IJG porting to PCA problem!

[cloud007]

I want to port huffman decoding part using PCA version IPP,but met the exception of:

When read about 70 rows,the following cal return error code -2.
Can some one tell me what's the cause?


int used_bits_len = 0,last_bits_left,marker = 0;
int valid_pre_fetched_bits = 0;
unsigned int pre_fetched_bits = 0;

stat = ippiDecodeHuffman8x8_Direct_JPEG_1u16s_C1(
cinfo->src->next_input_byte,
&used_bits_len,
(Ipp16s *)&block[0],
&state.last_dc_val[ci],
&marker,
&pre_fetched_bits,
&valid_pre_fetched_bits,
dctbl->pHuffTbl,
actbl->pHuffTbl
);

Thanks

[Vladimir_Dudnik]

Hi Cloud,

Please take a look on JPEG-cross-arch sample, it should be available in IPP v4.1. This sample was dedicated to work on both Windows XP on IA and Windows CE on PCA. So, it may give some hints how to use IPP functions in JPEG codec.

Regards,

Vladimir

[cloud007]

Thanks.It's ok now

[Vladimir_Dudnik]

Hi Cloud,

That's great you solve your issue. By the way, how you find IPP performance on Xscale platform? What about functionality? Is it cover your needs?

Regards,

Vladimir