Hi, I am studing NvDecodeGL sample to find sth about the decode performance like decode delay.
I noticed that there are 2 parameters:
ulNumDecodeSurfaces, its init value is 20, will be adjusted to 6 in the code.
ulNumOutputSurfaces, its default value is 2, the max value is 8 commentted by the code.
Does anyone know what these 2 parameters mean? Do/How they impact the decode performance? How can I adjust these 2 parameters to get good performance?
BTW, I put some print codes in the decodePicture function, and got some results like:
CurrPicIdx=0, ref_pic_flag=1, intra_pic_flag=1, decode time = 0.104495
CurrPicIdx=1, ref_pic_flag=1, intra_pic_flag=0, decode time = 0.052375
CurrPicIdx=2, ref_pic_flag=0, intra_pic_flag=0, decode time = 0.024016
CurrPicIdx=3, ref_pic_flag=0, intra_pic_flag=0, decode time = 8.990667
CurrPicIdx=4, ref_pic_flag=0, intra_pic_flag=0, decode time = 16.952734
CurrPicIdx=5, ref_pic_flag=1, intra_pic_flag=0, decode time = 6.985843
CurrPicIdx=2, ref_pic_flag=0, intra_pic_flag=0, decode time = 6.819008
CurrPicIdx=3, ref_pic_flag=0, intra_pic_flag=0, decode time = 0.045222
CurrPicIdx=4, ref_pic_flag=0, intra_pic_flag=0, decode time = 15.547287
CurrPicIdx=2, ref_pic_flag=1, intra_pic_flag=0, decode time = 7.243120
CurrPicIdx=0, ref_pic_flag=0, intra_pic_flag=0, decode time = 7.127895
CurrPicIdx=3, ref_pic_flag=0, intra_pic_flag=0, decode time = 0.045988
CurrPicIdx=4, ref_pic_flag=0, intra_pic_flag=0, decode time = 15.211319
CurrPicIdx=0, ref_pic_flag=1, intra_pic_flag=0, decode time = 7.439847
CurrPicIdx=1, ref_pic_flag=0, intra_pic_flag=0, decode time = 6.656772
The input stream is IPBBBPBBBPBBBPB. The CurrPicIdx is always smaller than ulNumDecodeSurfaces, but I can not find any other relationship between them, do they have any other relationship? Why is the CurrPicIdx not in order? Why the differences of the decode time is so big?
Are there any other configurations impacting the decode performance?
Thank you very much!