ClangFormat: apply to source, most of intern
[blender.git] / intern / cycles / kernel / split / kernel_buffer_update.h
index 18eec63..e777433 100644 (file)
@@ -41,132 +41,133 @@ CCL_NAMESPACE_BEGIN
 ccl_device void kernel_buffer_update(KernelGlobals *kg,
                                      ccl_local_param unsigned int *local_queue_atomics)
 {
-       if(ccl_local_id(0) == 0 && ccl_local_id(1) == 0) {
-               *local_queue_atomics = 0;
-       }
-       ccl_barrier(CCL_LOCAL_MEM_FENCE);
-
-       int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0);
-       if(ray_index == 0) {
-               /* We will empty this queue in this kernel. */
-               kernel_split_params.queue_index[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS] = 0;
-       }
-       char enqueue_flag = 0;
-       ray_index = get_ray_index(kg, ray_index,
-                                 QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS,
-                                 kernel_split_state.queue_data,
-                                 kernel_split_params.queue_size,
-                                 1);
+  if (ccl_local_id(0) == 0 && ccl_local_id(1) == 0) {
+    *local_queue_atomics = 0;
+  }
+  ccl_barrier(CCL_LOCAL_MEM_FENCE);
+
+  int ray_index = ccl_global_id(1) * ccl_global_size(0) + ccl_global_id(0);
+  if (ray_index == 0) {
+    /* We will empty this queue in this kernel. */
+    kernel_split_params.queue_index[QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS] = 0;
+  }
+  char enqueue_flag = 0;
+  ray_index = get_ray_index(kg,
+                            ray_index,
+                            QUEUE_HITBG_BUFF_UPDATE_TOREGEN_RAYS,
+                            kernel_split_state.queue_data,
+                            kernel_split_params.queue_size,
+                            1);
 
 #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;
-       }
+  /* 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) {
+  if (ray_index != QUEUE_EMPTY_SLOT) {
 #endif
 
