Intel® C++ Compiler

possible icc bug with for loop condition?

Hi,

I have a program which does 1-byte vector sum using SSE intrinsics.

When I compile it with Intel compiler (ICC 15.0.1, x86_64, Linux, SandyBridge CPU) it segfaults, it looks like the end condition of the loop is not checked correctly. Same code works with GCC.

My only optimization flag -s "-O3" (When compiling with -O1 the program works).

icc compile error on OS X

The code:
--------------------------
#include

int main() {
#ifdef __clang_version__
printf("%s\n", __clang_version__);
#endif
}
---------------------------

__clang_version__ is a builtin macro defined by clang, and is supposed to be irrelevant to icc. But icc emits a compile error when compiling this code:

test.c(5): error: extra text after expected end of number
printf("%s\n", __clang_version__);
^

This looks like a bug.

Specs:
OS X 10.10.3
icc 15.0.2

Why is my project still being linked using LINK instead of xilink?

I've just converted a project that used the Microsoft compiler to use the Intel one. I'm using Visual Studio 2013 and Intel C++ 15.

Now, looking at compiler output, I see the following:

1>icl: : warning #10210: problem with Microsoft compilation of 'Everything.cpp'
1>Everything.obj : warning LNK4229: invalid directive '/Intel(R) compiler non-linkable IL object file' encountered; ignored
1>LINK : fatal error LNK1561: entry point must be defined
========== Build: 0 succeeded, 1 failed, 0 up-to-date, 0 skipped ==========

Subscribe to Intel® C++ Compiler