Compilador C++ Intel®

transient files with some prefix?


I'm on Win 8.1 x64 ICC 15.0.3, MSVC 2013.

When I break compilation, sometimes ICC left 4-charactes (readable characters) internal compilation files, they are removed when compilation is unsuccessful.

But when I break it they are left-overs.

This interfere with GIT ignore rules, since it seems their names are randomly-generated.

e.g. when I do report from Amplifier or Inspector, the result directory created is prefixed by letter 'r' (==report?) so I can easily ignore them via configuration of GIT (or possibly other VCS too).

Unexistent warning


The following program does not output anything when compiled with icpc, with the flags "-std=c++11 -c -Wall":

extern int foo(char*);

int main() {
    return 0;

In comparison, here is the output with g++ 4.9.2:

cpp:4:12: warning: deprecated conversion from string constant to 'char*' [-Wwrite-strings]

And with clang 3.5:

cpp:4:9: warning: ISO C++11 does not allow conversion from string literal to 'char *' [-Wc++11-compat-deprecated-writable-strings]

Thank you.

2015 Update 2 upgrade removes old versions, but doesn't install anything new

Much like 3 or 4 months ago:

Repair doesn't work.  Uninstall and reinstall doesn't work either, even with a valid license or a serial number it says there's no components selected to install.  Now I'm downloading the full installer from the registration center to see what that does.

Lame QA guys....


error : assertion failed at: "shared/cfe/edgglue/edg_expr.c", line 847

Windows 8.1 64 VS2013

I need some workaround tips for:

error : assertion failed at: "shared/cfe/edgglue/edg_expr.c", line 845
error : assertion failed at: "shared/cfe/edgglue/edg_expr.c", line 847

errors if compiled with:

Intel(R) C++ Compiler XE for applications running on IA-32, version 15.0.2 Package ID: w_ccompxe_2015.2.179

Intel(R) C++ Compiler XE for IA-32, version 14.0.4 Package ID: w_ccompxe_2013_sp1.4.237

no errors if compiled with:

ld: section PUBLIC_KEYS loaded at [0000000000000610,0000000000000a63] overlaps section BOOT_VECT loaded at [0000000000000500,000


I encountered an error msg. when I used ICC to compile SOFIA. The following are the steps of how we comply with the instructions to install ICC, set the environment, and compile.

1. Install ICC compiler V15 (file name: l_comp_embd_p_15.0.0.037.tgz)

2. Install ips tool (file name: ips2013_external_req.tar.gz) and move to /opt/*

3. eepo sync SOFIA code

4. source build/

5. lunch Sf3gr_svb_garnet-userdebug

6. source ~/intel/system_studio_2015.0.037/bin/ ia32 (ICC installed path)

Is there any advantage to use malloc under 64 compared to win32 ?


I hope this is the right forum. I have a C++ program that is relatively memory hungry but works fine and is stable under linux 64. When I try to run under Win32, I often get a failure of malloc (cannot allocate memory segment). I must admit that I allocate large blocks ( 50 MB) but, again, it works in Linux.

Computation offload design decisions

So, I just read the article about How to offload computation to Intel(R) Graphics Technology and now I am wondering why the support for offloading to GPU was not done through extending existing __declspec(cpu_dispatch(...)) and __declspec(cpu_specific(...))?

For example:
__declspec(cpu_dispatch(generic, core_2_duo_sse4_1, core_4th_gen_avx, gfx)


floating point exception

The attached very simple program crashed when compiled with icc (icc version 15.0.2 (gcc version 4.9.0 compatibility)), under Ubuntu 14.10.

Step to reproduce

  1. Download the attached source file bug.c
  2. Compile it: icc bug.c -o bug
  3. Execute it: bug
  4. Result:

0 1 -> 1
Floating point exception (core dumped)

When compiled with other compilers as gcc or clang, the program behaves as expected, without crashing. The source code does not contain any floating point variable.


Assine o Compilador C++ Intel®