michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 7.1.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
pross has joined #ffmpeg-devel
thilo has quit [Ping timeout: 248 seconds]
thilo has joined #ffmpeg-devel
cone-553 has quit [Quit: transmission timeout]
Anthony_ZO has joined #ffmpeg-devel
Kei_N has joined #ffmpeg-devel
Kei_N_ has quit [Ping timeout: 248 seconds]
Traneptora has quit [Quit: Quit]
Traneptora has joined #ffmpeg-devel
^Neo has quit [Ping timeout: 260 seconds]
minimal has quit [Quit: Leaving]
k777__ has quit [Remote host closed the connection]
Quackdoc has joined #ffmpeg-devel
Quackdoc has quit [Client Quit]
jamrial has quit []
_whitelogger has joined #ffmpeg-devel
cone-481 has joined #ffmpeg-devel
<cone-481>
ffmpeg Lynne master:7b0156201bf4: vulkan: fix logging level when erroring upon creating shader module
<haasn>
ramiro: do you think it's a good idea to delete SWS_OP_SWAP_BYTES in favor of a more generic SWS_OP_SHUFFLE taking an arbitrary 128 bit byte-wise shuffle mask?
<haasn>
for x86 this would make a lot of sense but I dunno about other platforms, especially ARM
<haasn>
seems like the standard implementation of bswap on rvv is using a vrgather instruction, so basically the same perf
<haasn>
same situation as x86 I mean
<haasn>
and with VLEN >= 128 being the de-facto "standard", the same 16 byte limit seems appropriate
<haasn>
neon has dedicated instructions though, VREV32 etc
<haasn>
I guess the best compromise is to implement both SWS_OP_SWAP_BYTES and SWS_OP_SHUFFLE, and let implementations choose to compile one as the other
<haasn>
but I still want to introduce an SWS_OP_SHUFFLE since it will allow the optimizer to short circuit read_packed + swizzle + write_packed as read_planar + shuffle + write_planar
<haasn>
hmm, maybe as a compromise, we can just add an explicit order array to SWS_OP_SWAP_BYTES (and maybe rename it to SHUFFLE_BYTES) + add the restriction that it can only shuffle within the pixel type being operated on
<haasn>
that way, SWAP_BYTES { .type = U32 } becomes SHUFFLE_BYTES { .type = U32, .order = { 3, 2, 1, 0 }} and everything else remains the same, but we can still promote byte-level shuffles to OP_SHUFFLE_BYTES
<haasn>
I think I like that compromise the best
mark4o has joined #ffmpeg-devel
markh has quit [Ping timeout: 268 seconds]
mark4o is now known as markh
kasper93 has quit [Quit: kasper93]
kasper93 has joined #ffmpeg-devel
abdu has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
mkver has quit [Ping timeout: 265 seconds]
mkver has joined #ffmpeg-devel
minimal has joined #ffmpeg-devel
<jamrial>
mkver: maybe ff_progress_frame_get_buffer() should have an assert() to ensure the frame wasn't previously allocated
<mkver>
ff_progress_frame_alloc() already has it.
<jamrial>
ah
abdu54 has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
kasper93 has quit [Ping timeout: 272 seconds]
kasper93 has joined #ffmpeg-devel
kasper93_ has joined #ffmpeg-devel
kasper93 has quit [Killed (tantalum.libera.chat (Nickname regained by services))]
kasper93_ is now known as kasper93
<JEEB>
hmm, > src/libavutil/hwcontext_vulkan.c:82:5: error: unknown type name ‘VkPhysicalDeviceShaderExpectAssumeFeatures’; did you mean ‘VkPhysicalDeviceShaderExpectAssumeFeaturesKHR’
<JEEB>
so vulkan configure check passed, but this failed :D
<JEEB>
will try to clear my build root
kasper93_ has joined #ffmpeg-devel
kasper93__ has joined #ffmpeg-devel
<JEEB>
yeh, with a cleared build root still fails
kasper93 is now known as Guest3656
kasper93__ is now known as kasper93
Guest3656 has quit [Ping timeout: 252 seconds]
<JEEB>
OK, I had vulkan and vk_video dirs in my custom sysroot from feb 2024 so that might be the cause (configure check catching the one in base sysroot)
<JEEB>
hmm
kasper93_ has quit [Ping timeout: 252 seconds]
<JEEB>
but if it was the include order since the pkg-config check fails it should fail the header based check
<JEEB>
OK, now cleaned up the vulkan stuff in my additional-deps sysroot
<JEEB>
yeh, now it builds.
Sean_McG has joined #ffmpeg-devel
abdu54 has quit [Ping timeout: 240 seconds]
abdu has joined #ffmpeg-devel
Taripe is now known as ePirat
amogh has quit [Quit: Connection closed for inactivity]