New versions of free software used in the GUI's

[Jamaika]

Originally Posted by TDS

I thought I'd give this a try, however, using in with a particular app, it failed on several "jobs", on my Ryzen 7950X...not sure if it's the AVX only.

Also, where is ffprobe ??

I think I'll be sticking with the BtbN master builds.

Regards

I know it should be AVX2. AVX512 is only for dav1d/libjxl. I don't know if the latest boards have AVX512 because it was supposed to be rejected.

I don't use Ffprobe. I am not adding it because the zip files are smaller. If anyone is interested, I can add it once.

BtbN is the showcase of ffmpeg. I don't have time to add all the additions.

[TDS]

Originally Posted by Jamaika

I don't use Ffprobe. I am not adding it because the zip files are smaller. If anyone is interested, I can add it once.

May I suggest that you use 7-zip instead of WinZip..

for example, your current .zip is 79,837 Kb, and the same using 7z, is only 53,225 Kb.

[Jamaika]

fmpeg version 7.1.0-1c45104f Copyright (c) 2000-2024 the FFmpeg developers {opencl not default}
built with GCC: (GNU) 11.3.1 20221227 (SIMD) AVX512ICL
configuration: -std=gnu++11 -march=x86-64-v4 -ftree-vectorize -g0 -O3 -fPIC -mgfni -mvaes -mvpclmulqdq -mavx512f -mavx512cd -mavx512dq -mavx512er -mavx512pf -mavx512bw -mavx512vl -mavx512vnni -mavx512ifma -mavx512vbmi -mavx512vbmi2 -mavx512vpopcntdq -mavx512bitalg
libraries: -lavdevice_x64 -lavfilter_x64 -lavformat_x64 -lavcodec_x64 -lswresample_x64 -lswscale_x64 -lavutil_x64 -lpostproc_x64 -ldavs2_x64 -ldav1d_x64 -luavs3d_x64 -lxevd_x64 -laom_x64 -lkvazaar_x64 -lopenh264_x64 -lsvtav1_x64 -ltheora_x64 -lvpx_x64 -luvg266_x64 -luavs3e_x64 -lx264_opencl_0810bit_x64 -lx265_081012bit_x64 -lxavs_x64 -lxavs2_x64 -lxeve_x64 -lxvid_x64 -lcelt_opus_silk_x64 -lcelt_x64 -lcodec2_x64 -lmp3lame_x64 -lgsm_x64 -lvorbis_x64 -lfdk-aac_x64 -lshine_x64 -lsnappy_x64 -lspeex_x64 -ltwolame_x64 -ljxl_x64 -lpng_x64 -lwebp_x64 -lopenjpeg_x64 -llcms2_x64 -logg_x64 -lrsvgdec_x64 -lcairo_pixman_x64 -lass_x64 -l:freetype_harfbuzz_x64.a -l:fribidi_x64.a -lxml2_x64 -lvmaf_x64 -l:zlib_x64.a -l:bzip2_x64.a -l:lzma_x64.a -l: opencl_x64.a -l: opencv2_opencl_x64.a -l:quirc_x64.a -lqrencode_x64
libavutil 59. 15.100 / 59. 15.100
libavcodec 61. 5.102 / 61. 5.102
libavformat 61. 3.100 / 61. 3.100
libavdevice 61. 2.100 / 61. 2.100
libavfilter 10. 2.101 / 10. 2.101
libswscale 8. 2.100 / 8. 2.100
libswresample 5. 2.100 / 5. 2.100
libpostproc 58. 2.100 / 58. 2.100

Latest additions:
aom 3.8.2-a4420e5
dav1d 1.4.1-37d5243
brotli 1.1.0-c1c76e9
jpegxl 0.10.2-1f50b85
libass 0.17.1-d160f8d
glib 2.81.0-2d85a80
harfbuzz 8.4.0-a69a6cb
libxml2 2.13.0-f506ec6
freetype 2.13.2-12adfc2
lzma 5.7.0-73f629e
pango 1.52.2-bc98a43
libvpx 1.14.0-7d37ffa
snappy 1.2.0-52820ea
openh264 2.4.1-1debdae
kvazaar 2.3.1-0711762
opus 1.5.2-ddbe483
lcms 2.17-83f650b
highway 1.1.0-57cc2d8
nasm 2.16.03rc1

https://www.sendspace.com/file/uyhyoa

[TDS]

