A note about IPP bug fix list

[Ying_H_Intel]

Along with IPP 6.1 update 2 release, we publisha list of bug fixes, known issues, and limitations for each IPP version. You may like toreveiw it and add your commentsso more users can share.

The bug listisat http://software.intel.com/en-us/articles/intel-ipp-library-61-fixes-list/

Some notes:
1. Regarding UIC bug fix.

DPD200084538

UIC sample picnic.exe (64 bits) cannot be started in w_ipp-uic_p_6.1.1.044

The problemonly happenswith64bit picnic (winem64t/em64t). Theroot cause is64bit QT build problem. We rebuild theQT and UIC demo in w_ipp-uic_p_6.1.2.051, so the bug is gone.

DPD200138989

Running application in debug causes heap corruption at the end of functions that useUIC classes

Using all UIC source code and uic dll/lib from w_ipp-samples_p_6.1.035,in debug mode,program crash as hep corruption. The root cause is memory allocation issue in UIC. Fix it in the w_ipp-uic_p_6.1.2.051.

2 Regarding ippResizeSqrPixel

DPD200082866

Artifacts when usinng ippiResizeSqrtPixel_16_C1R with IPPI_INNTER_LINEAR for interpolation

The function give wrong result. 16 bit integers per pixel but the values only range from 0 to about 40

DPD200134979

ResizeSqrPixel function error

See http://software.intel.com/en-us/articles/resize-function-ippiresizesqrpixel-crashed-for-small-image/
and http://software.intel.com/en-us/forums/showthread.php?t=64819

DPD200133708

ippiResizeSqrPixel_8u_C1R performance issue

Improve theperformance of ippResizeSqrPixel when in multi-threads.

Any more notes are welcomed!

Regards,
Ying

[PaulF_IntelCorp]

Ying,

Thank you for posting that additional detail regarding the bug fixes. It is very helpful.

Note, the two links in item #DPD200134979 somehow got turned into only one link. Here's the missing one:

http://software.intel.com/en-us/articles/resize-function-ippiresizesqrpixel-crashed-for-small-image/

Paul

[Mikael_Grev]

Hello,

DPD200134712 could possibly affect my interaction with IPP. Is there any more info on that one? Or, since I have bought the product, can I get more info on the bugs fixed?

Cheers,
Mikael

[Vladimir_Dudnik]

Hi Mikhael,

do you mean issue with title "Distorted JPEG output"? There was an issue on JPEG sample code level. It is not related to IPP functions stability.

Regards,
Vladimir

[Mikael_Grev]

OK, good. Thanks.

[Gennady_F_Intel]

Quoting - Ying H (Intel)

Along with IPP 6.1 update 2 release, we publisha list of bug fixes, known issues, and limitations for each IPP version. You may like toreveiw it and add your commentsso more users can share.

The bug listisat http://software.intel.com/en-us/articles/intel-ipp-library-61-fixes-list/

Some notes:
1. Regarding UIC bug fix.

DPD200084538

UIC sample picnic.exe (64 bits) cannot be started in w_ipp-uic_p_6.1.1.044

The problemonly happenswith64bit picnic (winem64t/em64t). Theroot cause is64bit QT build problem. We rebuild theQT and UIC demo in w_ipp-uic_p_6.1.2.051, so the bug is gone.

DPD200138989

Running application in debug causes heap corruption at the end of functions that useUIC classes

Using all UIC source code and uic dll/lib from w_ipp-samples_p_6.1.035,in debug mode,program crash as hep corruption. The root cause is memory allocation issue in UIC. Fix it in the w_ipp-uic_p_6.1.2.051.

2 Regarding ippResizeSqrPixel

DPD200082866

Artifacts when usinng ippiResizeSqrtPixel_16_C1R with IPPI_INNTER_LINEAR for interpolation

The function give wrong result. 16 bit integers per pixel but the values only range from 0 to about 40

DPD200134979

ResizeSqrPixel function error

See http://software.intel.com/en-us/articles/resize-function-ippiresizesqrpixel-crashed-for-small-image/
and http://software.intel.com/en-us/forums/showthread.php?t=64819

DPD200133708

ippiResizeSqrPixel_8u_C1R performance issue

Improve theperformance of ippResizeSqrPixel when in multi-threads.

Any more notes are welcomed!

Regards,
Ying

It should be noticed that the runtime problem
"Artifacts when usinng ippiResizeSqrtPixel_16_C1R with IPPI_INNTER_LINEAR for interpolation"
marked as fixed in IPP v6.1 update 2 (18 Oct 2009), actually was not fixed in this update.
The fix will available in IPP v.7.0 beta.
--Gennady

[j_miles]

Quoting - Vladimir Dudnik (Intel)

Hi Mikhael,

do you mean issue with title "Distorted JPEG output"? There was an issue on JPEG sample code level. It is not related to IPP functions stability.

Regards,
Vladimir

I would like to have some more details on this regardless of it being on sample code level, please. Actually, I like to have details on all issues that are somehow related to "JPEG". Thanks.


