On Wed, 27 Feb 2008 08:22:07 +0530, Gv, Naveen wrote: > Dear Mr Paleino, > Thank you for your interest in MKL and your suggestions regarding Debian > distribution. Thank you for looking into this. > Although Intel has no current plans to release Intel MKL as free > software in the manner described in your message, perhaps there are > still ways for users of your molecular/computational biology packages to > benefit from the Intel MKL performance library. > > Your message refers to the dependency on Intel MKL as being optional - > does this mean that your package is designed to use Intel MKL if it is > available in the application environment? Not really -- it must be present at compile time -- and then at runtime. Since I've got no copy of Intel MKL, neither have the chance to have a computer with that library installed, I can't build programs with support for things I don't have. Moreover, Debian bases its infrastructure upon "build daemons": packages must build cleanly there to enter the official distribution. Even if a end-user has Intel MKL installed, our programs won't use it, because they had no such opportunity at compile-time. > Such a usage model has a precedent example in another domain: the OpenCV > open-source computer-vision library is designed to detect and use the > Intel Integrated Performance Primitives (Intel IPP) library, in its > runtime environment, to yield extra performance. Intel IPP is a sister > library product to Intel MKL. I'll look into OpenCV, thanks for the suggestion. Kind regards, David Paleino in behalf of Debian-Med Team -- . ''`. Debian maintainer | http://wiki.debian.org/DavidPaleino : :' : Linuxer #334216 --|-- http://www.hanskalabs.net/ `. `'` GPG: 1392B174 ----|---- http://snipr.com/qa_page `- 2BAB C625 4E66 E7B8 450A C3E1 E6AA 9017 1392 B174
Attachment:
signature.asc
Description: PGP signature