https://community.intel.com/t5/Software-Tuning-Performance/Test-results-for-CRT-function-sqrt-for-different-Floating-Point/m-p/942770#M2009 Intel C++ compiler supports <STRONG>four</STRONG> Floating Point Models: <STRONG>Fast</STRONG>, <STRONG>Precise</STRONG>, <STRONG>Strict</STRONG> and <STRONG>Fast=2</STRONG> and test results demonstrate how performance is affected when CRT-function <STRONG>sqrt</STRONG> is called ( results are in ASC order ): <STRONG>[ Fast=2 (/fp:fast=2) [Intel C++] ]</STRONG> 32-bit Windows platform CRT Sqrt - float - Calculating the Square Roots - 203 ticks Last Result: 134217728.000^0.5 = 11585.236 CRT Sqrt - double - Calculating the Square Roots - 750 ticks Last Result: 134217728.000^0.5 = 11585.237 <STRONG>[ Fast (/fp:fast) ]</STRONG> 32-bit Windows platform CRT Sqrt - float - Calculating the Square Roots - 219 ticks Last Result: 134217728.000^0.5 = 11585.236 CRT Sqrt - double - Calculating the Square Roots - 750 ticks Last Result: 134217728.000^0.5 = 11585.237 <STRONG>[ Precise (/fp:precise) ]</STRONG> 32-bit Windows platform CRT Sqrt - float - Calculating the Square Roots - 422 ticks Last Result: 134217728.000^0.5 = 11585.237 CRT Sqrt - double - Calculating the Square Roots - 750 ticks Last Result: 134217728.000^0.5 = 11585.237 <STRONG>[ Strict (/fp:strict) ]</STRONG> 32-bit Windows platform CRT Sqrt - float - Calculating the Square Roots - 875 ticks Last Result: 134217728.000^0.5 = 11585.237 CRT Sqrt - double - Calculating the Square Roots - 2969 ticks Last Result: 134217728.000^0.5 = 11585.237 As you can see <STRONG>Fast=2</STRONG> is the fastest Floating Point Model, and <STRONG>Strict</STRONG> is the slowest ( slower in 875 / 203 = ~4.3x ). Wed, 13 Feb 2013 01:55:10 GMT SergeyKostrov 2013-02-13T01:55:10Z https://community.intel.com/t5/Software-Tuning-Performance/Test-results-for-CRT-function-sqrt-for-different-Floating-Point/m-p/942770#M2009 Intel C++ compiler supports <STRONG>four</STRONG> Floating Point Models: <STRONG>Fast</STRONG>, <STRONG>Precise</STRONG>, <STRONG>Strict</STRONG> and <STRONG>Fast=2</STRONG> and test results demonstrate how performance is affected when CRT-function <STRONG>sqrt</STRONG> is called ( results are in ASC order ): <STRONG>[ Fast=2 (/fp:fast=2) [Intel C++] ]</STRONG> 32-bit Windows platform CRT Sqrt - float - Calculating the Square Roots - 203 ticks Last Result: 134217728.000^0.5 = 11585.236 CRT Sqrt - double - Calculating the Square Roots - 750 ticks Last Result: 134217728.000^0.5 = 11585.237 <STRONG>[ Fast (/fp:fast) ]</STRONG> 32-bit Windows platform CRT Sqrt - float - Calculating the Square Roots - 219 ticks Last Result: 134217728.000^0.5 = 11585.236 CRT Sqrt - double - Calculating the Square Roots - 750 ticks Last Result: 134217728.000^0.5 = 11585.237 <STRONG>[ Precise (/fp:precise) ]</STRONG> 32-bit Windows platform CRT Sqrt - float - Calculating the Square Roots - 422 ticks Last Result: 134217728.000^0.5 = 11585.237 CRT Sqrt - double - Calculating the Square Roots - 750 ticks Last Result: 134217728.000^0.5 = 11585.237 <STRONG>[ Strict (/fp:strict) ]</STRONG> 32-bit Windows platform CRT Sqrt - float - Calculating the Square Roots - 875 ticks Last Result: 134217728.000^0.5 = 11585.237 CRT Sqrt - double - Calculating the Square Roots - 2969 ticks Last Result: 134217728.000^0.5 = 11585.237 As you can see <STRONG>Fast=2</STRONG> is the fastest Floating Point Model, and <STRONG>Strict</STRONG> is the slowest ( slower in 875 / 203 = ~4.3x ). Wed, 13 Feb 2013 01:55:10 GMT https://community.intel.com/t5/Software-Tuning-Performance/Test-results-for-CRT-function-sqrt-for-different-Floating-Point/m-p/942770#M2009 SergeyKostrov 2013-02-13T01:55:10Z https://community.intel.com/t5/Software-Tuning-Performance/Test-results-for-CRT-function-sqrt-for-different-Floating-Point/m-p/942771#M2010 Here are source codes of the test case: #ifndef _WIN32_WINNT #define _WIN32_WINNT 0x0501 #endif #include "stdio.h" #include "tchar.h" #include "float.h" #include "math.h" #include "windows.h" #ifdef _WIN32_PLATFORM #define _WIN_PLATFORM _T("32-bit Windows platform") #endif #ifdef _WIN64_PLATFORM #define _WIN_PLATFORM _T("64-bit Windows platform") #endif #define _USE_CRT_SQRT_FLOAT_ #define _USE_CRT_SQRT_DOUBLE_ int <STRONG>_tmain</STRONG>( void ) { <STRONG>_tprintf</STRONG>( _T("%s\n"), _WIN_PLATFORM ); // Initializations unsigned int uiTicksStart = 0U; unsigned int uiControlWordx87 = 0U; int iV = 0; int iNumberOfIterations = 134217728; // 2^27 int iDummy = -1; float fRes = 0.0f; double dRes = 0.0L; uiControlWordx87 = <STRONG>_control87</STRONG>( _PC_53, _MCW_PC ); // <STRONG>Sub-Test 1 - CRT sqrt - float</STRONG> { #ifdef _USE_CRT_SQRT_FLOAT_ <STRONG>_tprintf</STRONG>( _T("CRT Sqrt - float - Calculating the Square Roots") ); uiTicksStart = ::<STRONG>GetTickCount</STRONG>(); for( iV = 0; iV &lt; iNumberOfIterations; iV++ ) { iDummy = -1; fRes = <STRONG>sqrt</STRONG>( ( float )iV ); } <STRONG>_tprintf</STRONG>( _T(" - %ld ticks\n"), ( int )( ::<STRONG>GetTickCount()</STRONG> - uiTicksStart ) ); <STRONG>_tprintf</STRONG>( _T("\tLast Result: %.3f^0.5 = %.3f\n"), ( float )iV, fRes ); #endif } // <STRONG>Sub-Test 2 - CRT sqrt - double</STRONG> { #ifdef _USE_CRT_SQRT_DOUBLE_ <STRONG>_tprintf</STRONG>( _T("CRT Sqrt - double - Calculating the Square Roots") ); uiTicksStart = ::<STRONG>GetTickCount</STRONG>(); for( iV = 0; iV &lt; iNumberOfIterations; iV++ ) { iDummy = -1; dRes = <STRONG>sqrt</STRONG>( ( double )iV ); } <STRONG>_tprintf</STRONG>( _T(" - %ld ticks\n"), ( int )( ::<STRONG>GetTickCount</STRONG>() - uiTicksStart ) ); <STRONG>_tprintf</STRONG>( _T("\tLast Result: %.3f^0.5 = %.3f\n"), ( double )iV, dRes ); #endif } <STRONG>_tprintf</STRONG>( _T("Press ESC to Exit...\n") ); <STRONG>_gettch</STRONG>(); return ( int )0; } Wed, 13 Feb 2013 02:02:00 GMT https://community.intel.com/t5/Software-Tuning-Performance/Test-results-for-CRT-function-sqrt-for-different-Floating-Point/m-p/942771#M2010 SergeyKostrov 2013-02-13T02:02:00Z