Cycles: add packed_float3 type for storage

Introduce a packed_float3 type for smaller storage that is exactly 3
floats, instead of 4. For computation float3 is still used since it can
use SIMD instructions.

Ref T92212

Differential Revision: https://developer.blender.org/D13243

1 files changed, 26 insertions, 0 deletions

diff --git a/intern/cycles/util/math_float3.h b/intern/cycles/util/math_float3.h
index 81550c5d03c..031aac1b5d4 100644
--- a/intern/cycles/util/math_float3.h
+++ b/intern/cycles/util/math_float3.h
@@ -222,6 +222,32 @@ ccl_device_inline float3 operator/=(float3 &a, float f)
   return a = a * invf;
 }
 
+#if !(defined(__KERNEL_METAL__) || defined(__KERNEL_CUDA__))
+ccl_device_inline packed_float3 operator*=(packed_float3 &a, const float3 &b)
+{
+  a = float3(a) * b;
+  return a;
+}
+
+ccl_device_inline packed_float3 operator*=(packed_float3 &a, float f)
+{
+  a = float3(a) * f;
+  return a;
+}
+
+ccl_device_inline packed_float3 operator/=(packed_float3 &a, const float3 &b)
+{
+  a = float3(a) / b;
+  return a;
+}
+
+ccl_device_inline packed_float3 operator/=(packed_float3 &a, float f)
+{
+  a = float3(a) / f;
+  return a;
+}
+#endif
+
 ccl_device_inline bool operator==(const float3 &a, const float3 &b)
 {
 #ifdef __KERNEL_SSE__