.h> >> -#include <CL/va_ext.h> >> +#include <CL/cl_va_api_media_sharing_intel.h> >> #include "hwcontext_vaapi.h" >> #endif >> > > This broke the build when both are available. > > $ make > CC libavutil/hwcontext_opencl.o > src/libavutil/hwcontext_opencl.c: In function ‘opencl_device_derive’: > src/libavutil/hwcontext_opencl.c:1236:5: error: duplicate case value > case AV_HWDEVICE_TYPE_VAAPI: > ^~~~ > src/libavutil/hwcontext_opencl.c:1205:5: note: previously used here > case AV_HWDEVICE_TYPE_VAAPI: > ^~~~ > src/libavutil/hwcontext_opencl.c: In function ‘opencl_map_to’: > src/libavutil/hwcontext_opencl.c:2831:5: error: duplicate case value > case AV_PIX_FMT_VAAPI: > ^~~~ > src/libavutil/hwcontext_opencl.c:2825:5: note: previously used here > case AV_PIX_FMT_VAAPI: > ^~~~ > src/libavutil/hwcontext_opencl.c: In function ‘opencl_frames_derive_to’: > src/libavutil/hwcontext_opencl.c:2873:5: error: duplicate case value > case AV_HWDEVICE_TYPE_VAAPI: > ^~~~ > src/libavutil/hwcontext_opencl.c:2866:5: note: previously used here > case AV_HWDEVICE_TYPE_VAAPI: > ^~~~ > make: *** [ffbuild/common.mak:60: libavutil/hwcontext_opencl.o] Error 1 > make: Target 'all' not remade because of errors. > > $ cat config.h | grep HAVE_OPENCL > #define HAVE_OPENCL_D3D11 0 > #define HAVE_OPENCL_DRM_ARM 0 > #define HAVE_OPENCL_DRM_BEIGNET 1 > #define HAVE_OPENCL_DXVA2 0 > #define HAVE_OPENCL_VAAPI_BEIGNET 1 > #define HAVE_OPENCL_VAAPI_INTEL_MEDIA 1 > > > I think in general the Beignet mapping is more useful if present since it has > far fewer constraints, so perhaps disable this one if Beignet is there?
configure | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/configure b/configure index 331393f8d5..c94f516224 100755 --- a/configure +++ b/configure @@ -6471,9 +6471,12 @@ if enabled_all opencl libdrm ; then fi if enabled_all opencl vaapi ; then - enabled opencl_drm_beignet && enable opencl_vaapi_beignet - check_type "CL/cl.h CL/cl_va_api_media_sharing_intel.h" "clCreateFromVA_APIMediaSurfaceINTEL_fn" && - enable opencl_vaapi_intel_media + if enabled opencl_drm_beignet ; then + enable opencl_vaapi_beignet + else + check_type "CL/cl.h CL/cl_va_api_media_sharing_intel.h" "clCreateFromVA_APIMediaSurfaceINTEL_fn" && + enable opencl_vaapi_intel_media + fi fi if enabled_all opencl dxva2 ; then -- 2.19.2 _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
