Encoder/decoder player at 14-bits

Encoder/decoder player at 14-bits

Looking for a little help.

For Xp Sp2 on a quad 2.4ghz machine, I would like to benchmark the IPP encoder, decoder, and player for at 14-bits and compare it with other h264 encoder/decoder which currently only go to 8-bits.

I do not have IPP lib or intel compilers for MS installed but I have signed up for a demo of the IPP lib for MS, but if someone could speed the process up and just compile the simple basic versions at 14-bits.

The source material, I basically have a directory of 4k x 6k @16-bit grayscale sequentiial images, say 1000 for the test. I'll convert to HD framesize and 14-bit. I basically just want to point to the directory encode and output to say c:\temp.

for the encoder, i'd like to try the lossless predictive, levels 5 and 5.1

the player, the default

or will standard players choke on trying to play a 14-bit encoded file? the VC-1 codec doesn't appear to support anything over 8-bits??

can anyone help?

14 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

installed the latest par studio demo, VStudio 2008
sample source w_ipp-samples_p_6.0.0.127.zip
all but the followind build

application/simple_player: Failed
application/umc_video_enc_con: Failed
image-codecs/uic failed code '0x2'
jpegview failed code '0x460'

anyone else had these problem?

i downloaded and tried to run the apps in w_ipp-uic_p_6.0.0.127

it bombs looking for ippcore6.0.dll Par Studio is ver 6.1

what a mess, nothing works and i have to spend hours searching for dll's

You may check with Intel IPP 6.1 beta that is available via the latest Intel software product Intel Parallel Studio beta, please register beta at: http://software.intel.com/en-us/intel-parallel-studio-home/

Please confirm if IPP 6.1 fixes this issue.

Thanks,
Ying

Quoting - Ying S (Intel)
You may check with Intel IPP 6.1 beta that is available via the latest Intel software product Intel Parallel Studio beta, please register beta at: http://software.intel.com/en-us/intel-parallel-studio-home/

Please confirm if IPP 6.1 fixes this issue.

Thanks,
Ying

Thanks. I have the installed the latest Parallel Studio which includes IPP 6.1 beta, that appears to be the problem. The demo application is looking for ver 6.0 of ippcore.

Also wikipedia states that the Intel IPP is 14-bits but in the documentation,

file:///C:/extract/ipp-samples/audio-video-codecs/codec/h264_enc/readme.htm

11 3 int 0 1; 8 12; 8 12

chroma_format_idc (0 - monochrom, 1 - 420); bit_depth_luma; bit_depth_chroma

12 5 int

0 3; 8 12; 0 1; 0 2^(bit_depth_aux + 9) -1 ; 0 2^(bit_depth_aux + 9) -1

aux_format_idc; bit_depth_aux; alpha_incr_flag; alpha_opaque_value; alpha

it indicates only 12-bits.

Quoting - Chaos

Thanks. I have the installed the latest Parallel Studio which includes IPP 6.1 beta, that appears to be the problem. The demo application is looking for ver 6.0 of ippcore.

Also wikipedia states that the Intel IPP is 14-bits but in the documentation,

file:///C:/extract/ipp-samples/audio-video-codecs/codec/h264_enc/readme.htm

11 3 int 0 1; 8 12; 8 12

chroma_format_idc (0 - monochrom, 1 - 420); bit_depth_luma; bit_depth_chroma

12 5 int

0 3; 8 12; 0 1; 0 2^(bit_depth_aux + 9) -1 ; 0 2^(bit_depth_aux + 9) -1

aux_format_idc; bit_depth_aux; alpha_incr_flag; alpha_opaque_value; alpha

it indicates only 12-bits.

Well, you should definitely update your PATH and make sure the ippcore-x.y.dll file is reachable using $PATH. Most probably, v6.1b updated it when you installed.

