Enable newer SSE instructions when code generation flags are enabled

indra/llcommon/lluuid.h

@@ -92,7 +92,7 @@ class LL_COMMON_API LLUUID
 	 */
 	LL_FORCE_INLINE __m128i load_unaligned_si128(const U8* p) const
 	{
-#if defined(AL_AVX)
+#if defined(AL_AVX) || defined(__SSE3__)
 		return _mm_lddqu_si128(reinterpret_cast<const __m128i*>(p));
 #else
 		return _mm_loadu_si128(reinterpret_cast<const __m128i*>(p));
@@ -102,7 +102,7 @@ class LL_COMMON_API LLUUID
 	BOOL isNull() const // Faster than comparing to LLUUID::null.
 	{
 		__m128i mm = load_unaligned_si128(mData);
-#if defined(AL_AVX)
+#if defined(AL_AVX) || defined(__SSE4_1__)
 		return _mm_test_all_zeros(mm, mm) != 0;
 #else
 		mm = _mm_cmpeq_epi8(mm, _mm_setzero_si128());
@@ -126,7 +126,7 @@ class LL_COMMON_API LLUUID
 		__m128i mm_right = load_unaligned_si128(rhs.mData);
 
 		__m128i mm_cmp = _mm_cmpeq_epi32(mm_left, mm_right);
-#if defined(AL_AVX)
+#if defined(AL_AVX) || defined(__SSE4_1__)
 		return _mm_test_all_ones(mm_cmp);
 #else
 		return _mm_movemask_epi8(mm_cmp) == 0xFFFF;

indra/llmath/llvector4a.inl

@@ -248,7 +248,7 @@ inline void LLVector4a::setCross3(const LLVector4a& a, const LLVector4a& b)
 // Set all elements to the dot product of the x, y, and z elements in a and b
 inline void LLVector4a::setAllDot3(const LLVector4a& a, const LLVector4a& b)
 {
-#if AL_AVX
+#if AL_AVX || defined(__SSE4_1__)
 	mQ = _mm_dp_ps(a.mQ, b.mQ, 0x7f);
 #else
 	// ab = { a[W]*b[W], a[Z]*b[Z], a[Y]*b[Y], a[X]*b[X] }
@@ -269,7 +269,7 @@ inline void LLVector4a::setAllDot3(const LLVector4a& a, const LLVector4a& b)
 // Set all elements to the dot product of the x, y, z, and w elements in a and b
 inline void LLVector4a::setAllDot4(const LLVector4a& a, const LLVector4a& b)
 {
-#if AL_AVX
+#if AL_AVX || defined(__SSE4_1__)
 	mQ = _mm_dp_ps(a.mQ, b.mQ, 0xff);
 #else
 	// ab = { a[W]*b[W], a[Z]*b[Z], a[Y]*b[Y], a[X]*b[X] }
@@ -290,7 +290,7 @@ inline void LLVector4a::setAllDot4(const LLVector4a& a, const LLVector4a& b)
 // Return the 3D dot product of this vector and b
 inline LLSimdScalar LLVector4a::dot3(const LLVector4a& b) const
 {
-#if AL_AVX
+#if AL_AVX || defined(__SSE4_1__)
 	return _mm_dp_ps(mQ, b.mQ, 0x7f);
 #else
 	const LLQuad ab = _mm_mul_ps( mQ, b.mQ );
@@ -304,7 +304,7 @@ inline LLSimdScalar LLVector4a::dot3(const LLVector4a& b) const
 // Return the 4D dot product of this vector and b
 inline LLSimdScalar LLVector4a::dot4(const LLVector4a& b) const
 {
-#if AL_AVX
+#if AL_AVX || defined(__SSE4_1__)
 	return _mm_dp_ps(mQ, b.mQ, 0xff);
 #else
 	// ab = { w, z, y, x }

indra/llprimitive/llmaterialid.h

@@ -54,7 +54,7 @@ class LLMaterialID
  */
 	LL_FORCE_INLINE __m128i load_unaligned_si128(const U8* p) const
 	{
-#if defined(AL_AVX)
+#if defined(AL_AVX) || defined(__SSE3__)
 		return _mm_lddqu_si128(reinterpret_cast<const __m128i*>(p));
 #else
 		return _mm_loadu_si128(reinterpret_cast<const __m128i*>(p));
@@ -67,7 +67,7 @@ class LLMaterialID
 		__m128i mm_right = load_unaligned_si128(rhs.mID);
 
 		__m128i mm_cmp = _mm_cmpeq_epi32(mm_left, mm_right);
-#if defined(AL_AVX)
+#if defined(AL_AVX) || defined(__SSE4_1__)
 		return _mm_test_all_ones(mm_cmp);
 #else
 		return _mm_movemask_epi8(mm_cmp) == 0xFFFF;
@@ -131,7 +131,7 @@ class LLMaterialID
 	bool isNull() const
 	{
 		__m128i mm = load_unaligned_si128(mID);
-#if defined(AL_AVX)
+#if defined(AL_AVX) || defined(__SSE4_1__)
 		return _mm_test_all_zeros(mm, mm) != 0;
 #else
 		mm = _mm_cmpeq_epi8(mm, _mm_setzero_si128());