-       ccl_global char *ray_state = kernel_split_state.ray_state;
-       ccl_global PathState *state = &kernel_split_state.path_state[ray_index];
-       PathRadiance *L = &kernel_split_state.path_radiance[ray_index];
-       ccl_global Ray *ray = &kernel_split_state.ray[ray_index];
-       ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index];
-       bool ray_was_updated = false;
-
-       if(IS_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER)) {
-               ray_was_updated = true;
-               uint sample = state->sample;
-               uint buffer_offset = kernel_split_state.buffer_offset[ray_index];
-               ccl_global float *buffer = kernel_split_params.tile.buffer + buffer_offset;
-
-               /* accumulate result in output buffer */
-               kernel_write_result(kg, buffer, sample, L);
-
-               ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE);
-       }
-
-       if(kernel_data.film.cryptomatte_passes) {
-               /* Make sure no thread is writing to the buffers. */
-               ccl_barrier(CCL_LOCAL_MEM_FENCE);
-               if(ray_was_updated && state->sample - 1 == kernel_data.integrator.aa_samples) {
-                       uint buffer_offset = kernel_split_state.buffer_offset[ray_index];
-                       ccl_global float *buffer = kernel_split_params.tile.buffer + buffer_offset;
-                       ccl_global float *cryptomatte_buffer = buffer + kernel_data.film.pass_cryptomatte;
-                       kernel_sort_id_slots(cryptomatte_buffer, 2 * kernel_data.film.cryptomatte_depth);
-               }
-       }
-
-       if(IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) {
-               /* We have completed current work; So get next work */
-               ccl_global uint *work_pools = kernel_split_params.work_pools;
-               uint total_work_size = kernel_split_params.total_work_size;
-               uint work_index;
-
-               if(!get_next_work(kg, work_pools, total_work_size, ray_index, &work_index)) {
-                       /* If work is invalid, this means no more work is available and the thread may exit */
-                       ASSIGN_RAY_STATE(ray_state, ray_index, RAY_INACTIVE);
-               }
-
-               if(IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) {
-                       ccl_global WorkTile *tile = &kernel_split_params.tile;
-                       uint x, y, sample;
-                       get_work_pixel(tile, work_index, &x, &y, &sample);
-
-                       /* Store buffer offset for writing to passes. */
-                       uint buffer_offset = (tile->offset + x + y*tile->stride) * kernel_data.film.pass_stride;
-                       kernel_split_state.buffer_offset[ray_index] = buffer_offset;
-
-                       /* Initialize random numbers and ray. */
-                       uint rng_hash;
-                       kernel_path_trace_setup(kg, sample, x, y, &rng_hash, ray);
-
-                       if(ray->t != 0.0f) {
-                               /* Initialize throughput, path radiance, Ray, PathState;
-                                * These rays proceed with path-iteration.
-                                */
-                               *throughput = make_float3(1.0f, 1.0f, 1.0f);
-                               path_radiance_init(L, kernel_data.film.use_light_pass);
-                               path_state_init(kg,
-                                               AS_SHADER_DATA(&kernel_split_state.sd_DL_shadow[ray_index]),
-                                               state,
-                                               rng_hash,
-                                               sample,
-                                               ray);
+    ccl_global char *ray_state = kernel_split_state.ray_state;
+    ccl_global PathState *state = &kernel_split_state.path_state[ray_index];
+    PathRadiance *L = &kernel_split_state.path_radiance[ray_index];
+    ccl_global Ray *ray = &kernel_split_state.ray[ray_index];
+    ccl_global float3 *throughput = &kernel_split_state.throughput[ray_index];
+    bool ray_was_updated = false;
+
+    if (IS_STATE(ray_state, ray_index, RAY_UPDATE_BUFFER)) {
+      ray_was_updated = true;
+      uint sample = state->sample;
+      uint buffer_offset = kernel_split_state.buffer_offset[ray_index];
+      ccl_global float *buffer = kernel_split_params.tile.buffer + buffer_offset;
+
+      /* accumulate result in output buffer */
+      kernel_write_result(kg, buffer, sample, L);
+
+      ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE);
+    }
+
+    if (kernel_data.film.cryptomatte_passes) {
+      /* Make sure no thread is writing to the buffers. */
+      ccl_barrier(CCL_LOCAL_MEM_FENCE);
+      if (ray_was_updated && state->sample - 1 == kernel_data.integrator.aa_samples) {
+        uint buffer_offset = kernel_split_state.buffer_offset[ray_index];
+        ccl_global float *buffer = kernel_split_params.tile.buffer + buffer_offset;
+        ccl_global float *cryptomatte_buffer = buffer + kernel_data.film.pass_cryptomatte;
+        kernel_sort_id_slots(cryptomatte_buffer, 2 * kernel_data.film.cryptomatte_depth);
+      }
+    }
+
+    if (IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) {
+      /* We have completed current work; So get next work */
+      ccl_global uint *work_pools = kernel_split_params.work_pools;
+      uint total_work_size = kernel_split_params.total_work_size;
+      uint work_index;
+
+      if (!get_next_work(kg, work_pools, total_work_size, ray_index, &work_index)) {
+        /* If work is invalid, this means no more work is available and the thread may exit */
+        ASSIGN_RAY_STATE(ray_state, ray_index, RAY_INACTIVE);
+      }
+
+      if (IS_STATE(ray_state, ray_index, RAY_TO_REGENERATE)) {
+        ccl_global WorkTile *tile = &kernel_split_params.tile;
+        uint x, y, sample;
+        get_work_pixel(tile, work_index, &x, &y, &sample);
+
+        /* Store buffer offset for writing to passes. */
+        uint buffer_offset = (tile->offset + x + y * tile->stride) * kernel_data.film.pass_stride;
+        kernel_split_state.buffer_offset[ray_index] = buffer_offset;
+
+        /* Initialize random numbers and ray. */
+        uint rng_hash;
+        kernel_path_trace_setup(kg, sample, x, y, &rng_hash, ray);
+
+        if (ray->t != 0.0f) {
+          /* Initialize throughput, path radiance, Ray, PathState;
+         * These rays proceed with path-iteration.
+         */
+          *throughput = make_float3(1.0f, 1.0f, 1.0f);
+          path_radiance_init(L, kernel_data.film.use_light_pass);
+          path_state_init(kg,
+                          AS_SHADER_DATA(&kernel_split_state.sd_DL_shadow[ray_index]),
+                          state,
+                          rng_hash,
+                          sample,
+                          ray);
 #ifdef __SUBSURFACE__
-                               kernel_path_subsurface_init_indirect(&kernel_split_state.ss_rays[ray_index]);
+          kernel_path_subsurface_init_indirect(&kernel_split_state.ss_rays[ray_index]);
 #endif
-                               ASSIGN_RAY_STATE(ray_state, ray_index, RAY_REGENERATED);
-                               enqueue_flag = 1;
-                       }
-                       else {
-                               ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE);
-                       }
-               }
-       }
+          ASSIGN_RAY_STATE(ray_state, ray_index, RAY_REGENERATED);
+          enqueue_flag = 1;
+        }
+        else {
+          ASSIGN_RAY_STATE(ray_state, ray_index, RAY_TO_REGENERATE);
+        }
+      }
+    }
 
 #ifndef __COMPUTE_DEVICE_GPU__
-       }
+  }
 #endif
 
-       /* Enqueue RAY_REGENERATED rays into QUEUE_ACTIVE_AND_REGENERATED_RAYS;
-        * These rays will be made active during next SceneIntersectkernel.
-        */
-       enqueue_ray_index_local(ray_index,
-                               QUEUE_ACTIVE_AND_REGENERATED_RAYS,
-                               enqueue_flag,
-                               kernel_split_params.queue_size,
-                               local_queue_atomics,
-                               kernel_split_state.queue_data,
-                               kernel_split_params.queue_index);
+  /* Enqueue RAY_REGENERATED rays into QUEUE_ACTIVE_AND_REGENERATED_RAYS;
+   * These rays will be made active during next SceneIntersectkernel.
+   */
+  enqueue_ray_index_local(ray_index,
+                          QUEUE_ACTIVE_AND_REGENERATED_RAYS,
+                          enqueue_flag,
+                          kernel_split_params.queue_size,
+                          local_queue_atomics,
+                          kernel_split_state.queue_data,
+                          kernel_split_params.queue_index);
 }
 
 CCL_NAMESPACE_END