ippiDecodeHuffman8x8_Direct_JPEG_1u16s_C1 seems ineffective!

[cloud007]

After I replace the huffman decoding part of IJG with IPP function piDecodeHuffman8x8_Direct_JPEG_1u16s_C1,the decode speed is almost same as orginal IJG.
Can someone tell me where does the optimization performed in this function?And if erver compared with that of IJG?

[Vladimir_Dudnik]

We did not have IJG sample for Xscale, so we did not compare IJG with using IPP vs original. We are preparing IJG sample for IPP v5.0 for IXP on Linux big-endian, so I hope you'll find it useful. Preliminary the release is expected sometime at the end of this year.

Regards,

Vladimir