Merge remote-tracking branch 'b/main'

extern/hipew/include/hiprtew.h

@@ -79,6 +79,7 @@ typedef hiprtError(thiprtSetFuncTable)(hiprtContext context,
  hiprtFuncDataSet set);
 typedef hiprtError(thiprtDestroyFuncTable)(hiprtContext context,
  hiprtFuncTable funcTable);
+typedef void(thiprtSetLogLevel)( hiprtLogLevel level );
 
 /* Function declarations. */
 extern thiprtCreateContext *hiprtCreateContext;
@@ -94,6 +95,7 @@ extern thiprtGetSceneBuildTemporaryBufferSize *hiprtGetSceneBuildTemporaryBuffer
 extern thiprtCreateFuncTable *hiprtCreateFuncTable;
 extern thiprtSetFuncTable *hiprtSetFuncTable;
 extern thiprtDestroyFuncTable *hiprtDestroyFuncTable;
+extern thiprtSetLogLevel *hiprtSetLogLevel;
 
 /* HIPEW API. */
 

extern/hipew/src/hiprtew.cc

@@ -41,6 +41,7 @@ thiprtGetSceneBuildTemporaryBufferSize *hiprtGetSceneBuildTemporaryBufferSize;
 thiprtCreateFuncTable *hiprtCreateFuncTable;
 thiprtSetFuncTable *hiprtSetFuncTable;
 thiprtDestroyFuncTable *hiprtDestroyFuncTable;
+thiprtSetLogLevel *hiprtSetLogLevel;
 
 static void hipewHipRtExit(void)
 {
@@ -89,6 +90,7 @@ bool hiprtewInit()
  HIPRT_LIBRARY_FIND(hiprtCreateFuncTable)
  HIPRT_LIBRARY_FIND(hiprtSetFuncTable)
  HIPRT_LIBRARY_FIND(hiprtDestroyFuncTable)
+ HIPRT_LIBRARY_FIND(hiprtSetLogLevel)
 
  result = true;
 #endif

extern/xdnd/xdnd.c

@@ -464,8 +464,9 @@ static char *concat_string_list (char **t, int *bytes)
  break;
  if (!(t[n][0]))
  break;
- strcpy (s + l, t[n]);
- l += strlen (t[n]) + 1;
+ int t_size = strlen (t[n]) + 1;
+ memcpy (s + l, t[n], t_size);
+ l += t_size;
  }
  *bytes = l;
  s[l] = '\0';

intern/cycles/device/hiprt/device_impl.cpp

@@ -96,6 +96,8 @@ HIPRTDevice::HIPRTDevice(const DeviceInfo &info, Stats &stats, Profiler &profile
  set_error(string_printf("Failed to create HIPRT Function Table"));
  return;
  }
+
+ hiprtSetLogLevel(hiprtLogLevelNone);
 }
 
 HIPRTDevice::~HIPRTDevice()
@@ -261,8 +263,7 @@ string HIPRTDevice::compile_kernel(const uint kernel_features, const char *name,
  linker_options.append(" --offload-arch=").append(arch);
  linker_options.append(" -fgpu-rdc --hip-link --cuda-device-only ");
  string hiprt_ver(HIPRT_VERSION_STR);
- string hiprt_bc;
- hiprt_bc = hiprt_path + "\\hiprt" + hiprt_ver + "_amd_lib_win.bc";
+ string hiprt_bc = hiprt_path + "\\dist\\bin\\Release\\hiprt" + hiprt_ver + "_amd_lib_win.bc";
 
  string linker_command = string_printf("clang++ %s \"%s\" %s -o \"%s\"",
  linker_options.c_str(),

intern/cycles/device/hiprt/queue.cpp

@@ -61,6 +61,8 @@ bool HIPRTDeviceQueue::enqueue(DeviceKernel kernel,
  0),
  "enqueue");
 
