Cycles: Split path initialization into own kernel

This makes it easier to initialize things correctly in the data_init kernel
before they are needed by path tracing.

2 files changed, 104 insertions, 101 deletions

diff --git a/intern/cycles/kernel/split/kernel_data_init.h b/intern/cycles/kernel/split/kernel_data_init.h
index 5604363dcd9..982c7be2008 100644
--- a/intern/cycles/kernel/split/kernel_data_init.h
+++ b/intern/cycles/kernel/split/kernel_data_init.h
@@ -18,33 +18,6 @@ CCL_NAMESPACE_BEGIN
 
 /* Note on kernel_data_initialization kernel
  * This kernel Initializes structures needed in path-iteration kernels.
- * This is the first kernel in ray-tracing logic.
- *
- * Ray state of rays outside the tile-boundary will be marked RAY_INACTIVE
- *
- * Its input and output are as follows,
- *
- * Un-initialized rng---------------|--- kernel_data_initialization ---|--- Initialized rng
- * Un-initialized throughput -------|                                  |--- Initialized throughput
- * Un-initialized L_transparent ----|                                  |--- Initialized L_transparent
- * Un-initialized PathRadiance -----|                                  |--- Initialized PathRadiance
- * Un-initialized Ray --------------|                                  |--- Initialized Ray
- * Un-initialized PathState --------|                                  |--- Initialized PathState
- * Un-initialized QueueData --------|                                  |--- Initialized QueueData (to QUEUE_EMPTY_SLOT)
- * Un-initialized QueueIndex -------|                                  |--- Initialized QueueIndex (to 0)
- * Un-initialized use_queues_flag---|                                  |--- Initialized use_queues_flag (to false)
- * Un-initialized ray_state --------|                                  |--- Initialized ray_state
- * parallel_samples --------------- |                                  |--- Initialized per_sample_output_buffers
- * rng_state -----------------------|                                  |--- Initialized work_array
- * data ----------------------------|                                  |--- Initialized work_pool_wgs
- * start_sample --------------------|                                  |
- * sx ------------------------------|                                  |
- * sy ------------------------------|                                  |
- * sw ------------------------------|                                  |
- * sh ------------------------------|                                  |
- * stride --------------------------|                                  |
- * queuesize -----------------------|                                  |
- * num_samples ---------------------|                                  |
  *
  * Note on Queues :
  * All slots in queues are initialized to queue empty slot;
@@ -137,80 +110,6 @@ void KERNEL_FUNCTION_FULL_NAME(data_init)(
 		 */
 		*use_queues_flag = 0;
 	}
-
-	int ray_index = ccl_global_id(0) + ccl_global_id(1) * ccl_global_size(0);
-
-	/* This is the first assignment to ray_state;
-	 * So we dont use ASSIGN_RAY_STATE macro.
-	 */
-	kernel_split_state.ray_state[ray_index] = RAY_ACTIVE;
-
-	unsigned int my_sample;
-	unsigned int pixel_x;
-	unsigned int pixel_y;
-	unsigned int tile_x;
-	unsigned int tile_y;
-	unsigned int my_sample_tile;
-
-	unsigned int work_index = 0;
-	/* Get work. */
-	if(!get_next_work(kg, &work_index, ray_index)) {
-		/* No more work, mark ray as inactive */
-		kernel_split_state.ray_state[ray_index] = RAY_INACTIVE;
-
-		return;
-	}
-
-	/* Get the sample associated with the work. */
-	my_sample = get_work_sample(kg, work_index, ray_index) + start_sample;
-
-	my_sample_tile = 0;
-
-	/* Get pixel and tile position associated with the work. */
-	get_work_pixel_tile_position(kg, &pixel_x, &pixel_y,
-	                        &tile_x, &tile_y,
-	                        work_index,
-	                        ray_index);
-	kernel_split_state.work_array[ray_index] = work_index;
-
-	rng_state += kernel_split_params.offset + pixel_x + pixel_y*stride;
-
-	ccl_global float *per_sample_output_buffers = kernel_split_state.per_sample_output_buffers;
-	per_sample_output_buffers += ((tile_x + (tile_y * stride)) + (my_sample_tile)) * kernel_data.film.pass_stride;
-
-	/* Initialize random numbers and ray. */
-	kernel_path_trace_setup(kg,
-	                        rng_state,
-	                        my_sample,
-	                        pixel_x, pixel_y,
-	                        &kernel_split_state.rng[ray_index],
-	                        &kernel_split_state.ray[ray_index]);
-
-	if(kernel_split_state.ray[ray_index].t != 0.0f) {
-		/* Initialize throughput, L_transparent, Ray, PathState;
-		 * These rays proceed with path-iteration.
-		 */
-		kernel_split_state.throughput[ray_index] = make_float3(1.0f, 1.0f, 1.0f);
-		kernel_split_state.L_transparent[ray_index] = 0.0f;
-		path_radiance_init(&kernel_split_state.path_radiance[ray_index], kernel_data.film.use_light_pass);
-		path_state_init(kg,
-		                &kernel_split_state.sd_DL_shadow[ray_index],
-		                &kernel_split_state.path_state[ray_index],
-		                &kernel_split_state.rng[ray_index],
-		                my_sample,
-		                &kernel_split_state.ray[ray_index]);
-#ifdef __KERNEL_DEBUG__
-		debug_data_init(&kernel_split_state.debug_data[ray_index]);
-#endif
-	}
-	else {
-		/* These rays do not participate in path-iteration. */
-		float4 L_rad = make_float4(0.0f, 0.0f, 0.0f, 0.0f);
-		/* Accumulate result in output buffer. */
-		kernel_write_pass_float4(per_sample_output_buffers, my_sample, L_rad);
-		path_rng_end(kg, rng_state, kernel_split_state.rng[ray_index]);
-		ASSIGN_RAY_STATE(kernel_split_state.ray_state, ray_index, RAY_TO_REGENERATE);
-	}
 }
 
