The following Q&A session is the result of the Intel® IPP 7.0 beta webinar presented in May of 2010 by Paul Fischer of Intel Corporation.
You can download and view a recording of the webinar.
|Q:||What are the advantages of JPEG XR compared to JPEG 2000/JPEG?|
|A:||JPEG XR was designed with embedded/mobile devices in mind. In comparison to standard JPEG it provides:
• Higher compression (100:1 vs. 10:1)
• Broader data ranges -More image formats (8-16-32 bpp vs. 8-12 bpp)(bpp- bit-per-pixel)
• Advanced decoding features
• Richer color support
Go here for additional information on JPEG XR support in the IPP library.
|Q:||Is Intel® IPP available on Mac OS* X?|
|A:||Yes, Intel® IPP has always been available for Mac OS* X, but we no longer offer it as a standalone product. The library is included as part of the Intel® Compiler Professional Edition for Mac OS* X product.|
|Q:||Do you have recipes (command line examples) on how the BOOST libraries could be compiled using the optimized ipp_zlib,ipp_bzip2, etc., if this is at all possible?|
|A:||We do not have examples of using IPP in the BOOST libraries, but we are reviewing the BOOST libraries to evaluate how IPP could be used as a potential feature enhancement in future releases of the IPP library. If you have some specific suggestions, please let us know via the IPP fourm or priviately via Premier Support.|
|Q:||Would there be any conflict if one installs Intel® IPP 7.0 beta and Composer XE 2011 beta, given that they now install in the same directory structure?|
|A:||There will be no conflict, there are specific directories within the Composer directory that are specific to the IPP library. Additionally, if you wish to install more than one copy of the IPP library on your machine, you can do so (this is referred to as a "side-by-side" installation).|
|Q:||Is DMIP- Deferred Mode Image Processing(ever) going to be available as a static library - and preferably as source code?|
|A:||No, DMIP will not available as a static library or as source code in this version of the IPP library.|
|Q:||When you said the OpenMP static library has been deprecated, does that mean the ipp*_t.lib files are going away?|
|A:||The static OpenMP library files are still provided with this version of the product; however, we strongly advise that you transition from the static version of the OpenMP library to the dynamic version. Deprecating the static version of the OpenMP library means the ipp*_t.lib files will no longer be included in the product in the next major IPP release which will be delivered no sooner than 1 year after the IPP 7.0 release.|
|Q:||Now that different instruction sets such as SSE 4.1, 4.2 and AES have been merged into single library functions, how do you dispatch them for the correct CPU?|
|A:||All of the functions in the p8/y8 libraries are tuned and built for SSE4.1. For that small subset of functions that also have optimizations for SSE4.2 and/or AES-NI a small internal dispatcher inside the p8/y8 libraries is used. SSE4.2 and AES-NI instruction sets are used in a limited number of functions (not more than ~100 out of ~12000), so to minimized the library size limited number of functions branch internally to the appropriate optimization. The overhead is about 10-14 cpu clocks and is negligible in comparison with the function execution time.|
|Q:||Is "intel64" in the new directory structure the same as "em64t" in the old directory structure?|
|A:||IPP 7.0 beta establishes a new directory structure to simplify integration of the Intel® IPP stand-alone products with the Intel Compiler Professional Editions. The new directory structure also align with other of Intel software products, like Intel® Compiler, Intel® MKL, Intel® TBB etc. Click here to get more information on new directory structure.|
To get more information about IPP 7.0 Beta click here