As an addition, whining to get answers faster on a forum when you obviously have no clue about IPP (trying to actually use it is not so hard, provided you have a bit of C knowledge), might not be the most intelligent way of getting those answers.

Regards.

Quoting - elhefe38

Well, you should definitely update your PATH and make sure the ippcore-x.y.dll file is reachable using $PATH. Most probably, v6.1b updated it when you installed.

As an addition, whining to get answers faster on a forum when you obviously have no clue about IPP (trying to actually use it is not so hard, provided you have a bit of C knowledge), might not be the most intelligent way of getting those answers.

Regards.

It's not whining, it is inquiring to find out how to get fast answers, if you had a clue , you would look at the number of posts which the replies are ZERO. Most of the time at commercial forums, you have to pay if you want real answers and not answers from people like you.

Now your post is just flamebait, if you really want me to nuke you back to the stone age continue your ignorant and childish posting.

Just because I haven't posted here doesn't mean I haven't been using the Intel Linux compilers since version 5.

So look "brown belt", don't f_ck with people you have no clue about.

Obviously there is a bug in Par Studio Demo and IPP ver 6 on WinDoZe, if the tests for the project didn't require WinBloze, I wouldn't be posting here.

Quoting - Chaos

It's not whining, it is inquiring to find out how to get fast answers, if you had a clue , you would look at the number of posts which the replies are ZERO. Most of the time at commercial forums, you have to pay if you want real answers and not answers from people like you.

Now your post is just flamebait, if you really want me to nuke you back to the stone age continue your ignorant and childish posting.

Just because I haven't posted here doesn't mean I haven't been using the Intel Linux compilers since version 5.

So look "brown belt", don't f_ck with people you have no clue about.

Obviously there is a bug in Par Studio Demo and IPP ver 6 on WinDoZe, if the tests for the project didn't require WinBloze, I wouldn't be posting here.

1- I you need fast answers from Intel people, why don't you try the commercial support (IIRC, if you registered the IPP trial, you have access to it) ? If you just need informations that may be relevant to your problems or not, use the forum, but do not complain you do not get answers in a timely fashion.

2- Had you looked in the forum, many topics deal with "dll not found". Under Windows, or Linux, a dll/so not found usually results from a messed PATH/LD_LIBRARY_PATH env. variable, which I suggested for you to check.

3- No need to make fun of people that use Windows. Some have no choices but to use it. Every day. Not just for a single test.

4- Badmouthing me might make you feel better (and if it did, well good for you). I however doubt it will actually solve your problem. At least I made a suggestion for you to check. Now who is being childish (c'mon, "Windoze", "Winbloze"... this is so 90's... ;) ?

Quoting - elhefe38

1- I you need fast answers from Intel people, why don't you try the commercial support (IIRC, if you registered the IPP trial, you have access to it) ? If you just need informations that may be relevant to your problems or not, use the forum, but do not complain you do not get answers in a timely fashion.

2- Had you looked in the forum, many topics deal with "dll not found". Under Windows, or Linux, a dll/so not found usually results from a messed PATH/LD_LIBRARY_PATH env. variable, which I suggested for you to check.

3- No need to make fun of people that use Windows. Some have no choices but to use it. Every day. Not just for a single test.

4- Badmouthing me might make you feel better (and if it did, well good for you). I however doubt it will actually solve your problem. At least I made a suggestion for you to check. Now who is being childish (c'mon, "Windoze", "Winbloze"... this is so 90's... ;) ?

Ha ha, well I'm not up on the latest rants again M$

My BSD box runs everyday 24/7 never bombs, no virus and thrashes any DOS pig derivative machine. My kernel is highly tuned and deterministic with memory mgmt that isn't fragmented to hell and back.

