Skip to content

Conversation

@riv19
Copy link

@riv19 riv19 commented Oct 11, 2025

Fix error from issue #27 :

[alarm@localhost WebRip999]$ vspipe -c y4m tmp.vpy tets.y4m                                              
Warning: Failed to load /usr/lib/vapoursynth/libneo-f3kdb.so. Error given: /usr/lib/vapoursynth/libneo-f3kdb.so: undefined symbol: process_plane_impl_sse4_16bit_interleaved

No need for additional range checks, because at f3kdb.hpp:152 the code won't allow the array index to go beyond "C" optimization without detecting them on CPU.
UPD: but CPU capabilities return SSE support on ARM (it is hardcoded)... So that I had to fix it in another commit.

@riv19 riv19 force-pushed the bug/undefined_symbols_non_x86 branch 5 times, most recently from 35a0256 to 9fe8c89 Compare October 11, 2025 13:08
    Warning: Failed to load /usr/lib/vapoursynth/libneo-f3kdb.so. Error
    given: /usr/lib/vapoursynth/libneo-f3kdb.so: undefined symbol:
    process_plane_impl_sse4_16bit_interleaved
@riv19 riv19 force-pushed the bug/undefined_symbols_non_x86 branch 5 times, most recently from fd0cd71 to e794d16 Compare October 12, 2025 12:00
@riv19
Copy link
Author

riv19 commented Oct 12, 2025

Unfortunately this SIMD optimization introduces artifacts on ARM. I had to set "opt=0" in Vapoursynth to disable.
But this is another topic - at least, it is possible to use the plugin on non-x86.
Ready for merge.

Use sse2neon.h for NEON support.
@riv19 riv19 force-pushed the bug/undefined_symbols_non_x86 branch from e794d16 to f46c649 Compare October 12, 2025 12:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant