Can I remove gcc compiler after tensorflow installation? - tensorflow

I'm sorry if this is not the right place to ask this. If it is not, let me know and I'll close this question.
I'm trying to set up an environment for a bot we are working on, and part of that will require installing gcc on a server so we can install tensorflow. However, my company has decided that compilers are unsafe, so part of the process for production will require removing the compiler from the server after a set amount of time. After tensor flow is installed, will it be safe to remove the gcc compiler from the server, or will it break my installation?

Provided that tesnorflow itself does not invoke gcc during operation, it should be safe. Unless shared libraries tensorflow depends on are removed with it (can happen if you system packages them as part of the gcc package -- e.g. libstdc++, libgomp, etc).
I'd recommend checking tensorflow executables with ldd -r to find out (after removing gcc -- or query the package manager whether any of the output libraries is owned by the gcc package).

Related

Cannot disable CUDA build and the process stops

Trying to build ArrayFire examples, everything goes well until I get to the CUDA ones. They are supposed to be skipped, since I have an AMD processor/GPU. However, during the build process, the CUDA section is built anyway, failing for obvious reasons, interrupting the rest of the process.
I could manually change the CMakeLists.txt files. However, is there a higher level way to let the build system (cmake) know that I do not have a CUDA compatible GPU?
It looks like the ArrayFire_CUDA_FOUND and CUDA_FOUND macros are erroneously defined on my system.
The ArrayFire CMake build provides a flag to disable the CUDA backend. Simply set AF_BUILD_CUDA to NO via -DAF_BUILD_CUDA=NO at the command line to disable CUDA.

Compile Tensorflow programs with custom compiler

I'm trying to compile a very simple Tensorflow program (which only prints the Tensorflow version) with my company's c compiler but the libtensorflow.so I downloaded from Tensorflow's offical website is incompatible with our c compiler.
My company's c compiler is pretty much just a standard gcc but gcc can compile the program and our custom compiler cannot.
My colleague told me I have two options: (1) replace Bazel's compiler with our compiler and use Bazel to compile the program or (2) Compile the program with Bazel first then compile the program using our compiler and include the pb.h files generated by Bazel (because those bazel files can only be generated by Bazel).
I'm not sure how to do (!) but I tried (2). The problem with (2) is I got erros saying the protoc was generated by an older version and I'm not sure how to change to the right version.
Some additional information: (1) The OS is Linux, (2) I do not have the privilege to use sudo commands, (3) I cannot access system directories (e.g. /usr/local)
Is there any hope I can make this work? You may ask why not just build the program with Bazel. It's because our company's program needs to be run by our company's simulator and the simulator only accepts program generated by our company's compiler.
Your only option is to build tensorflow with Bazel, and tell Bazel to use your C/C++ compiler. The easiest way is to set the CC and CXX environment variables to point to your compiler's executable. If it is really a drop-in replacement of GCC, then it should work and after building you should get a tensorflow binary compiled with your custom compiler.
If special flags are needed then you should make a custom toolchain in Bazel to tell it how to use your compiler, it is a bit complex but not much. Instructions for that are at https://github.com/bazelbuild/bazel/wiki/Building-with-a-custom-toolchain

msys2 and mingw64 gcc version mismatch? (msys-perl)

I just installed msys2 and mingw64, with their development packages. I really need perl-Gtk3. Perl is msys2 and compiled with gcc-4.9.x, Gtk and friends are mingw and compiled with gcc-5.
Perl complains "Glib.c: loadable library and perl binaries are mismatched (got handshake key 0xde00080, needed 0xdd80080)" when building Glib. Should this work?
Thanks.
PS ... mingw-w64-x86_64-perl is simply unable to compile. And yes, I'm careful to use a mingw shell vs an msys shell.
Are you still having this problem? I have been able to build a Perl dev environment in MinGW64, current as of this time.
I have been able to build Perl Gtk2 / Gtk3 applications in that environment and the GUIs work. (Both Gtk2 and Gtk3 based). These applications are used in a production environment with several thousand desktop users. The application runs on OSX, Windows, and Linux, and can be packed into a binary for release as an "executable" for those operating systems. The details here are for the Windows version.
I do this by either installing the requisite system packages first with pacman, then as necessary rebuilding whatever system library packages that I may have modified, from source, using makepkg-mingw.
Then I build the requisite Perl modules using the CPAN shell, and the "look" command.
I use pkg-config to detect what library and header files are needed.
I then build (at minimum), the Perl Glib, Pango, Cairo, Gtk2, and Gtk3 modules using the perl Makefile.PL command.
The LIBS and INC options need to be added to that command to create a Makefile that includes the correct header files, and links to the correct libraries. The EXTRALIBS and LDLOADLIBS sections of the Makefile needs be correct.
Also ExtUtils::MM_Win32.pm ExtUtils::Liblist::Kid.pm needed to be edited due to the different archname reported by the MinGW64 perl.
I am only giving a general answer, because I was thinking offing a YouTube video on this. If this is a desired topic I will.

Can a library that uses CMake also be built with SCons?

