Fix #35684: cycles unable to use full 6GB of memory on NVidia Titan GPU. We now

use arrays instead of textures for general storage on this card (image textures
are still stored as texture). Textures were found to be faster on older cards,
but the limits on 1D texture size have not increased along with the memory size,
which meant that the full 6 GB could not be used.

The performance actually seems to be slightly better with arrays in some tests
on Titan. For older cards there seems to be a bit of a mix, some are better and
others not. We may change those to use arrays too, but more testing is needed,
only Titan and Tesla K20 (sm_35) is changed for now.

The fact that arrays are faster is a bit surprising, as others found textures
to be faster on Kepler. However even if they were, the memory limitation is
more important to solve anyway.
https://research.nvidia.com/publication/understanding-efficiency-ray-traversal-gpus-kepler-and-fermi-addendum

1 files changed, 4 insertions, 0 deletions

diff --git a/intern/cycles/kernel/kernel_globals.h b/intern/cycles/kernel/kernel_globals.h
index ab0a717b592..b5e691eb615 100644
--- a/intern/cycles/kernel/kernel_globals.h
+++ b/intern/cycles/kernel/kernel_globals.h
@@ -66,7 +66,11 @@ typedef struct KernelGlobals {
 __constant__ KernelData __data;
 typedef struct KernelGlobals {} KernelGlobals;
 
+#ifdef __KERNEL_CUDA_TEX_STORAGE__
 #define KERNEL_TEX(type, ttype, name) ttype name;
+#else
+#define KERNEL_TEX(type, ttype, name) const __constant__ __device__ type *name;
+#endif
 #define KERNEL_IMAGE_TEX(type, ttype, name) ttype name;
 #include "kernel_textures.h"