Wednesday, January 13, 2016

Why enabled pentium4 level optimization and openmp can not use it?




            


I want to optimize the double operation with sse2, but as long as / Qopenmp / QaxN exist, the program up and running on the wrong individual parameters that are so. Read some information, some say caused openmp mkl library and compiled using the conflicts openmp library? But obviously I use intel's own compiler ah? Depressed ing ... brothers to think of a way

Hardware Environment: pentium D820
Software environment: winxp professional sp2, vs2005 (vs2003 installed but not used), intel c / c ++ complier 9.1
Compile parameters: icl -Qvc8 "-Qlocation, link, C: \ Program Files \ Microsoft Visual Studio 8 \ VC \ Bin" / c / O2 / D "WIN32" / D "NDEBUG" / D "_WINDOWS" / D " _MBCS "/ FD / EHsc / MT / GS / GR / Fo" Release / "/ W3 / nologo / Wp64 / Zi / Gd / Qopenmp / QaxN
Reply:
potential_serial.exe The unhandled exception at 0x00401444: 0xC0000005: Access violation reading location
occurs 0xffffffffCompilation error appears as: 00401444 movapd xmmword ptr [esp + edx * 8 + 1F4Ch], xmm7 which esp value 8764464, edx value 0
How 0xffffffff read it?
Reply:
Reduce unnecessary parameters, use the command line to try to compile the following statement:

icl * .cpp / Qopenmp / O2 / QaxN
Reply:
Or not, ah, suspected vs2005 in openmp and intel mkl was a conflict in
Reply:
Multiple 1F4Ch not 16 can not be used __mm_load_pd use __mm_loadu_pd
Reply:
Stumped legendary case 16 aligned, I try to go back, thank you!

No comments:

Post a Comment