#126396 – Fix #126394: ffmpeg on win64 is built without SIMD optimizations – blender

Since ee1b2f53cc the ffmpeg libraries for Windows x64 are built
effectively without CPU specific SIMD optimizations.
--arch=x64 is not an architectures that ffmpeg configure understands,
so it falls back to “nothing is known, turn any architecture specific
bits off” code path.

This will require actually rebuilding windows_x64 ffmpeg libraries to
take effect. Once that is done, e.g. on Ryzen 5950X av_get_cpu_flags()
will start to return all the bits up and including AVX2, instead of
zero bits.

Jesse on chat also says this, which makes sense and someone should maybe do it:

Need to add a test for this. Maybe put the cpu flag checks in the same test that checks that we’re enabling the right codecs for ffmpeg

Since ee1b2f53cc the ffmpeg libraries for Windows x64 are built
effectively without CPU specific SIMD optimizations.
`–arch=x64` is not an architectures that ffmpeg configure understands,
so it falls back to “nothing is known, turn any architecture specific
bits off” code path.

This will require actually rebuilding windows_x64 ffmpeg libraries to
take effect. Once that is done, e.g. on Ryzen 5950X `av_get_cpu_flags()`
will start to return all the bits up and including AVX2, instead of
zero bits.

Jesse on chat also says this, which makes sense and someone should maybe do it:

> Need to add a test for this. Maybe put the cpu flag checks in the same test that checks that we’re enabling the right codecs for ffmpeg