Apple contribution for OSX SSE and iOS NEON optimizations unit tests, thanks to Jordan Hubbard, Ian Ollmann and Hristo Hristov.

For OSX: cd build ./premake_osx xcode4 for iOS: cd build ./ios_build.sh ./ios_run.sh Also integrated the branches/StackAllocation to make it easier to multi-thread collision detection in the near future. It avoids changing the btCollisionObject while performing collision detection. As this is a large patch, some stuff might be temporarily broken, I'll keep an eye out on issues.
2012-06-07 00:56:30 +00:00
parent 777b92a2ad
commit 73b217fb07
323 changed files with 30730 additions and 13635 deletions
--- a/Test/Source/Tests/Test_3x3mulVM.cpp
+++ b/Test/Source/Tests/Test_3x3mulVM.cpp
@@ -0,0 +1,112 @@
+//
+//  Test_3x3mulVM.cpp
+//  BulletTest
+//
+//  Copyright (c) 2011 Apple Inc.
+//
+
+
+
+#include "LinearMath/btScalar.h"
+#if defined (BT_USE_SSE_IN_API) || defined (BT_USE_NEON)
+
+
+#include "Test_3x3mulVM.h"
+#include "vector.h"
+#include "Utils.h"
+#include "main.h"
+#include <math.h>
+#include <string.h>
+
+#include <LinearMath/btMatrix3x3.h>
+
+#define LOOPCOUNT 1000
+#define ARRAY_SIZE 128
+
+static inline btSimdFloat4 rand_f4(void)
+{
+    return btAssign128( RANDF_01, RANDF_01, RANDF_01, BT_NAN );      // w channel NaN
+}
+
+static btVector3 M3x3mulVM_ref(  const btVector3 &v, const btMatrix3x3 &m)
+{
+	return btVector3(m.tdotx(v), m.tdoty(v), m.tdotz(v));
+}
+
+int Test_3x3mulVM(void)
+{
+    // Init an array flanked by guard pages
+    btVector3   in1[ARRAY_SIZE];
+    btMatrix3x3 in2[ARRAY_SIZE];
+    btVector3   out[ARRAY_SIZE];
+    btVector3   out2[ARRAY_SIZE];
+    
+    // Init the data
+    size_t i, j;
+    for( i = 0; i < ARRAY_SIZE; i++ )
+    {
+        in1[i] = btVector3(rand_f4());   
+        in2[i] = btMatrix3x3(rand_f4(), rand_f4(), rand_f4() );   
+        
+        out[i] = M3x3mulVM_ref(in1[i], in2[i]);
+        out2[i] = (in1[i] * in2[i]);
+
+		if( fabsf(out[i].m_floats[0] - out2[i].m_floats[0]) + 
+			fabsf(out[i].m_floats[1] - out2[i].m_floats[1]) +
+			fabsf(out[i].m_floats[2] - out2[i].m_floats[2]) +
+			fabsf(out[i].m_floats[3] - out2[i].m_floats[3]) > FLT_EPSILON*4 )
+		{	
+			vlog( "Error - M3x3mulVM result error! ");
+            vlog( "failure @ %ld\n", i);
+			vlog( 	"\ncorrect = (%10.4f, %10.4f, %10.4f, %10.4f) "
+					"\ntested  = (%10.4f, %10.4f, %10.4f, %10.4f) \n", 
+					out[i].m_floats[0], out[i].m_floats[1], out[i].m_floats[2], out[i].m_floats[3], 
+					out2[i].m_floats[0], out2[i].m_floats[1], out2[i].m_floats[2], out2[i].m_floats[3]);
+		
+			return 1;
+		}
+    }
+    
+    uint64_t scalarTime, vectorTime;
+    uint64_t startTime, bestTime, currentTime;
+    bestTime = -1LL;
+    scalarTime = 0;
+    for (j = 0; j < LOOPCOUNT; j++) 
+    {
+        startTime = ReadTicks();
+        for( i = 0; i < ARRAY_SIZE; i++ )
+            out[i] = M3x3mulVM_ref(in1[i], in2[i]);
+        currentTime = ReadTicks() - startTime;
+        scalarTime += currentTime;
+        if( currentTime < bestTime )
+            bestTime = currentTime;
+    }
+    if( 0 == gReportAverageTimes )
+        scalarTime = bestTime;        
+    else
+        scalarTime /= LOOPCOUNT;
+    
+    bestTime = -1LL;
+    vectorTime = 0;
+    for (j = 0; j < LOOPCOUNT; j++) 
+    {
+        startTime = ReadTicks();
+        for( i = 0; i < ARRAY_SIZE; i++ )
+            out2[i] = (in1[i] * in2[i]);
+        currentTime = ReadTicks() - startTime;
+        vectorTime += currentTime;
+        if( currentTime < bestTime )
+            bestTime = currentTime;
+    }
+    if( 0 == gReportAverageTimes )
+        vectorTime = bestTime;        
+    else
+        vectorTime /= LOOPCOUNT;
+    
+    vlog( "Timing:\n" );
+    vlog( "\t    scalar\t    vector\n" );
+    vlog( "\t%10.2f\t%10.2f\n", TicksToCycles( scalarTime ) / ARRAY_SIZE, TicksToCycles( vectorTime ) / ARRAY_SIZE );
+    
+    return 0;
+}
+#endif //BT_USE_SSE