$ git clone -n https://gitee.com/leacyand/ffmpeg.git /var/lib/mock/dist-an23-epao-build-548022-87732/root/chroot_tmpdir/scmroot/ffmpeg
Cloning into '/var/lib/mock/dist-an23-epao-build-548022-87732/root/chroot_tmpdir/scmroot/ffmpeg'...
$ git fetch origin 5ccf2087596964792ac1dab06308bcab1724a6a6:KOJI_FETCH_HEAD
From https://gitee.com/leacyand/ffmpeg
 * [new ref]         5ccf2087596964792ac1dab06308bcab1724a6a6 -> KOJI_FETCH_HEAD
$ git reset --hard KOJI_FETCH_HEAD
HEAD is now at 5ccf208 This patch adds hand-optimized RISC-V Vector (Zve32x) code for hevc_add_res. All checkasm tests passed. Benchmarks were performed on Sophon SG2044 with native FFmpeg timers: hevc_add_res_4x4_8_c: 26.4 ( 1.00x) hevc_add_res_4x4_8_rvv_i32: 22.9 ( 1.15x) hevc_add_res_8x8_8_c: 88.1 ( 1.00x) hevc_add_res_8x8_8_rvv_i32: 29.6 ( 2.98x) hevc_add_res_16x16_8_c: 318.9 ( 1.00x) hevc_add_res_16x16_8_rvv_i32: 50.2 ( 6.35x) hevc_add_res_32x32_8_c: 1815.4 ( 1.00x) hevc_add_res_32x32_8_rvv_i32:158.4 (11.46x)