hoelee/universal-video-transcode
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

gate: give every run its own raw frame files

Browse Source 
The three tiers run side by side by design (copy is I/O, x265 is CPU, nvenc is the GPU ASIC) and
they share the stage directory - but the gate's raw frame files had FIXED names, so two concurrent
gates overwrote each other's decodes. Measured: a comparison reported 100% of frames below 0.85 -
which reads exactly like a systematic defect and was nothing but two runs writing ssim_ref.yuv at
once. The raws are now tagged with the pid and the file tag. A single-file conversion failed its
gate on this and the 'defect' was entirely my own harness.
This commit is contained in:
hoelee
2026-10-07 14:03:33 +08:00
parent 66a80e2752
commit bcbffd6e4d
1 changed files with 7 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/library_run.py
+7 -2
View File
@@ -340,8 +340,13 @@ def ssim_vs_source(src_local, out_local, info, workdir, tag, dur=SSIM_WINDOW):
"""
start = max(0.0, (info["dur"] / 2) - dur / 2)
cmp_scale = "480:270"
ref = os.path.join(workdir, "ssim_ref.yuv")
out = os.path.join(workdir, "ssim_out.yuv")
# The raw frame files MUST be per-run: the tiers run side by side (by design - copy is I/O,
# x265 is CPU, nvenc is the GPU ASIC) and they share this stage directory, so fixed names let
# two concurrent gates overwrite each other's decodes. Measured: a comparison came out with
# 100% of frames below 0.85 (a convincing-looking 'systematic defect') that was nothing but two
# runs writing ssim_ref.yuv at the same time.
ref = os.path.join(workdir, "ssim_%d_%s_ref.yuv" % (os.getpid(), tag))
out = os.path.join(workdir, "ssim_%d_%s_out.yuv" % (os.getpid(), tag))
counts = []
# Decode WITHOUT an input seek, and cut the window with the trim filter on absolute PTS.
#