I eventually found the problem on the M$ with IPP is that there are numerous errors in the Intel env.bat ippenv.bat etc etc,files that link in wrong files, wrong SDK, wrong VS, etc, etc I ended up creating a custom env.bat that fixes everything and now it compiles.

there is still the issue that wikipedia states that the IPP is 14-bit but all the documentation and examples limit the codec to 12-bit. are the examples 'crippled'?

http://en.wikipedia.org/wiki/H.264

link with summary of 264 codecs

also looking through examples and trying to find predictive lossless encoding.

after hacking through Makefiles and drilling down through multiple layers of Intel bat files, there are still nasty incompatabilities and wrong pointers with libiomp5md.lib vcomp.dll and libguide40.dll

this is why its called demo software.

And the proof, you give some people a vi or cmd prompt, they can build anything.

++ Custom Intel IPP custom environment is found at "C:ProgFileIntelParallel Studio
Composerippia32toolsenvippenv.bat"
Start components compile
******* coreumc : PASSED
******* corevm : PASSED
******* corevm_plus : PASSED
******* pipelineumc_pipeline : PASSED
******* codecaac_dec : PASSED
******* codecaac_dec_int : PASSED
******* codecaac_enc : PASSED
******* codecaac_enc_int : PASSED
******* codecac3_dec : PASSED
******* codecavi_spl : PASSED
******* codecavs_common : PASSED
******* codecavs_dec : PASSED
******* codecavs_enc : PASSED
******* codecavs_spl : PASSED
******* codeccolor_space_converter : PASSED
******* codeccommon : PASSED
******* codecdemuxer : PASSED
******* codecdv100_dec : PASSED
******* codecdv100_enc : PASSED
******* codecdv50_dec : PASSED
******* codecdv50_enc : PASSED
******* codecdv_dec : PASSED
******* codecdv_enc : PASSED
******* codech261_dec : PASSED
******* codech261_enc : PASSED
******* codech263_dec : PASSED
******* codech263_enc : PASSED
******* codech264_dec : PASSED
******* codech264_enc : PASSED
******* codech264_spl : PASSED
******* codecjpeg : PASSED
******* codecme : PASSED
******* codecmp3_dec : PASSED
******* codecmp3_dec_int : PASSED
******* codecmp3_enc : PASSED
******* codecmp3_enc_int : PASSED
******* codecmpeg2_dec : PASSED
******* codecmpeg2_enc : PASSED
******* codecmpeg2_mux : PASSED
******* codecmpeg4_dec : PASSED
******* codecmpeg4_enc : PASSED
******* codecmpeg4_mux : PASSED
******* codecmpeg4_spl : PASSED
******* codecscene_analyzer : PASSED
******* codecspl_common : PASSED
******* codecvc1_common : PASSED
******* codecvc1_dec : PASSED
******* codecvc1_enc : PASSED
******* codecvc1_spl : PASSED
******* ioaudio_renders : PASSED
******* iomedia_buffers : PASSED
******* ioumc_io : PASSED
******* iovideo_renders : PASSED
******* applicationsimple_player : PASSED
******* applicationumc_h264_dec_con : PASSED
******* applicationumc_video_enc_con : PASSED
+-----------------------------------------------------------------------------+
|-OK audio-video-codecs Tue Apr 7 10:39:16 PM 2009
+-----------------------------------------------------------------------------+

Quoting - Chaos

And the proof, you give some people a vi or cmd prompt, they can build anything.

Waow dude... you managed to compile a sample ? How impressive. It sure takes a top-notch FreeBSD hacker to reach this level of performance I guess...
Is there anything else you wanted to share with us, dumb Windoze/Winbloze/Window$ (you forgot this one ! lol) programmers ?

Quoting - Chaos

it's not FreeBSD either. FreeBSD is a UNIX derivative NOT true BSD

Hoo I am deeply sorry for my mistake Mr BSD integrist. Thanks a lot for your very valuable contributions to this forum.

Quoting - Chaos

Let's see you debug kernel mode drivers.

You know what ? I already did. And not just with an easy OS like Linux.

Leave a Comment

Please sign in to add a comment. Not a member? Join today