Cycles: OpenCL kernel split

This commit contains all the work related on the AMD megakernel split work
which was mainly done by Varun Sundar, George Kyriazis and Lenny Wang, plus
some help from Sergey Sharybin, Martijn Berger, Thomas Dinges and likely
someone else which we're forgetting to mention.

Currently only AMD cards are enabled for the new split kernel, but it is
possible to force split opencl kernel to be used by setting the following
environment variable: CYCLES_OPENCL_SPLIT_KERNEL_TEST=1.

Not all the features are supported yet, and that being said no motion blur,
camera blur, SSS and volumetrics for now. Also transparent shadows are
disabled on AMD device because of some compiler bug.

This kernel is also only implements regular path tracing and supporting
branched one will take a bit. Branched path tracing is exposed to the
interface still, which is a bit misleading and will be hidden there soon.

More feature will be enabled once they're ported to the split kernel and
tested.

Neither regular CPU nor CUDA has any difference, they're generating the
same exact code, which means no regressions/improvements there.

Based on the research paper:

  https://research.nvidia.com/sites/default/files/publications/laine2013hpg_paper.pdf

Here's the documentation:

  https://docs.google.com/document/d/1LuXW-CV-sVJkQaEGZlMJ86jZ8FmoPfecaMdR-oiWbUY/edit

Design discussion of the patch:

  https://developer.blender.org/T44197

Differential Revision: https://developer.blender.org/D1200

1 files changed, 137 insertions, 0 deletions

diff --git a/intern/cycles/kernel/kernel_direct_lighting.cl b/intern/cycles/kernel/kernel_direct_lighting.cl
new file mode 100644
index 00000000000..8bdc7dc0fd1
--- /dev/null
+++ b/intern/cycles/kernel/kernel_direct_lighting.cl
@@ -0,0 +1,137 @@
+/*
+ * Copyright 2011-2015 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.
+ */
+
+#include "kernel_split.h"
+
+/*
+ * Note on kernel_ocl_path_trace_direct_lighting kernel.
+ * This is the eighth kernel in the ray tracing logic. This is the seventh
+ * of the path iteration kernels. This kernel takes care of direct lighting
+ * logic. However, the "shadow ray cast" part of direct lighting is handled
+ * in the next kernel.
+ *
+ * This kernels determines the rays for which a shadow_blocked() function associated with direct lighting should be executed.
+ * Those rays for which a shadow_blocked() function for direct-lighting must be executed, are marked with flag RAY_SHADOW_RAY_CAST_DL and
+ * enqueued into the queue QUEUE_SHADOW_RAY_CAST_DL_RAYS
+ *
+ * The input and output are as follows,
+ *
+ * rng_coop -----------------------------------------|--- kernel_ocl_path_trace_direct_lighting --|--- BSDFEval_coop
+ * PathState_coop -----------------------------------|                                            |--- ISLamp_coop
+ * shader_data --------------------------------------|                                            |--- LightRay_coop
+ * ray_state ----------------------------------------|                                            |--- ray_state
+ * Queue_data (QUEUE_ACTIVE_AND_REGENERATED_RAYS) ---|                                            |
+ * kg (globals + data) ------------------------------|                                            |
+ * queuesize ----------------------------------------|                                            |
+ *
+ * note on shader_DL : shader_DL is neither input nor output to this kernel; shader_DL is filled and consumed in this kernel itself.
+ * Note on Queues :
+ * This kernel only reads from the QUEUE_ACTIVE_AND_REGENERATED_RAYS queue and processes
+ * only the rays of state RAY_ACTIVE; If a ray needs to execute the corresponding shadow_blocked
+ * part, after direct lighting, the ray is marked with RAY_SHADOW_RAY_CAST_DL flag.
+ *
+ * State of queues when this kernel is called :
+ * state of queues QUEUE_ACTIVE_AND_REGENERATED_RAYS and QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS will be same
+ * before and after this kernel call.
+ * QUEUE_SHADOW_RAY_CAST_DL_RAYS queue will be filled with rays for which a shadow_blocked function must be executed, after this
+ * kernel call. Before this kernel call the QUEUE_SHADOW_RAY_CAST_DL_RAYS will be empty.
+ */
+__kernel void kernel_ocl_path_trace_direct_lighting(
+	ccl_global char *globals,
+	ccl_constant KernelData *data,
+	ccl_global char *shader_data,           /* Required for direct lighting */
+	ccl_global char *shader_DL,             /* Required for direct lighting */
+	ccl_global uint *rng_coop,              /* Required for direct lighting */
+	ccl_global PathState *PathState_coop,   /* Required for direct lighting */
+	ccl_global int *ISLamp_coop,            /* Required for direct lighting */
+	ccl_global Ray *LightRay_coop,          /* Required for direct lighting */
+	ccl_global BsdfEval *BSDFEval_coop,     /* Required for direct lighting */
+	ccl_global char *ray_state,             /* Denotes the state of each ray */
+	ccl_global int *Queue_data,             /* Queue memory */
+	ccl_global int *Queue_index,            /* Tracks the number of elements in each queue */
+	int queuesize                           /* Size (capacity) of each queue */
+	)
+{
+	ccl_local unsigned int local_queue_atomics;
+	if(get_local_id(0) == 0 && get_local_id(1) == 0) {
+		local_queue_atomics = 0;
+	}
+	barrier(CLK_LOCAL_MEM_FENCE);
+
+	char enqueue_flag = 0;
+	int ray_index = get_global_id(1) * get_global_size(0) + get_global_id(0);
+	ray_index = get_ray_index(ray_index, QUEUE_ACTIVE_AND_REGENERATED_RAYS, Queue_data, queuesize, 0);
+
+#ifdef __COMPUTE_DEVICE_GPU__
+	/* If we are executing on a GPU device, we exit all threads that are not required
+	 * If we are executing on a CPU device, then we need to keep all threads active
+	 * since we have barrier() calls later in the kernel. CPU devices,
+	 * expect all threads to execute barrier statement.
+	 */
+	if(ray_index == QUEUE_EMPTY_SLOT)
+		return;
+#endif
+
+#ifndef __COMPUTE_DEVICE_GPU__
+	if(ray_index != QUEUE_EMPTY_SLOT) {
+#endif
+		if(IS_STATE(ray_state, ray_index, RAY_ACTIVE)) {
+			/* Load kernel globals structure and ShaderData structure */
+			KernelGlobals *kg = (KernelGlobals *)globals;
+			ShaderData *sd = (ShaderData *)shader_data;
+			ShaderData *sd_DL  = (ShaderData *)shader_DL;
+
+			ccl_global PathState *state = &PathState_coop[ray_index];
+
+			/* direct lighting */
+#ifdef __EMISSION__
+			if((kernel_data.integrator.use_direct_light && (ccl_fetch(sd, flag) & SD_BSDF_HAS_EVAL))) {
+				/* sample illumination from lights to find path contribution */
+				ccl_global RNG* rng = &rng_coop[ray_index];
+				float light_t = path_state_rng_1D(kg, rng, state, PRNG_LIGHT);
+				float light_u, light_v;
+				path_state_rng_2D(kg, rng, state, PRNG_LIGHT_U, &light_u, &light_v);
+
+#ifdef __OBJECT_MOTION__
+				light_ray.time = sd->time;
+#endif
+				LightSample ls;
+				light_sample(kg, light_t, light_u, light_v, ccl_fetch(sd, time), ccl_fetch(sd, P), state->bounce, &ls);
+
+				Ray light_ray;
+				BsdfEval L_light;
+				bool is_lamp;
+				if(direct_emission(kg, sd, &ls, &light_ray, &L_light, &is_lamp, state->bounce, state->transparent_bounce, sd_DL)) {
+					/* write intermediate data to global memory to access from the next kernel */
+					LightRay_coop[ray_index] = light_ray;
+					BSDFEval_coop[ray_index] = L_light;
+					ISLamp_coop[ray_index] = is_lamp;
+					/// mark ray state for next shadow kernel
+					ADD_RAY_FLAG(ray_state, ray_index, RAY_SHADOW_RAY_CAST_DL);
+					enqueue_flag = 1;
+				}
+			}
+#endif
+		}
+#ifndef __COMPUTE_DEVICE_GPU__
+	}
+#endif
+
+#ifdef __EMISSION__
+	/* Enqueue RAY_SHADOW_RAY_CAST_DL rays */
+	enqueue_ray_index_local(ray_index, QUEUE_SHADOW_RAY_CAST_DL_RAYS, enqueue_flag, queuesize, &local_queue_atomics, Queue_data, Queue_index);
+#endif
+}