Cycles: Tweaks to support CUDA 8 toolkit

All the changes are mainly giving explicit tips on inlining functions,
so they match how inlining worked with previous toolkit.

This make kernel compiled by CUDA 8 render in average with same speed
as previous kernels. Some scenes are somewhat faster, some of them are
somewhat slower. But slowdown is within 1% so far.

On a positive side it allows us to enable newer generation cards on
buildbots (so GTX 10x0 will be officially supported soon).

5 files changed, 53 insertions, 28 deletions

diff --git a/intern/cycles/kernel/bvh/bvh_shadow_all.h b/intern/cycles/kernel/bvh/bvh_shadow_all.h
index 1d6fa303d3e..e9eeff31ecc 100644
--- a/intern/cycles/kernel/bvh/bvh_shadow_all.h
+++ b/intern/cycles/kernel/bvh/bvh_shadow_all.h
@@ -37,11 +37,16 @@
  *
  */
 
-ccl_device bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
-                                            const Ray *ray,
-                                            Intersection *isect_array,
-                                            const uint max_hits,
-                                            uint *num_hits)
+#ifndef __KERNEL_GPU__
+ccl_device
+#else
+ccl_device_inline
+#endif
+bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
+                                 const Ray *ray,
+                                 Intersection *isect_array,
+                                 const uint max_hits,
+                                 uint *num_hits)
 {
 	/* todo:
 	 * - likely and unlikely for if() statements
diff --git a/intern/cycles/kernel/bvh/bvh_subsurface.h b/intern/cycles/kernel/bvh/bvh_subsurface.h
index 18978efcfa3..d9623c94b2e 100644
--- a/intern/cycles/kernel/bvh/bvh_subsurface.h
+++ b/intern/cycles/kernel/bvh/bvh_subsurface.h
@@ -35,12 +35,17 @@
  *
  */
 
-ccl_device void BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
-                                            const Ray *ray,
-                                            SubsurfaceIntersection *ss_isect,
-                                            int subsurface_object,
-                                            uint *lcg_state,
-                                            int max_hits)
+#ifndef __KERNEL_GPU__
+ccl_device
+#else
+ccl_device_inline
+#endif
+void BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
+                                 const Ray *ray,
+                                 SubsurfaceIntersection *ss_isect,
+                                 int subsurface_object,
+                                 uint *lcg_state,
+                                 int max_hits)
 {
 	/* todo:
 	 * - test if pushing distance on the stack helps (for non shadow rays)
diff --git a/intern/cycles/kernel/bvh/bvh_traversal.h b/intern/cycles/kernel/bvh/bvh_traversal.h
index 68a11b65ad7..b1a52968a26 100644
--- a/intern/cycles/kernel/bvh/bvh_traversal.h
+++ b/intern/cycles/kernel/bvh/bvh_traversal.h
@@ -40,16 +40,21 @@
  *
  */
 
-ccl_device bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
-                                            const Ray *ray,
-                                            Intersection *isect,
-                                            const uint visibility
+#ifndef __KERNEL_GPU__
+ccl_device
+#else
+ccl_device_inline
+#endif
+bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
+                                 const Ray *ray,
+                                 Intersection *isect,
+                                 const uint visibility
 #if BVH_FEATURE(BVH_HAIR_MINIMUM_WIDTH)
-                                            , uint *lcg_state,
-                                            float difl,
-                                            float extmax
+                                 , uint *lcg_state,
+                                 float difl,
+                                 float extmax
 #endif
-                                            )
+                                 )
 {
 	/* todo:
 	 * - test if pushing distance on the stack helps (for non shadow rays)
diff --git a/intern/cycles/kernel/bvh/bvh_volume.h b/intern/cycles/kernel/bvh/bvh_volume.h
index 03499e94347..107373c17dc 100644
--- a/intern/cycles/kernel/bvh/bvh_volume.h
+++ b/intern/cycles/kernel/bvh/bvh_volume.h
@@ -36,10 +36,15 @@
  *
  */
 
-ccl_device bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
-                                            const Ray *ray,
-                                            Intersection *isect,
-                                            const uint visibility)
+#ifndef __KERNEL_GPU__
+ccl_device
+#else
+ccl_device_inline
+#endif
+bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
+                                 const Ray *ray,
+                                 Intersection *isect,
+                                 const uint visibility)
 {
 	/* todo:
 	 * - test if pushing distance on the stack helps (for non shadow rays)
diff --git a/intern/cycles/kernel/bvh/bvh_volume_all.h b/intern/cycles/kernel/bvh/bvh_volume_all.h
index 7eddc2891d0..1f6515c9862 100644
--- a/intern/cycles/kernel/bvh/bvh_volume_all.h
+++ b/intern/cycles/kernel/bvh/bvh_volume_all.h
@@ -36,11 +36,16 @@
  *
  */
 
-ccl_device uint BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
-                                            const Ray *ray,
-                                            Intersection *isect_array,
-                                            const uint max_hits,
-                                            const uint visibility)
+#ifndef __KERNEL_GPU__
+ccl_device
+#else
+ccl_device_inline
+#endif
+uint BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg,
+                                 const Ray *ray,
+                                 Intersection *isect_array,
+                                 const uint max_hits,
+                                 const uint visibility)
 {
 	/* todo:
 	 * - test if pushing distance on the stack helps (for non shadow rays)