I have a quite large x64 c++ program that is compiled with Visual Studio. There are two ways how do we compile it - one is with sln solution file (from IDE or through msbuild), another is through Makefile in VS command prompt (which then internally uses Visual Studio's cl.exe and link.exe).
Somehow I got into situation that binary build through Makefile produces executable with /AVX instructions enabled, and I don't want that, so I want to find out how to get rid of it.
I try to compile and link it without any parameters (especially without /arch:AVX). I hope that I compile and link the same sources and libs in both sln and Makefile.
What is the easy way how to find out where the AVX comes from? Project includes dozens of h/cpp files, around 40 own libs, 10 windows libs. How would you proceed?
The simple way, find an old PC without AVX, run your app in debugger, it’ll crash with invalid instruction runtime error and will trap to debugger. This way you’ll immediately find out what code compiled into that instruction.
Another, harder way, write a C# console app that disassembles the compiled DLL or EXE, and searches for AVX instruction. You’ll need Gee.External.Capstone nuget package for disassembling, PeNet package for parsing into sections. Also this code to disassemble. The disassembleStreamEx method will return you the stream of the instructions. Search for X86InstructionGroup.AVX or X86InstructionGroup.AVX2 instructions, print addresses, then use e.g. WinDBG + PDB file to translate addresses into source code location.
Update: before doing that, search your codebase for files containing _mm256_, and *.asm files. If you’re lucky you won’t have to do anything else.
Related
I'd like to build a keyboard layout that runs on Win7 and later without using MSKLC.
I downloaded the Keyboard Layout Samples, and although it says it requires VS2013 Preview and WDK8.1 Preview, it builds using VS2012 and WDK8 after changing the "Platform Toolset" property from "WindowsApplicationForDrivers8.1" to "WindowsApplicationForDrivers8.0".
But: Checking the kbdus.dll built with Dependency Viewer reveals that it imports quite a lot of stuff from kernel32.dlland from msvcr110.dll. This is in contrast to kbdus.dll from the Windows\system32 directory, which, as one might expect, does not import anything and exports one single function.
My first question: Is it an issues that the dll built imports from msvcr110.dll?
I guess it is, as Win7 does not include this dll. (Honestly, I'm wondering how this can be, because these are official Microsoft samples.)
Tweaking around with compiler and linker settings, I managed to get rid of imports from msvcr110.dll, but the dll built still imports a lot of stuff and is quite large (> 70kB), compared to the original kbdus.dll (7kB). It's the same with x86 and x64, Debug and Release builds. I'm looking for the /compileandlinkanddonothingfancy switch.
My second question: How can these dlls be built correctly?
(Note: I found this question, but it is about WDK7. That will be what I try next, if I don't get it with WDK8.0)
I think one should pass /noentry to the linker.
I have much trouble compiling OpenCV 245 with GPU support. With some effort I managed to have cuda support up and running, but now I am stuck on opencl, here is the problem:
At some point during the compilation, the file kernels.cpp is generated, containing all kernel functions as strings. For what I understand, they are converted automatically from the .cl files with the cl2cpp.cmake script.
What I don't understand is that one file is excluded from the build: nonfree_surf.cl (which is on my disk, alongside all other .cl files) is not included, either in the visual studio project, or in the kernels.cpp that is automatically generated. This leads to an undefined error at link time.
I have tried manually adding the nonfree_surf.cl to the visual studio project. This does not change anything. In the CMakeLists.txt for ocl module, all the cl files seem to be automatically added with the line:
file(GLOB CL_FILES "${CMAKE_CURRENT_SOURCE_DIR}/src/kernels/*.cl")
I have tried manually adding nonfree_surf to CL_FILES, with no effect.
I have very little knowledge of CMake, hence I don't understand well what is going on. Can somebody give me a clue how I could find the reason of this behavior, namely:
Why are all .cl files added to my VS project, except nonfree_surf.cl ?
How can I correct this ?
Maybe I can execute cl2cpp.cmake script manually ? If so, how ?
Managed it by manually running the script:
cmake -DOUTPUT_PATH=c:/opencv/kernels2.cpp -DCL_DIR:PATH=c:/opencv/modules/ocl/src/kernels -P "c:\opencv\modules\ocl\cl2cpp.cmake"
For some reason (probably the same that makes nonfree_surf not be processed), not every necessary function are processed this way, so I juste copy-pasted the nonfree_surf string into kernels.cpp, and proceeded with the build.
If anyone needs the binaries, since they are a pain to compile, here they are:
Opencv 2.4.5 binaries compiled with VS2010 x86 (WIN32) including ocl and gpu library.
Is it possible to compile libexif with Visual Studio 2010? I have been trying to do so and have been running into a whole slew of problems. I cannot find any information about whether anybody has successfully done this before. I know I can use MinGW to compile the library, but I am in a situation where I need it to be compiled with Visual Studio and then need to link to it from a Visual C++ app. Is this possible?
