Cycles: Use native saturate function for CUDA

This more a workaround for CUDA optimizer which can't optimize clamp(x, 0, 1)
into a single instruction and uses 4 instructions instead.

Original patch by @lockal with own modification:

  Don't make changes outside of the kernel. They don't make any difference
  anyway and term saturate() has a bit different meaning outside of kernel.

This gives around 2% of speedup in Barcelona file, but in more complex shader
setups with lots of math nodes with clamping speedup could be much nicer.

Subscribers: dingto

Projects: #cycles

Differential Revision: https://developer.blender.org/D1224

1 files changed, 2 insertions, 2 deletions

diff --git a/intern/cycles/kernel/svm/svm_closure.h b/intern/cycles/kernel/svm/svm_closure.h
index 8afdb73fb38..0d2d155f827 100644
--- a/intern/cycles/kernel/svm/svm_closure.h
+++ b/intern/cycles/kernel/svm/svm_closure.h
@@ -347,7 +347,7 @@ ccl_device void svm_node_closure_bsdf(KernelGlobals *kg, ShaderData *sd, float *
 				sc->N = N;
 
 				/* sigma */
-				sc->data0 = clamp(param1, 0.0f, 1.0f);
+				sc->data0 = saturate(param1);
 				sc->data1 = 0.0f;
 				sc->data2 = 0.0f;
 				sd->flag |= bsdf_ashikhmin_velvet_setup(sc);
@@ -655,7 +655,7 @@ ccl_device void svm_node_mix_closure(ShaderData *sd, float *stack, uint4 node)
 	decode_node_uchar4(node.y, &weight_offset, &in_weight_offset, &weight1_offset, &weight2_offset);
 
 	float weight = stack_load_float(stack, weight_offset);
-	weight = clamp(weight, 0.0f, 1.0f);
+	weight = saturate(weight);
 
 	float in_weight = (stack_valid(in_weight_offset))? stack_load_float(stack, in_weight_offset): 1.0f;