I want to use KDL (Kinematics and Dynamics Library) in robot control box. But robot control box uses SCons as their build system while KDL uses CMake.
It turned out that the control box doesn't have CMake installed. Should I install CMake in the control box? Or write SCons file for compiling KDL?
====================================================
My question is ambiguous. Sorry for that. And unfortunately, I cannot show the link of Control Box, it's not public. Here is link of KDL installation manual.
http://www.orocos.org/kdl/installation-manual
Let me make it more clear.
Forget all of previous question above and all about Control box, KDL. Let's say that you want to use one library. But the library can be built using CMake according to installation manual. Your PC doesn't have CMake installed but it has SCons, and unfortunately you should not install CMake on your PC.
If you can only use SCons, what can you do?
I know this situation is not usual, I want to know your opinion.
To answer your initial question: Yes, you should always try to install CMake, if that is a build requirement for you library and if you need to build that library from the sources.
To answer your later question: Replacing or rewriting the build system scripts is a major effort and not advisable. In general there is no script to convert build-systems. Such script might help to make the manual transformation. If you have a look at LLVM's effort to replace Autotools by CMake or Boost replacing it's own build system by CMake, you find out it takes several people several years and still not everybody is satisfied.
Often you don't need to build the library yourself. Either there are already built packages from the project directly of from your distribution (Debian etc. packages) or third party packagers like Mac Ports or NuGet.
In your case KDL provides Debian/Ubuntu packages.
Additional KDL is part of ROS, which is experimental in Homebrew for OS X.

How to set up a dev. environment for ARM board AT91SAM7-EX256 under x86 Linux?

I am a newbie to embedded developement, as figure shown. I have a small ARM board, AT91SAM7-EX256. I have also a JTAG programmer dongle, too. I am using Linux (Ubuntu x86_32) on my notebook and desktop machine. I'm using CodeSourcery Lite for cross-compiling to ARM-Linux.
Am I right that I can't use this Linux-target cross-compiler to make binary or hex files for the small ARM board (it comes without any operating system)? Should I use the version called ARM EABI instead?
As I see, it's a "generic" ARM compiler. I've read some docs, and there're lot of options to specify the processor type and instruction set (thumb, etc.), there will be no problem with it. But how can I tell the compiler, how should the image (bin/hex) looks like for the specific board (startup, code/data blocks etc.)? (In assemblers, there're the org and load directives for it.)
What software do I need to capture some debug messages from the board on my PC? I don't want to on-board debugging, I just need some detailed run-time signal, more than just blinking leds.
I have an option to use MS-Windows, I can get a dedicated machine for it. Do you recommend it, is it much easier?
Can I use inline assembly somehow in my C code? I dunno anything about that. Can I use C++ or just C?
I have also a question, which don't need to answer: are there really 4096 kind of GNU compilers and cross-compilers (from Linux_x86_32 -> Linux_x86_32, Linux_x86_32 -> Linux_ARM, OSX -> Linux_ARM, PPC_Linux -> OSX) and 16 different GNU compiler sources (as many target platforms/processors exists) around? The signs says "yes", but I can't believe it. Correct me, and show me the GNU compiler which can produce object file for any platform/processor, and the universal linker which can produce executable for any platform.
While Windows is not a "better" platform do this kind of embedded development on, it may be easier to start with since you can get a pre-built environment to work with. For example, Yagarto (which I would recommend).
Setting up an embedded development environment on Linux can require a considerable amount of knowledge, but it's not impossible.
To answer your questions:
Your Linux cross-compiler comes with libraries to build executables for a Linux environment. You have hinted that you want to build a bare-metal executable for this board. While you can do this with your compiler, it will just confuse things. I recommend building a baremetal cross-compiler. Since you're building your own baremetal executable (and thus you are the operating system, the ABI doesn't matter since you're generating all of the code and not interoperating with other previously built code.
There are several versions of the ARM instruction set (and Thumb). You need to generate code for your particular processor. If you generate the code for a newer version of the instruction set, you will likely generate code which generates a reserved instruction exception. Most prebuilt gcc cross-compiler toolchains for ARM are "multilib" and will build for a variety of architectures in both ARM and Thumb.
Not sure exactly what you're looking for here. This is a bare metal platform. You can use the debugger channel to send messages if you're debugging on target, or you'll need to build your own communication channel into the firmware you write (i.e. uart support).
See above.
Yes. See here for details on gcc's extended inline assembly syntax. You can do this in C++ and C. You can also simply link pure assembly files.
There is no universal gcc compiler / linker. You need a uniquely built compiler for each host / target combination you use.
Finally, please take a look at Atmel's documentation. They have a wealth of information on developing for this target as well as a board package with the needed linker directives and example programs. Note of course the package is for Atmel's own eval board, but it will get you started.
http://sam7stuff.blogspot.com/
I use either of the codesourcery lite versions. But I have no use for the gcc library nor a C library, I just need a compiler.
In the gcc 3 days newlib was great, modify two files worth of system support (simple open, close, read, putc type stuff) and you could compile just about anything, but with gcc 4.x you cannot even go back and cross compile gcc 3.x, you have to install an old linux distro in a virtual machine.
To get the gcc library yes you probably want to use the eabi version not the version with linux gnueabi in the file names.
You might also consider llvm (if you dont need a C library, and you will still need binutils), hmm, I wonder if newlib compiles with llvm.
I prefer to avoid getting trapped in sandboxes, learn the tools and how to manipulate the linker, etc to build your binaries.