04:33 < blakjak888> Hi, I am having real trouble getting MLPACK to compile using any methods on the website. The latest I have tried is to use MLPACK 3.4.1 with Boost 1.74.0 however I am getting some strange linker errors telling me there are "one or more multiply defined symbols found" in almost every mlpack_xxx.exe it tries to build. I am also getting boost linker

04:33 < blakjak888> errors stating that the function basic_iarchive(unsingt int) is already defined boost_serialization-vc140-mt-gd.lib from the xxx1_73.dll. I have no idea where it is getting 1_73 from. Any ideas?

05:44 < blakjak888> I found the problem. I had to remove VCPKG integration as it was causing conflicting libraries to be linked.

05:46 < blakjak888> I finally used CMAKE-GUI to tune the parametes for the VC++ projects, however no matter what I did, I could not get it to download STB into the deps directory. I ended up cloning it off GitHUB. Anyone know what might have caused that?

10:57 < zoq> blakjak888: can you check if you can download - http://mlpack.org/files/stb-2.22/stb_image.h?

15:05 < AakashkaushikGi4> Hi, while searching about gpu support for mlpack i found out about bandicoot and i did check some of it on gitlab but as i am new to gitlab and also bandicoot are there are any beginner issues i can help with or what kind of knowledge do i need to start helping with it ?

15:28 < rcurtin> AakashkaushikGi4: cool that you found it---bandicoot is still in development though so there aren't really any great starter issues

15:29 < rcurtin> more help would be great, but at least for now it's a lot of template metaprogramming and CUDA/OpenCL development

15:30 < rcurtin> I've just about finished some nice kernels for accu() and dot() that seem to perform well, but there is an invalid memory usage issue in my CUDA kernel that I haven't figured out yet :(

15:41 < AakashkaushikGi4> @rcurtin i think i can get myself familiar with template meta programing and CUDA/OpenCL and will read through the bandicoot code and let you know if can help ?

15:43 < AakashkaushikGi4> I also saw the issue of some kernels returning values instead of void

16:02 < rcurtin> yeah, feel free; but keep in mind in many cases there are not clear answers yet, since we are still designing the library

16:03 < rcurtin> so it may be that if we take the time to change return types, we might find it was better to leave it the way it was for some unknown reason

16:03 < rcurtin> so if you do work on it, just remember that pretty much everything is open for discussion still

16:22 < AakashkaushikGi4> Sure, I will keep that in mind, I am interested in this because it can enable proper GPU support for mlpack if i am not wrong, and will help me learn a lot.

16:46 < rcurtin> awesome :)

16:57 < jeffin143[m]> @zoq , @rcurtin may be tag some issues with hacktoberfest

16:57 < jeffin143[m]> Probably good first issues wala