Hello WillyBilly06，
My chip is ESP32-d0wd, set to run at 240MHz. I can smoothly decode LHDC V5 at 44.1kHz/48kHz/96kHz with your decoder, but switching to 192kHz will cause Bluetooth active drop packet and watchdog is triggered. At the same time, I checked your README and determined that the single core decoding stereo has reached the performance limit of ESP32 at 192kHz.
Therefore, for 192kHz, I split the left and right channels according to what you said in the README, decoded them in different cores, and finally mixed the outputs. However, the only effect is the watchdog is not triggered, Bluetooth is still continuously droping packets, i2s is still hungry, and the sound is still lagging.
I set LHDCV5_DEC_PROFILE=1 to enable PROF, and observed that the single channel PROF is about 6.6ms, but the final single frame decoding time is even longer, exceeding the LHDCV 5ms per frame limit. I feel that even if I split the left and right channel decoding, ESP32-d0wd can't process each frame of data within the allowed time.
According to the README, your decoder does not have integer division. I am not sure if there is still room for efficiency optimization in the decoder?
I would like to know what effect you have on ESP32-d0wd when testing 192kHz? What effect on the ESP32-S31 with stronger performance?
I don't have ESP32-S31, but I guess it to maybe usable.
Hello WillyBilly06，
My chip is ESP32-d0wd, set to run at 240MHz. I can smoothly decode LHDC V5 at 44.1kHz/48kHz/96kHz with your decoder, but switching to 192kHz will cause Bluetooth active drop packet and watchdog is triggered. At the same time, I checked your README and determined that the single core decoding stereo has reached the performance limit of ESP32 at 192kHz.
Therefore, for 192kHz, I split the left and right channels according to what you said in the README, decoded them in different cores, and finally mixed the outputs. However, the only effect is the watchdog is not triggered, Bluetooth is still continuously droping packets, i2s is still hungry, and the sound is still lagging.
I set LHDCV5_DEC_PROFILE=1 to enable PROF, and observed that the single channel PROF is about 6.6ms, but the final single frame decoding time is even longer, exceeding the LHDCV 5ms per frame limit. I feel that even if I split the left and right channel decoding, ESP32-d0wd can't process each frame of data within the allowed time.
According to the README, your decoder does not have integer division. I am not sure if there is still room for efficiency optimization in the decoder?
I would like to know what effect you have on ESP32-d0wd when testing 192kHz? What effect on the ESP32-S31 with stronger performance?