Add optimized noise implementation from Intel for AVX2/FMA3-capable CPUs.

Also overhauled optimized noise for AVX/FMA4-capable CPUs, as well as CPU type identification code.

Author: c-lipka

platform/x86/avx2fma3noise.cpp

@@ -0,0 +1,64 @@
+//******************************************************************************
+///
+/// @file platform/x86/avx2fma3noise.h
+///
+/// This file contains declarations related to implementations of the noise
+/// generator optimized for the AVX2 and FMA3 instruction set.
+///
+/// @copyright
+/// @parblock
+///
+/// Persistence of Vision Ray Tracer ('POV-Ray') version 3.7.
+/// Copyright 1991-2015 Persistence of Vision Raytracer Pty. Ltd.
+///
+/// POV-Ray is free software: you can redistribute it and/or modify
+/// it under the terms of the GNU Affero General Public License as
+/// published by the Free Software Foundation, either version 3 of the
+/// License, or (at your option) any later version.
+///
+/// POV-Ray is distributed in the hope that it will be useful,
+/// but WITHOUT ANY WARRANTY; without even the implied warranty of
+/// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+/// GNU Affero General Public License for more details.
+///
+/// You should have received a copy of the GNU Affero General Public License
+/// along with this program.  If not, see <http://www.gnu.org/licenses/>.
+///
+/// ----------------------------------------------------------------------------
+///
+/// POV-Ray is based on the popular DKB raytracer version 2.12.
+/// DKBTrace was originally written by David K. Buck.
+/// DKBTrace Ver 2.0-2.12 were written by David K. Buck & Aaron A. Collins.
+///
+/// @endparblock
+///
+//******************************************************************************
+
+#ifndef POVRAY_AVX2FMA3NOISE_H
+#define POVRAY_AVX2FMA3NOISE_H
+
+#include "syspovconfigbase.h"
+#include "backend/frame.h"
+
+#ifdef TRY_OPTIMIZED_NOISE_AVX2FMA3
+
+namespace pov
+{
+
+bool AVX2FMA3NoiseSupported();
+
+void AVX2FMA3NoiseInit();
+
+/// Optimized Noise function using AVX2 and FMA3 instructions.
+/// @author Optimized by Intel
+DBL AVX2FMA3Noise(const Vector3d& EPoint, int noise_generator);
+
+/// Optimized DNoise function using AVX2 and FMA3 instructions.
+/// @author Optimized by Intel
+void AVX2FMA3DNoise(Vector3d& result, const Vector3d& EPoint);
+
+}
+
+#endif // TRY_OPTIMIZED_NOISE_AVX2FMA3
+
+#endif // POVRAY_AVX2FMA3NOISE_H

platform/x86/avx2fma3noise.h

@@ -5,6 +5,8 @@
 /// This file contains implementations of the noise generator optimized for the
 /// AVX and FMA4 instruction set.
 ///
+/// @author Original optimizations by AMD
+///
 /// @copyright
 /// @parblock
 ///
@@ -37,13 +39,17 @@
 #include "syspovconfigbase.h"
 #include "avxfma4noise.h"
 
+#ifdef MACHINE_INTRINSICS_H
+#include MACHINE_INTRINSICS_H
+#endif
+
 #include "core/material/pattern.h"
 #include "core/material/texture.h"
 #include "cpuid.h"
 
 /*****************************************************************************/
 
-#ifdef TRY_OPTIMIZED_NOISE
+#ifdef TRY_OPTIMIZED_NOISE_AVXFMA4
 
 /********************************************************************************************/
 /* AMD Specific optimizations: Its found that more than 50% of the time is spent in         */
@@ -688,5 +694,5 @@ void AVXFMA4DNoise(Vector3d& result, const Vector3d& EPoint)
 
 }
 
-#endif
+#endif // TRY_OPTIMIZED_NOISE_AVXFMA4
 

platform/x86/avxfma4noise.cpp

@@ -40,7 +40,7 @@
 #include "syspovconfigbase.h"
 #include "backend/frame.h"
 
-#ifdef TRY_OPTIMIZED_NOISE
+#ifdef TRY_OPTIMIZED_NOISE_AVXFMA4
 
 namespace pov
 {
@@ -58,6 +58,6 @@ void AVXFMA4DNoise(Vector3d& result, const Vector3d& EPoint);
 
 }
 
-#endif // TRY_OPTIMIZED_NOISE
+#endif // TRY_OPTIMIZED_NOISE_AVXFMA4
 
 #endif // POVRAY_AVXFMA4NOISE_H