 CCL_NAMESPACE_END
diff --git a/intern/cycles/kernel/split/kernel_path_init.h b/intern/cycles/kernel/split/kernel_path_init.h
new file mode 100644
index 00000000000..e613db214ed
--- /dev/null
+++ b/intern/cycles/kernel/split/kernel_path_init.h
@@ -0,0 +1,104 @@
+/*
+ * Copyright 2011-2017 Blender Foundation
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+CCL_NAMESPACE_BEGIN
+
+/* This kernel initializes structures needed in path-iteration kernels.
+ * This is the first kernel in ray-tracing logic.
+ *
+ * Ray state of rays outside the tile-boundary will be marked RAY_INACTIVE
+ */
+
+ccl_device void kernel_path_init(KernelGlobals *kg) {
+	int ray_index = ccl_global_id(0) + ccl_global_id(1) * ccl_global_size(0);
+
+	/* This is the first assignment to ray_state;
+	 * So we dont use ASSIGN_RAY_STATE macro.
+	 */
+	kernel_split_state.ray_state[ray_index] = RAY_ACTIVE;
+
+	unsigned int my_sample;
+	unsigned int pixel_x;
+	unsigned int pixel_y;
+	unsigned int tile_x;
+	unsigned int tile_y;
+	unsigned int my_sample_tile;
+
+	unsigned int work_index = 0;
+	/* Get work. */
+	if(!get_next_work(kg, &work_index, ray_index)) {
+		/* No more work, mark ray as inactive */
+		kernel_split_state.ray_state[ray_index] = RAY_INACTIVE;
+
+		return;
+	}
+
+	/* Get the sample associated with the work. */
+	my_sample = get_work_sample(kg, work_index, ray_index) + kernel_split_params.start_sample;
+
+	my_sample_tile = 0;
+
+	/* Get pixel and tile position associated with the work. */
+	get_work_pixel_tile_position(kg, &pixel_x, &pixel_y,
+	                        &tile_x, &tile_y,
+	                        work_index,
+	                        ray_index);
+	kernel_split_state.work_array[ray_index] = work_index;
+
+	ccl_global uint *rng_state = kernel_split_params.rng_state;
+	rng_state += kernel_split_params.offset + pixel_x + pixel_y*kernel_split_params.stride;
+
+	ccl_global float *per_sample_output_buffers = kernel_split_state.per_sample_output_buffers;
+	per_sample_output_buffers += (tile_x + tile_y * kernel_split_params.stride + my_sample_tile)
+	                             * kernel_data.film.pass_stride;
+
+	/* Initialize random numbers and ray. */
+	kernel_path_trace_setup(kg,
+	                        rng_state,
+	                        my_sample,
+	                        pixel_x, pixel_y,
+	                        &kernel_split_state.rng[ray_index],
+	                        &kernel_split_state.ray[ray_index]);
+
+	if(kernel_split_state.ray[ray_index].t != 0.0f) {
+		/* Initialize throughput, L_transparent, Ray, PathState;
+		 * These rays proceed with path-iteration.
+		 */
+		kernel_split_state.throughput[ray_index] = make_float3(1.0f, 1.0f, 1.0f);
+		kernel_split_state.L_transparent[ray_index] = 0.0f;
+		path_radiance_init(&kernel_split_state.path_radiance[ray_index], kernel_data.film.use_light_pass);
+		path_state_init(kg,
+		                &kernel_split_state.sd_DL_shadow[ray_index],
+		                &kernel_split_state.path_state[ray_index],
+		                &kernel_split_state.rng[ray_index],
+		                my_sample,
+		                &kernel_split_state.ray[ray_index]);
+#ifdef __KERNEL_DEBUG__
+		debug_data_init(&kernel_split_state.debug_data[ray_index]);
+#endif
+	}
+	else {
+		/* These rays do not participate in path-iteration. */
+		float4 L_rad = make_float4(0.0f, 0.0f, 0.0f, 0.0f);
+		/* Accumulate result in output buffer. */
+		kernel_write_pass_float4(per_sample_output_buffers, my_sample, L_rad);
+		path_rng_end(kg, rng_state, kernel_split_state.rng[ray_index]);
+		ASSIGN_RAY_STATE(kernel_split_state.ray_state, ray_index, RAY_TO_REGENERATE);
+	}
+}
+
+CCL_NAMESPACE_END
+