Apart from the above specific request, I would also like to express my opinion on the bug list:
I do think it is a good initiative but I find the list lacking and non-descriptive. I would suggest that you do provide some more details on each issue. Add 1-3 lines of additional description to each item as this will enhance the information level and enable users to be able to better make a decision on whether or not a given issue applies to their usage. Users can then ask for (even) more information if necessary. Many of the current issue titles are simply not sufficient to deduct anything. For example, what does DPD200128941 "Small destination block size optimization" concern, and for which function(s)? Or how about DPD200131858 "Problems with data compression"? Or DPD200133943 "Improve MPEG2Mux", etc ...?
Some issues may not really need much more information (although right now, we cannot even be sure of that), e.g. issue DPD200130212 "Add const specifier in DataWriter::PutData() method" that probably describes well what the issue concerns and the related changes.

I have noticed several other issue that may (or not) apply to our use but it is impossible to deduct that from the little information given and I would then have to ask for details on tens of issues. With a relatively little additional effort I believe you could reduce your support time and at the same time have more pleased users.


Jay

[PaulF_IntelCorp]

Quoting - j_miles

...I would suggest that you do provide some more details on each issue. Add 1-3 lines of additional description to each item as this will enhance the information level and enable users to be able to better make a decision on whether or not a given issue applies to their usage...

Hello Jay,

Providing what can be interpreted as an adequate description of each item in the bug-fix list for every user is impractical. For many of the bugs the descriptions can be quite long and involved. That is why we ask that you post a message on the IPP user forum and reference the "DPD" number of interest if you have questions regarding specific items listed in the bug-fix tables.

As additional information regarding the bugs are posted to the forums we will provide links to those explanations from the bug-fix list so that other users can also find that additional information.

Paul

[j_miles]

Quoting - Paul Fischer (Intel)

Hello Jay,

Providing what can be interpreted as an adequate description of each item in the bug-fix list for every user is impractical. For many of the bugs the descriptions can be quite long and involved. That is why we ask that you post a message on the IPP user forum and reference the "DPD" number of interest if you have questions regarding specific items listed in the bug-fix tables.

As additional information regarding the bugs are posted to the forums we will provide links to those explanations from the bug-fix list so that other users can also find that additional information.

Paul

Hello Paul,

Well, I agree that what's adequate for everyone will be too much due to varying interests. Note that I am also suggesting just a few lines where applicable because some items are (supposedly) adequately described (at this level) through the issue title. I think the effort of doing it to begin with is very small (the person responsible for the issue writes the small title/description) and would save you some support time in the end. There will obviously be the few odd issues that are simply too complicated to describe in few lines, You should agree that a title such as the one for issue DPD200131858 "Problems with data compression" is not very descriptive and a little more (useful) information would help users of data compression functionality a lot. It should at least be possible to deduce whether or not the problem is at sample level (and which sample) or at IPP function level and for which optimizations (if applicable). Additionally, the description should provide an indication of the problem and its severity (e.g. access violations, wrong error codes returned in certain situations, rounding errors, etc.). Whether or not something is a new feature (although I would presume that the service releases are only for bug-fixes..?) or a bug-fix, or just a documentation enhancement would also be valuable.

Another suggestion is to also provide information back to the submitters of issues to Premier Support (and forum issues as well) on what the related DPD issue number is for a given release fixing the given support problem. It would help the submitters identify their issue and the release containing the fix as well. I see several issues that may be related to (or identical to) support cases that we have submitted but I cannot know, so a "link" between the two issues would be beneficial for the support case submitter. This info would be added to the support case and not here. I also note some issue that may be related to forum issues...

As an example (incidentally, an item I am also interested in) look at issue DPD200130124 entitled "JPEG color conversion functions" from v. 6.1-2: There are several things I could think of when I see this; e.g. a feature request of adding more functions for YCbCr to RGB conversion, fixing problems (e.g. rounding errors or optimization issues) in existing functions, etc. And depending on one's usage of such functions and what the issue actually concerns, it may be relevant or it may not be relevant at all. The title+description should reflect that.


- Jay


PS. I am still interested in details on the "JPEG" issues (and others). Let me be more specific; issues DPD200130124, DPD200134755, DPD200134712, DPD200134360, DPD200134372, DPD200134376, and DPD200133366. Thanks.

[PaulF_IntelCorp]

Hello Jay,

I agree, some of those bug fix titles leave a bit to be desired. We only started posting these lists a few months ago and are trying to get more descriptive titles into the list. I'm expecting that as we move forward the descriptions will get better, not worse! We also recently moved to a new bug tracking tool, which raised some havoc on making the list.

Our practice is to inform premiere support customers of bug fixes when a release is made, where that bug fix has a corresponding premiere support number. Not every bug fix in the list has a premiere support issue associated with it, or the issue number may not have been properly tracked when the bug was originally entered into the bug tracking system. We have purposely removed customer-specificID information from the bug fix titles in order to maintain customer privacy. Not all customers want the rest of the world to know of their issues or even, in some cases, that these tools are part of their product.

I'll do my best to try and get some detailed descriptions of your list posted on the KB and will provide a link to the list when the descriptions are available.

Thanks for the feedback,

Paul

[j_miles]

Quoting Paul Fischer (Intel)

...

I'll do my best to try and get some detailed descriptions of your list posted on the KB and will provide a link to the list when the descriptions are available.

Thanks for the feedback,

Paul

Got it. Thanks a lot.

- Jay