+ debug_enqueue_end();
+
  return !(hiprt_device_->have_error());
 }
 

intern/cycles/kernel/device/hiprt/bvh.h

@@ -46,11 +46,11 @@ ccl_device_intersect bool scene_intersect(KernelGlobals kg,
  GET_TRAVERSAL_STACK()
 
  if (visibility & PATH_RAY_SHADOW_OPAQUE) {
- GET_TRAVERSAL_ANY_HIT(table_closest_intersect, 0)
+ GET_TRAVERSAL_ANY_HIT(table_closest_intersect, 0, ray->time)
  hit = traversal.getNextHit();
  }
  else {
- GET_TRAVERSAL_CLOSEST_HIT(table_closest_intersect, 0)
+ GET_TRAVERSAL_CLOSEST_HIT(table_closest_intersect, 0, ray->time)
  hit = traversal.getNextHit();
  }
  if (hit.hasHit()) {
@@ -157,13 +157,13 @@ ccl_device_intersect bool scene_intersect_shadow_all(KernelGlobals kg,
  payload.in_state = state;
  payload.max_hits = max_hits;
  payload.visibility = visibility;
- payload.prim_type = PRIMITIVE_TRIANGLE;
+ payload.prim_type = PRIMITIVE_NONE;
  payload.ray_time = ray->time;
  payload.num_hits = 0;
  payload.r_num_recorded_hits = num_recorded_hits;
  payload.r_throughput = throughput;
  GET_TRAVERSAL_STACK()
- GET_TRAVERSAL_ANY_HIT(table_shadow_intersect, 1)
+ GET_TRAVERSAL_ANY_HIT(table_shadow_intersect, 1, ray->time)
  hiprtHit hit = traversal.getNextHit();
  num_recorded_hits = payload.r_num_recorded_hits;
  throughput = payload.r_throughput;
@@ -201,7 +201,7 @@ ccl_device_intersect bool scene_intersect_volume(KernelGlobals kg,
 
  GET_TRAVERSAL_STACK()
 
- GET_TRAVERSAL_CLOSEST_HIT(table_volume_intersect, 3)
+ GET_TRAVERSAL_CLOSEST_HIT(table_volume_intersect, 3, ray->time)
  hiprtHit hit = traversal.getNextHit();
  // return hit.hasHit();
  if (hit.hasHit()) {

intern/cycles/kernel/device/hiprt/common.h

@@ -53,45 +53,42 @@ struct LocalPayload {
 # endif
 
 # ifdef HIPRT_SHARED_STACK
-# define GET_TRAVERSAL_ANY_HIT(FUNCTION_TABLE, RAY_TYPE) \
+# define GET_TRAVERSAL_ANY_HIT(FUNCTION_TABLE, RAY_TYPE, RAY_TIME) \
  hiprtSceneTraversalAnyHitCustomStack<Stack> traversal(kernel_data.device_bvh, \
  ray_hip, \
  stack, \
  visibility, \
  hiprtTraversalHintDefault, \
  &payload, \
  kernel_params.FUNCTION_TABLE, \
- RAY_TYPE); \
- hiprtSceneTraversalAnyHitCustomStack<Stack> traversal_simple( \
- kernel_data.device_bvh, ray_hip, stack, visibility);
-# define GET_TRAVERSAL_CLOSEST_HIT(FUNCTION_TABLE, RAY_TYPE) \
+ RAY_TYPE, \
+  RAY_TIME);
+
+# define GET_TRAVERSAL_CLOSEST_HIT(FUNCTION_TABLE, RAY_TYPE, RAY_TIME) \
  hiprtSceneTraversalClosestCustomStack<Stack> traversal(kernel_data.device_bvh, \
  ray_hip, \
  stack, \
  visibility, \
  hiprtTraversalHintDefault, \
  &payload, \
  kernel_params.FUNCTION_TABLE, \
- RAY_TYPE); \
- hiprtSceneTraversalClosestCustomStack<Stack> traversal_simple( \
- kernel_data.device_bvh, ray_hip, stack, visibility);
+ RAY_TYPE, \
+ RAY_TIME);
 # else
 # define GET_TRAVERSAL_ANY_HIT(FUNCTION_TABLE) \
  hiprtSceneTraversalAnyHit traversal(kernel_data.device_bvh, \
  ray_hip, \
  visibility, \
  FUNCTION_TABLE, \
  hiprtTraversalHintDefault, \
- &payload); \
- hiprtSceneTraversalAnyHit traversal_simple(kernel_data.device_bvh, ray_hip, visibility);
+ &payload);
 # define GET_TRAVERSAL_CLOSEST_HIT(FUNCTION_TABLE) \
  hiprtSceneTraversalClosest traversal(kernel_data.device_bvh, \
  ray_hip, \
  visibility, \
  FUNCTION_TABLE, \
  hiprtTraversalHintDefault, \
- &payload); \
- hiprtSceneTraversalClosest traversal_simple(kernel_data.device_bvh, ray_hip, visibility);
+ &payload);
 # endif
 
