Fuse the three jacobian dot products into one horizontal reduction. (#2151)

SolveVelocityConstraintGetTotalLambda computed three independent Vec3::Dot
calls and summed the scalar results. Since Dot(a,b) = sum_xyz(a*b), the sum of
the dots equals a single reduction of the component-wise sum of products, so
the two redundant horizontal reductions can be removed from the latency-bound
critical path. Accumulate component-wise into a Vec3 and call ReduceSum() once.

Performance uplift about 3% for Pyramid test with 300 iterations.

Cross-platform determinism is preserved but results changed (fewer roundings),
so the determinism check hashes have been regenerated.

Author: asenkew

.github/workflows/determinism_check.yml

@@ -1,16 +1,16 @@
 name: Determinism Check
 
 env:
-  CONVEX_VS_MESH_HASH: '0x918f27cacae9752b'
-  HIGH_SPEED_HASH: '0x58d8d892536f4021'
-  RAGDOLL_HASH: '0x3be6ea76c07a2ac8'
-  PYRAMID_HASH: '0x74d0118836ac0892'
-  CHARACTER_VIRTUAL_HASH: '0xc1d4e05b164b1221'
-  CONVEX_VS_MESH_HASH_DOUBLE: '0x3729e7890b93a289'
-  HIGH_SPEED_HASH_DOUBLE: '0x183f48cb94739ce'
-  RAGDOLL_HASH_DOUBLE: '0x8dd6bc501859a228'
-  PYRAMID_HASH_DOUBLE: '0xf07f04597a102a82'
-  CHARACTER_VIRTUAL_HASH_DOUBLE: '0xa0b16727aa16fdb0'
+  CONVEX_VS_MESH_HASH: '0x30ea0fd16accfad0'
+  HIGH_SPEED_HASH: '0x64d4f332a3a94503'
+  RAGDOLL_HASH: '0x82a4ecc9e5424430'
+  PYRAMID_HASH: '0x7d979c44bb41a1e4'
+  CHARACTER_VIRTUAL_HASH: '0x37d910bfeae010e8'
+  CONVEX_VS_MESH_HASH_DOUBLE: '0x725342da41935e02'
+  HIGH_SPEED_HASH_DOUBLE: '0xd0e6a90aa77edbf1'
+  RAGDOLL_HASH_DOUBLE: '0x3c5e5cdf1473a471'
+  PYRAMID_HASH_DOUBLE: '0xcd3653dde7ab6582'
+  CHARACTER_VIRTUAL_HASH_DOUBLE: '0x9b1c94e2c4b6db97'
   EMSCRIPTEN_VERSION: 6.0.8
   NODE_VERSION: 24.x
   UBUNTU_CLANG_VERSION: clang++-18

Jolt/Physics/Constraints/ConstraintPart/ContactConstraintPart.h

@@ -169,21 +169,22 @@ class ContactConstraintPart : public ContactConstraintPart1<Type1>, public Conta
 	JPH_INLINE float			SolveVelocityConstraintGetTotalLambda(Vec3Arg inLinearVelocity1, Vec3Arg inAngularVelocity1, Vec3Arg inLinearVelocity2, Vec3Arg inAngularVelocity2, Vec3Arg inWorldSpaceAxis) const
 	{
 		// Calculate jacobian multiplied by linear velocity
-		float jv;
+		Vec3 acc;
 		if constexpr (Type1 != EMotionType::Static && Type2 != EMotionType::Static)
-			jv = inWorldSpaceAxis.Dot(inLinearVelocity1 - inLinearVelocity2);
+			acc = inWorldSpaceAxis * (inLinearVelocity1 - inLinearVelocity2);
 		else if constexpr (Type1 != EMotionType::Static)
-			jv = inWorldSpaceAxis.Dot(inLinearVelocity1);
+			acc = inWorldSpaceAxis * inLinearVelocity1;
 		else if constexpr (Type2 != EMotionType::Static)
-			jv = inWorldSpaceAxis.Dot(-inLinearVelocity2);
+			acc = inWorldSpaceAxis * (-inLinearVelocity2);
 		else
 			JPH_ASSERT(false); // Static vs static is nonsensical!
 
-		// Calculate jacobian multiplied by angular velocity
 		if constexpr (Type1 != EMotionType::Static)
-			jv += Vec3::sLoadFloat3Unsafe(this->mR1PlusUxAxis).Dot(inAngularVelocity1);
+			acc += Vec3::sLoadFloat3Unsafe(this->mR1PlusUxAxis) * inAngularVelocity1;
 		if constexpr (Type2 != EMotionType::Static)
-			jv -= Vec3::sLoadFloat3Unsafe(this->mR2xAxis).Dot(inAngularVelocity2);
+			acc -= Vec3::sLoadFloat3Unsafe(this->mR2xAxis) * inAngularVelocity2;
+
+		float jv = acc.ReduceSum();
 
 		// Lagrange multiplier is:
 		//