AviSynth+ 3.7.3 (r4071, 3.7, x86_64) (3.7.3.0)

https://gitlab.com/uvz/AviSynthPlus-Builds

[TDS]

Originally Posted by Jamaika

fmpeg version 7.1.0-1c45104f Copyright (c) 2000-2024 the FFmpeg developers {opencl not default}
built with GCC: (GNU) 11.3.1 20221227 (SIMD) AVX512ICL

Nice to see that you've added ffprobe

However, this appears to be an AVX512 build, and I just thought I'd see what it did on my old AVX CPU, and it's didn't work, as expected.

And again, may I suggest that you use 7-zip to pack the archive a lot smaller.

https://www.7-zip.org/a/7z2404-x64.exe

or

https://www.7-zip.org/a/7z2404.exe

[Jamaika]

Originally Posted by TDS

However, this appears to be an AVX512 build, and I just thought I'd see what it did on my old AVX CPU, and it's didn't work, as expected.

My ffmpeg won't work with these gcc options. Only AVX512ICL or above.
I don't know if there is AVX4/AVX1024.
Supposedly security can be bypassed programmatically. Then all SSE2/AVX/AVX2/AVX512 boards are supported. I didn't do it. I don't know if it's safe.

[LigH]

If you compile software with AVX512 instructions enabled, then a processor supporting AVX512 is required to just even start this application because already general, not really time-critical code written in a high-level language like C/C++ etc. might have been compiled with AVX512 instructions. A measurable speed advantage is highly doubtful.

This is in contrast to e.g. generic x265 builds which support AVX512 instructions optionally for the most time-critical routines in the encoder (via manually optimized assembler code), but only when you enable it with an additional parameter.

[Jamaika]

Originally Posted by LigH

This is in contrast to e.g. generic x265 builds which support AVX512 instructions optionally for the most time-critical routines in the encoder (via manually optimized assembler code), but only when you enable it with an additional parameter.

For me it isn't only assembler that is unfinished. I didn't remove warming for AVX512.

#if X265_ARCH_X86
if (!strcasecmp(value, "avx512"))
{
p->cpuid = X265_NS::cpu_detect(true);
if (!(p->cpuid & X265_CPU_AVX512))
x265_log(p, X265_LOG_WARNING, "AVX512 is not supported\n");
}
else
{
if (bValueWasNull)
p->cpuid = atobool(value);
else
p->cpuid = parseCpuName(value, bError, false);
}
#else
if (bValueWasNull)
p->cpuid = atobool(value);
else
p->cpuid = parseCpuName(value, bError, false);
#endif

Info highway for Ryzen 7.
// The ZEN4-optimized AVX3 target is numerically lower than AVX3_DL and is thus
// considered better. Do not enable it unless the user explicitly requests it -
// we do not want to choose the ZEN4 path on Intel because it could be slower.
#if defined(HWY_WANT_AVX3_ZEN4) && HWY_BASELINE_AVX3_DL != 0
#define HWY_BASELINE_AVX3_ZEN4 HWY_AVX3_ZEN4
#else
#define HWY_BASELINE_AVX3_ZEN4 0
#endif

Info dav1d. There isn't cpu AVX512
#if ARCH_X86_64
/* We only support >128-bit SIMD on x86-64. */
if (X(r.ecx, 0x18000000)) /* OSXSAVE/AVX */ {
const uint64_t xcr0 = dav1d_cpu_xgetbv(0);
if (X(xcr0, 0x00000006)) /* XMM/YMM */ {
if (cpu.max_leaf >= 7) {
dav1d_cpu_cpuid(&r, 7, 0);
if (X(r.ebx, 0x00000128)) /* BMI1/BMI2/AVX2 */ {
flags |= DAV1D_X86_CPU_FLAG_AVX2;
if (X(xcr0, 0x000000e0)) /* ZMM/OPMASK */ {
if (X(r.ebx, 0xd0230000) && X(r.ecx, 0x00005f42))
flags |= DAV1D_X86_CPU_FLAG_AVX512ICL;
}
}
}
}
}
#endif