Recent long testvector pipelines with 10dB attenuation timed out. This is due to a testcase not ending (decoder enters an endless loop somewhere), see e.g.: https://forge.3gpp.org/rep/sa4/audio/ivas-basop/-/jobs/212307. How to reproduce: Build ref and fxp codec: ``` git checkout 20231128_Update_Ittiam make -j mv ./IVAS_cod IVAS_cod_ref mv ./IVAS_dec IVAS_dec_ref git checkout main make -j ``` Get ltv testvectors: ``` git clone https://forge.3gpp.org/rep/ivas-codec-pc/ivas-pc-testfiles.git cd ivas-pc-testfiles git checkout longer-testv-on-runners cd - ``` Get scaling script from float repo and scale testvectors (10 dB attenuation): ``` git clone https://forge.3gpp.org/rep/ivas-codec-pc/ivas-codec.git cd ivas-codec git checkout basop-ci-branch python3 tests/scale_pcm.py ../ivas-pc-testfiles 0.3162 ``` Run ref encoder: ``` ./IVAS_cod_ref -ism_masa 4 2 ivas-pc-testfiles/ltv48_OMASA_4ISM_2TC_ISM1.csv NULL NULL ivas-pc-testfiles/ltv48_OMASA_4ISM_2TC_ISM4.csv ivas-pc-testfiles/ltv48_OMASA_4ISM_2TC.met 64000 48 ivas-pc-testfiles/ltv48_OMASA_4ISM_2TC.wav bit ``` Run fxp decoder - will not finish, but enter endless loop or similar: ``` ./IVAS_dec HOA2 48 bit out.wav ``` Running the ref decoder for comparison works.