Intel® C++ Compiler

odd message from linker about sincosf using icc15.0

I'm using icc 15.0.2 on 64bit linux (Centos 6.5) and when I go to link a .so I get this message:

 

646   1 sincosf{ic=SINCOSF}( (F32) phi_8909_V$17a, (F32 *)sinphi_8909_V$17b, (F32 *)cosphi_8909_V$17c );  [CALL_CONVENTION_UNIX_ABI]

There's no indication if this is a warning or an error. That is all that comes out of stderr.

I don't get this using 14.0 or 12.1.

My code does make a call to sincosf()

- James

 

 

 

 

 

SSE horizontal sum

Dear Intel developers,

I have to do an horizontal sum of float by using SSE and adding the results on another float. I wrote this:

 

float x=0;
float denom_arr_tmp[4];
_mm128 denom_tmp;

for(.....) {

//calculate denom_tmp

}
_mm_store_ps(denom_arr_tmp, denom_tmp);
x+= denom_arr_tmp[0] + denom_arr_tmp[1] + denom_arr_tmp[2] + denom_arr_tmp[3]

I'm not sure is the best way. What is the fastest way to do float horizontal sum?

Thanks.

valid code corrupts compiler and causes bug in template deduction

The following code below works fine in clang or gcc, but fails with intel compiler 2016.

Note if you comment out the section of code labeled block A (which has nothing to do with anything and isn't referenced) the code compiles. It seems that somehow it's corrupting icc. Please suggest a work around and let me know when you have a patch available.

Visual Studio 2013: cannot open file libmmd.lib on vc++ projects

I downloaded and installed the trial of Parallel Studio XE 2015 for Windows, and tried to use it to compile a Visual Studio 2013 solution.  This solution contains about 50 vc++ projects.  Some use MFC, others are test projects.  I started by right clicking the solution in the Solution Explorer, and selecting Intel Compiler -> Use Intel C++.  I didn't want to use the Intel Compiler for the test projects, so I set them back to use VC++ by right clicking on the project, selecting Intel Compiler -> Use Visual C++.  There are a few other production projects where I did this as well.

Application generating core dump

i am currently working on one application which is generating core dump files. Application is compiled using icpc(2013) and running on Intel(R) Xeon(R)  E5-2670 processor. if i am using return 0 then application is generating core dump file but instead of return 0; if i use exit(EXIT_SUCCESS); then no core dump files are generated. In case of return 0; application is executed successfully(correct output) without giving segmentation fault or any other error message.

icc -mmic: x86_64-k1om-linux-ld: cannot find -lselinux

compile with "icc -mmic" failed.

 

......
x86_64-k1om-linux-ld: skipping incompatible /usr/lib64/libselinux.so when searching for -lselinux
x86_64-k1om-linux-ld: cannot find -lselinux

......

[root@amax gio]# find / -name libselinux.so
/usr/lib64/libselinux.so

 

---------------------

 

Should I recompile libselinux.so with "icc -mmic"?

I tried to comiple with icc -mmic, and without "-lselinux". And "setenforce 0". Then another error.

x86_64-k1om-linux-ld: warning: xxx.so, needed by yyy.so, not found (try using -rpath or -rpath-link)

 

I use "icc -mmic" to compile it.

Why there are so maney warnings "x86_64-k1om-linux-ld: warning: xxx.so, needed by ../glib/.libs/libglib-2.0.so, not found"?

I'm sure ../glib/.libs/libglib-2.0.so exits.

The errors caused by that?

 

 

S’abonner à Intel® C++ Compiler