Is there anyways to get the AVX intrinsics included via x86intrin.h in GCC 4.8.2 without turning on -mavx for the whole compilation unit? I only want to use them in a function that I've marked target("avx"). More recent versions of gcc seem to do this correctly. But with 4.8.2 even if I manually #define __AVX__ (ew), it complains about the __builtin* functions not being defined...
Related
Here's the code:
main() {}
On gcc.godbolt.org, both GCC 10.1 and Clang 10 (with -Wall -Wextra -pedantic-errors -std=c++20) refuse to compile this.
GCC: error: ISO C++ forbids declaration of 'main' with no type [-Wpedantic]
Clang: error: C++ requires a type specifier for all declarations
But on my local machine, MinGW GCC happily accepts this code without any errors or warnings (with the same flags). This is not something new; this specific peculiarity was there for years.
Why does MinGW GCC behave differntly from the regular GCC in this case? Are there any flags to make it diagnose this error?
I got my GCC from MSYS2. It identifies as
# g++ --version
g++.exe (Rev3, Built by MSYS2 project) 10.1.0
Copyright (C) 2020 Free Software Foundation, Inc.
Clang 10 on the same machine does reject the code (the official binary, using libstdc++ from this GCC).
As noted by #ssbssa, MinGW GCC enables -fms-extensions by default, and it's one of the effects of that flag.
Compile with -fno-ms-extensions to fix that.
Intel Fortran compiler/linker has the optional flag -ipo-c or /Qipo-c which enables the generation of a single interprocedurally-optimized object file from all files, which can be later used for linking. Is there an equivalent flag to Intel's -ipo-cin gfortran?
GCC has -fwhole-program, does that work for gfortran?
Or if you don't want to pass all the Fortran source files on one giant command line, there's -flto link-time optimization which uses a linker "plugin" to run the optimizer on GIMPLE stored in .o files (instead of or as well as machine code).
LTO means you should pass all your optimization options to the invocation of gfortran that does the linking, as well as the gfortran -c that compiles to .o.
So you might use gfortran -ffast-math -O3 -march=native -flto to compile and link, assuming gfortran supports the same options as gcc. (And that -march=native is what you want: make an executable optimized for the computer you compiled on, which might SIGILL on other computers without all the ISA extensions this one supports.)
I'm trying to compile and run my code migrated from Unix to windows. My code is pure C++ and not using Qt classes. it is fine in Unix.
I'm also using Qt creator as an IDE and qmake.exe with -spec win32-g++ for compiling. As I have sse instructions within my code, I have to include emmintrin.h header.
I added:
QMAKE_FLAGS_RELEASE += -O3 -msse4.1 -mssse3 -msse3 -msse2 -msse
QMAKE_CXXFLAGS_RELEASE += -O3 -msse4.1 -mssse3 -msse3 -msse2 -msse
In the .pro file. I have been able to compile my code without errors. but after running it gives run-time error while going through some functions containing __m128 or like that.
When I open emmintrin.h, I see:
#ifndef __SSE2__
# error "SSE2 instruction set not enabled"
#else
and It is undefined after #else.
I don't know how to enable SSE in my computer.
Platform: Windows Vista
System type: 64-bit
Processor: intel(R) Core(TM) i5-2430M CPU # 2.40Hz
Does anyone know the solution?
Thanks in advance.
It sounds like your data is not 16 byte aligned, which is a requirement for SSE loads such as mm_load_ps. You can either:
use _mm_loadu_ps as a temporary workaround. On newer CPUs the performance hit for misaligned loads such as this is fairly small (on older CPUs it's much more significant), but it should still be avoided if possible
or
fix your memory alignment. On Windows/Visual Studio you can use the declspec(align(16)) attribute for static allocations or _aligned_malloc for dynamic allocations. For gcc and most other civilised platforms/compilers use __attribute__ ((align(16))) for the former and posix_memalign for the latter.
As this question shows, with g++, I can do g++ -S -masm=intel test.cpp.
Also, with clang, I can do clang++ -S test.cpp, but -masm=intel is not supported by clang (warning argument unused during compilation: -masm=intel). How do I get intel syntax with clang?
As noted below by #thakis, newer versions of Clang (3.5+) accept the -masm=intel argument.
For older versions, this should get clang to emit assembly code with Intel syntax:
clang++ -S -mllvm --x86-asm-syntax=intel test.cpp
You can use -mllvm <arg> to pass in llvm options from the clang command line. Sadly this option doesn't appear to be well documented, and thus I only found it by browsing through the llvm mailing lists.
As of clang r208683 (clang 3.5+), it understands -masm=intel. So if your clang is new enough, you can just use that.
Presuming you can have Clang emit normal LLVM byte codes, you can then use llc to compile to assembly language, and use its --x86-asm-syntax=intel option to get the result in Intel syntax.
i have some problem with my boost library. i m using freebsd and installed my boost using ports. Boost version is : 1.45 and i use g++47 as compiler. Also i have never defined BOOST DISABLE THREADS at there : /usr/local/include/boost/config/user.hpp .Also exactly my error is :
/usr/local/include/boost/config/requires_threads.hpp:29:4: error: #error "Threading support unavaliable: it has been explicitly disabled with BOOST_DISABLE_THREADS"
explicitly but where ?? And my compile command;
g++47 -O3 -Wall -std=c++0x -I. -Iinclude -I../include -I/usr/local/include -c -o Application.o src/Application.cpp
Thanks
The experimental GCC version 4.7 disables Boost.Threads. See: https://svn.boost.org/trac/boost/ticket/6165
Edit: It should be noted that as of the release version of GCC 4.7, and Boost higher than 1.48 (Boost_1_48_0 is still not working), threads works again.
See the ticket 6165 mentioned above by Joachim:
To define 'threads' support, GCC <= 4.6 defines _GLIBCXX__PTHREADSwhereas GCC >= 4.7 defines _GLIBCXX_HAS_GTHREADS. So, in order to compile older Boosts using any GCC more recent than 4.6 you need the patch libstdcpp3.hpp.patch enclosed in that ticket.
Another problem that could also prevent Boost on working with modern compiler is ticket 6940 (TIME_UTC has a special meanhing in C11, therefore Boost >= 1.50 use TIME_UTC_ instead)