Fix T93283: Cycles render error with CUDA CPU + GPU after recent optimization

BVH2 triangle intersection was broken on the GPU since packed floats can't
be loaded directly into SSE. The better long term solution for performance
would be to build a BVH2 for GPU and Embree for CPU, similar to what we do
for OptiX.

1 files changed, 1 insertions, 13 deletions

diff --git a/intern/cycles/kernel/light/light.h b/intern/cycles/kernel/light/light.h
index 2e7f862a715..3f7d0e0899e 100644
--- a/intern/cycles/kernel/light/light.h
+++ b/intern/cycles/kernel/light/light.h
@@ -676,19 +676,7 @@ ccl_device_forceinline void triangle_light_sample(KernelGlobals kg,
     ls->D = z * B + safe_sqrtf(1.0f - z * z) * safe_normalize(C_ - dot(C_, B) * B);
 
     /* calculate intersection with the planar triangle */
-    if (!ray_triangle_intersect(P,
-                                ls->D,
-                                FLT_MAX,
-#if defined(__KERNEL_SSE2__) && defined(__KERNEL_SSE__)
-                                (ssef *)V,
-#else
-                                V[0],
-                                V[1],
-                                V[2],
-#endif
-                                &ls->u,
-                                &ls->v,
-                                &ls->t)) {
+    if (!ray_triangle_intersect(P, ls->D, FLT_MAX, V[0], V[1], V[2], &ls->u, &ls->v, &ls->t)) {
       ls->pdf = 0.0f;
       return;
     }