To answer your question: Yes it is possible... but it is a bit of a hack. Libexif uses functions that MSVC has chosen not to implement. See my working example VS2010 project below (if you don't like downloading files then skip to my explanation of what needed changing to get it to work below):
https://www.dropbox.com/s/l6wowl8pouux01a/libexif-0.6.21_CompiledInVS2010%2BExample.7z?dl=0
To elaborate, the issues that needed a "hack" (as hinted in the LibExif readme-win32.txt documentation) are:
Libexif uses inline in several places which is not defined in VS for C, only C++ (see this)
Libexif uses snprintf extensively in the code which is not defined in VS (see here)
You need to create the config.h yourself without a ./configure command to help you. You could read through the script but most of it doesn't make sense for Windows VS2010.
You will need to define GETTEXT_PACKAGE because it's probably setup in the configure file. I just choose UTF-8, whether that is correct or not I'm not sure.
There was a random unsigned static * that needed to be moved from a .c file to the .h file as C in VS doesn't allow you to create new variables inside functions in the particular way they were trying to do.
Read the "readme-win32.txt" file. Advice is:
hack yourself a build system somehow. This seems to be the Windows way of doing things.
Don't get your hopes up. The *nix way of doing things is the configuration script that needs to be run first. It auto-generates source files to marry the library to the specific flavor of *nix. The configuration script is almost half a megabyte. Three times as much code as in the actual .c files :) You cannot reasonably get that working without MinGW so you can execute the script. Once you got that done, you've got a better shot at it with a VS solution. As long as it doesn't use too much C99 specific syntax.
I have the following problem:
I have a Matlab program in form of some set of *.m files. It is later compiled into executable and used. The problem is that occasionally the resulted executable just hangs and this behavior cannot be reproduced when debugging/running *.m files from IDE (even using the same input data).
To figure out what hapens I intended to:
compile (somehow) *.m files into C/C++
compile C/C++ as debug to get .exe and .pdb
And later when .exe hangs just 'attach' visual studio debugger to hanged .exe to check where it 'loops/waits'.
Unfortunaley Matlab Compiler (as I was told today) does not produce C/C++ code before creating executable. I was misleaded by -g option of mcc which according to the documenttion is supposed to do the following:
-g Generate Debugging Information
Include debugging symbol information for the C/C++ code generated by MATLAB Compiler.
It looks exacly like the thing I want to archive.
I would appreciate if someone could explain me that discapency or suggest how to archive what I am trying to do (if it can be done at all).
It is not possible to create a debuggable code in Matlab compiler, because the deployed code uses MCR. (Matlab virtual machine) .
See this question : Is there any way to debug compiled components using Matlab Debugger?
Since you don't have errors, but rather an infinite loop, the best solution in that case would be screen outputs, and hopefully you will trace the bug.
For my bachelors thesis, I am implementing a distributed version of an algorithm for factoring large integers (finding the prime factorisation). This has applications in e.g. security of the RSA cryptosystem.
My vision is, that clients (linux or windows) will download an application and compute some numbers (these are independant, thus suited for parallelization). The numbers (not found very often), will be sent to a master server, to collect these numbers. Once enough numbers have been collected by the master server, it will do the rest of the computation, which cannot be easily parallelized.
Anyhow, to the technicalities. I was thinking to use Boost::Asio to do a socket client/server implementation, for the clients communication with the master server. Since I want to compile for both linux and windows, I thought windows would be as good a place to start as any. So I downloaded the Boost library and compiled it, as it said on the Boost Getting Started page:
bootstrap
.\bjam
It all compiled just fine. Then I try to compile one of the tutorial examples, client.cpp, from Asio, found (here.. edit: cant post link because of restrictions). I am using the Visual C++ compiler from Microsoft Visual Studio 2008, like this:
cl /EHsc /I D:\Downloads\boost_1_42_0 client.cpp
But I get this error:
/out:client.exe
client.obj
LINK : fatal error LNK1104: cannot open file 'libboost_system-vc90-mt-s-1_42.lib'
Anyone have any idea what could be wrong, or how I could move forward? I have been trying pretty much all week, to get a simple client/server socket program for c++ working, but with no luck. Serious frustration kicking in.
Thank you in advance.
The reason the build is failing is because it cannot find the library file containing boost system. Boost includes a "handy" autolinking feature, such that when you include a header file for a binary libaray (as opposed to a header only library), boost automatically tells the compiler that it should link in the library. The downside to this is that boost doesn't tell the compiler where to find the library.
The short answer is to read a little further in the boost getting started guide. This page shows you how to add the necessary flags to the compiler command line: Getting started on windows: linking from the command line.
The first thing you have to do is find the .lib file. Boost hides them in a deep directory structure, so search for it starting in the directory you ran bjam from. Make note of the directory where the file is. You may also wish to use bootstrap --prefix=/some/install/location and bjam install to install boost somewhere other than the source directory in which you built it.
Are you building your project using a Visual Studio solution, or on the command line?
If you are using a solution file, find the link page in the solution properties. There should be a box where you can enter additional library paths. Add the directory in which you boost .lib files reside to this box.
If you are using cl on the command link, familiarize yourself with the command line options for cl and link. You can pass commands to the linker using the cl option /link, and the linker command you are looking for is /libpath.