Code refactor: use split variance calculation for mega kernels too.

There is no significant difference in denoised benchmark scenes and
denoising ctests, so might as well make it all consistent.

3 files changed, 10 insertions, 20 deletions

diff --git a/intern/cycles/device/device_cpu.cpp b/intern/cycles/device/device_cpu.cpp
index 72330b02a28..ff34f4f9ce4 100644
--- a/intern/cycles/device/device_cpu.cpp
+++ b/intern/cycles/device/device_cpu.cpp
@@ -176,10 +176,10 @@ public:
 	KernelFunctions<void(*)(KernelGlobals *, uchar4 *, float *, float, int, int, int, int)>       convert_to_byte_kernel;
 	KernelFunctions<void(*)(KernelGlobals *, uint4 *, float4 *, float*, int, int, int, int, int)> shader_kernel;
 
-	KernelFunctions<void(*)(int, TilesInfo*, int, int, float*, float*, float*, float*, float*, int*, int, int, bool)> filter_divide_shadow_kernel;
-	KernelFunctions<void(*)(int, TilesInfo*, int, int, int, int, float*, float*, int*, int, int, bool)>               filter_get_feature_kernel;
-	KernelFunctions<void(*)(int, int, float*, float*, float*, float*, int*, int)>                                     filter_detect_outliers_kernel;
-	KernelFunctions<void(*)(int, int, float*, float*, float*, float*, int*, int)>                                     filter_combine_halves_kernel;
+	KernelFunctions<void(*)(int, TilesInfo*, int, int, float*, float*, float*, float*, float*, int*, int, int)> filter_divide_shadow_kernel;
+	KernelFunctions<void(*)(int, TilesInfo*, int, int, int, int, float*, float*, int*, int, int)>               filter_get_feature_kernel;
+	KernelFunctions<void(*)(int, int, float*, float*, float*, float*, int*, int)>                               filter_detect_outliers_kernel;
+	KernelFunctions<void(*)(int, int, float*, float*, float*, float*, int*, int)>                               filter_combine_halves_kernel;
 
 	KernelFunctions<void(*)(int, int, float*, float*, float*, int*, int, int, float, float)> filter_nlm_calc_difference_kernel;
 	KernelFunctions<void(*)(float*, float*, int*, int, int)>                                 filter_nlm_blur_kernel;
@@ -563,8 +563,7 @@ public:
 				                              (float*) buffer_variance_ptr,
 				                              &task->rect.x,
 				                              task->render_buffer.pass_stride,
-				                              task->render_buffer.denoising_data_offset,
-				                              use_split_kernel);
+				                              task->render_buffer.denoising_data_offset);
 			}
 		}
 		return true;
@@ -587,8 +586,7 @@ public:
 				                            (float*) variance_ptr,
 				                            &task->rect.x,
 				                            task->render_buffer.pass_stride,
-				                            task->render_buffer.denoising_data_offset,
-				                            use_split_kernel);
+				                            task->render_buffer.denoising_data_offset);
 			}
 		}
 		return true;
diff --git a/intern/cycles/device/device_cuda.cpp b/intern/cycles/device/device_cuda.cpp
index e5464dcf34e..54e012191ae 100644
--- a/intern/cycles/device/device_cuda.cpp
+++ b/intern/cycles/device/device_cuda.cpp
@@ -1173,7 +1173,6 @@ public:
 		                   task->rect.z-task->rect.x,
 		                   task->rect.w-task->rect.y);
 
-		bool use_split_variance = use_split_kernel();
 		void *args[] = {&task->render_buffer.samples,
 		                &task->tiles_mem.device_pointer,
 		                &a_ptr,
@@ -1183,8 +1182,7 @@ public:
 		                &buffer_variance_ptr,
 		                &task->rect,
 		                &task->render_buffer.pass_stride,
-		                &task->render_buffer.denoising_data_offset,
-		                &use_split_variance};
+		                &task->render_buffer.denoising_data_offset};
 		CUDA_LAUNCH_KERNEL(cuFilterDivideShadow, args);
 		cuda_assert(cuCtxSynchronize());
 
@@ -1209,7 +1207,6 @@ public:
 		                   task->rect.z-task->rect.x,
 		                   task->rect.w-task->rect.y);
 
-		bool use_split_variance = use_split_kernel();
 		void *args[] = {&task->render_buffer.samples,
 		                &task->tiles_mem.device_pointer,
 				        &mean_offset,
@@ -1218,8 +1215,7 @@ public:
 		                &variance_ptr,
 		                &task->rect,
 		                &task->render_buffer.pass_stride,
-		                &task->render_buffer.denoising_data_offset,
-		                &use_split_variance};
+		                &task->render_buffer.denoising_data_offset};
 		CUDA_LAUNCH_KERNEL(cuFilterGetFeature, args);
 		cuda_assert(cuCtxSynchronize());
 
diff --git a/intern/cycles/device/opencl/opencl_base.cpp b/intern/cycles/device/opencl/opencl_base.cpp
index 7bdf81462b8..8095611f099 100644
--- a/intern/cycles/device/opencl/opencl_base.cpp
+++ b/intern/cycles/device/opencl/opencl_base.cpp
@@ -982,7 +982,6 @@ bool OpenCLDeviceBase::denoising_divide_shadow(device_ptr a_ptr,
 
 	cl_kernel ckFilterDivideShadow = denoising_program(ustring("filter_divide_shadow"));
 
-	char split_kernel = is_split_kernel()? 1 : 0;
 	kernel_set_args(ckFilterDivideShadow, 0,
 	                task->render_buffer.samples,
 	                tiles_mem,
@@ -993,8 +992,7 @@ bool OpenCLDeviceBase::denoising_divide_shadow(device_ptr a_ptr,
 	                buffer_variance_mem,
 	                task->rect,
 	                task->render_buffer.pass_stride,
-	                task->render_buffer.denoising_data_offset,
-	                split_kernel);
+	                task->render_buffer.denoising_data_offset);
 	enqueue_kernel(ckFilterDivideShadow,
 	               task->rect.z-task->rect.x,
 	               task->rect.w-task->rect.y);
@@ -1015,7 +1013,6 @@ bool OpenCLDeviceBase::denoising_get_feature(int mean_offset,
 
 	cl_kernel ckFilterGetFeature = denoising_program(ustring("filter_get_feature"));
 
-	char split_kernel = is_split_kernel()? 1 : 0;
 	kernel_set_args(ckFilterGetFeature, 0,
 	                task->render_buffer.samples,
 	                tiles_mem,
@@ -1025,8 +1022,7 @@ bool OpenCLDeviceBase::denoising_get_feature(int mean_offset,
 	                variance_mem,
 	                task->rect,
 	                task->render_buffer.pass_stride,
-	                task->render_buffer.denoising_data_offset,
-	                split_kernel);
+	                task->render_buffer.denoising_data_offset);
 	enqueue_kernel(ckFilterGetFeature,
 	               task->rect.z-task->rect.x,
 	               task->rect.w-task->rect.y);