model: update hevc 10bit QP list

Use a valid QP range accepted by the iHD driver for HEVC 10bit. Unfortunately, there is still a bug in ffmpeg and vpl that incorrectly forces QP in the wrong range... See intel/vpl-gpu-rt#297 Thus, until the issue is fixed, the ffmpeg-qsv encoder code needs to be modified, manually, to remove the QP range constraints so that the requested values can propogate to vpl and the driver to properly generate a trendline model. Signed-off-by: U. Artie Eoff <[email protected]>
uartie · Mar 21, 2024 · 57c7339 · 57c7339
1 parent 6c91591
commit 57c7339
Showing 1 changed file with 3 additions and 0 deletions.
diff --git a/model/encode/10bit/hevc.py b/model/encode/10bit/hevc.py
@@ -35,6 +35,9 @@ def initvars(self, _):
   @slash.parametrize(*TrendModelMixin.filter_spec(spec))
   def test(self, case):
     vars(self).update(case = case)
+    vars(self).update(
+      modelqps = [12, 13, 16, 18, 23, 31, 40, 42, 45, 48, 49, 51, 54, 57, 60, 61, 63]
+    )
     vars(self).update(spec[case].copy())
 
     # Some features/formats are only supported by VDENC (lowpower = 1).