64 bits support for IPP

myotun — Tue, 06 Apr 2010 20:28:17 GMT

I am currently using IPP 32 bits library togething with code samples provided from Intel for video encoding and decoding. I found that there are 32 bits dedicated c files under vm, e.g. vm_thread_linux32.c.

If I want to compile the code under 64 bits OS (Linux) using IPP 64 bits library (version 6.1.2.051), does Intel provide 64 bits version of the c files mentioned above or do I have to use only 32 bits version using multiple flags in make file (e.g. -DLINUX32 -DLINUX64) ? In the latter case, will there be any performance degradation because of using 32 bits version of source code?

Thanks

MT

64 bits support for IPP

Ying_H_Intel — Thu, 08 Apr 2010 09:31:13 GMT

Hi MT,

The sample of audio-video-codeccan be build directly underthe 64bit OS using IPP 64bit library. For example, if you are looking at the ReleaseNotes.htm under the directory of audio-video-codec, it mentions how to build 64bit application. or you can just buildthe sampleby buildem64t.sh, which is for 64bit ( I assume your platform is Intel 64 or em64t,
Ia32 - Ia32, like X86-32bit platform
em64t (or Intel 64) - for X86 64bit platform,
IA64 - it for Itanium 64bit platform.
).

Actually, the source code of audio-video-code are crossing 32bit and64bit, windows, linux and Mac OS. In other words, the same source code can be for both 32bit and 64bit application. The vm_thread_linux32.ccan be used in 64bit.(May be the name "linux32"bring a little confusion.)But it is not problem if yourun the build script "buildem64t.sh".

If you build the sample in 64bit with your self makefile, then yes, you need add both flag -D LINUX32 -DLINUX64 and other compiler options like -D __KERNEL_STRICT_NAMEetc, as the sample makefileused.

Regards,
Ying

topic 64 bits support for IPP in Intel® Integrated Performance Primitives

64 bits support for IPP

64 bits support for IPP