I need to do determine processor support for SSE2 prior installing a software. From what I understand, I came up with this:
bool TestSSE2(char * szErrorMsg)
{
__try
{
__asm
{
xorpd xmm0, xmm0 // executing SSE2 instruction
}
}
#pragma warning (suppress: 6320)
__except (EXCEPTION_EXECUTE_HANDLER)
{
if (_exception_code() == STATUS_ILLEGAL_INSTRUCTION)
{
_tcscpy_s(szErrorMsg,MSGSIZE, _T("Streaming SIMD Extensions 2(SSE2) is not supported by the CPU.\r\n Unable to launch APP"));
return false;
}
_tcscpy_s(szErrorMsg,MSGSIZE, _T("Streaming SIMD Extensions 2(SSE2) is not supported by the CPU.\r\n Unable to launch APP"));
return false;
}
return true;
}
Would this work? I'm not really sure how to test, since my CPU supports it, so I don't get false from the function call.
How do I determine processor support for SSE2?
Call CPUID with eax = 1 to load the feature flags in to edx. Bit 26 is set if SSE2 is available. Some code for demonstration purposes, using MSVC++ inline assembly (only for x86 and not portable!):
The most basic way to check for SSE2 support is by using the
CPUID
instruction (on platforms where it is available). Either using inline assembly or using compiler intrinsics.You can use the _cpuid function. All is explained in the MSDN.
I found this one by accident in the MSDN:
Windows-only, but if you are not interested in anything cross-platform, very simple.