 ccl_device_inline void set_intersect_point(KernelGlobals kg,

intern/cycles/kernel/integrator/mnee.h

@@ -775,7 +775,9 @@ ccl_device_forceinline bool mnee_path_contribution(KernelGlobals kg,
  surface_shader_bsdf_eval(kg, state, sd, wo, throughput, ls->shader);
 
  /* Update light sample with new position / direction and keep pdf in vertex area measure. */
- light_sample_update(kg, ls, vertices[vertex_count - 1].p);
+ const uint32_t path_flag = INTEGRATOR_STATE(state, path, flag);
+ light_sample_update(
+ kg, ls, vertices[vertex_count - 1].p, vertices[vertex_count - 1].n, path_flag);
 
  /* Save state path bounce info in case a light path node is used in the refractive interface or
  * light shader graph. */

intern/cycles/kernel/integrator/shade_dedicated_light.h

@@ -20,6 +20,8 @@ ccl_device_inline bool shadow_linking_light_sample_from_intersection(
  KernelGlobals kg,
  ccl_private const Intersection &ccl_restrict isect,
  ccl_private const Ray &ccl_restrict ray,
+ const float3 N,
+ const uint32_t path_flag,
  ccl_private LightSample *ccl_restrict ls)
 {
  const int lamp = isect.prim;
@@ -31,7 +33,7 @@ ccl_device_inline bool shadow_linking_light_sample_from_intersection(
  return distant_light_sample_from_intersection(kg, ray.D, lamp, ls);
  }
 
- return light_sample_from_intersection(kg, &isect, ray.P, ray.D, ls);
+ return light_sample_from_intersection(kg, &isect, ray.P, ray.D, N, path_flag, ls);
 }
 
 ccl_device_inline float shadow_linking_light_sample_mis_weight(KernelGlobals kg,
@@ -88,8 +90,11 @@ ccl_device bool shadow_linking_shade_light(KernelGlobals kg,
  ccl_private float &mis_weight,
  ccl_private int &ccl_restrict light_group)
 {
+ const uint32_t path_flag = INTEGRATOR_STATE(state, path, flag);
+ const float3 N = INTEGRATOR_STATE(state, path, mis_origin_n);
  LightSample ls ccl_optional_struct_init;
- const bool use_light_sample = shadow_linking_light_sample_from_intersection(kg, isect, ray, &ls);
+ const bool use_light_sample = shadow_linking_light_sample_from_intersection(
+ kg, isect, ray, N, path_flag, &ls);
  if (!use_light_sample) {
  /* No light to be sampled, so no direct light contribution either. */
  return false;
@@ -100,7 +105,6 @@ ccl_device bool shadow_linking_shade_light(KernelGlobals kg,
  return false;
  }
 
- const uint32_t path_flag = INTEGRATOR_STATE(state, path, flag);
  if (!is_light_shader_visible_to_path(ls.shader, path_flag)) {
  return false;
  }

intern/cycles/kernel/integrator/shade_light.h

@@ -24,20 +24,22 @@ ccl_device_inline void integrate_light(KernelGlobals kg,
  float3 ray_P = INTEGRATOR_STATE(state, ray, P);
  const float3 ray_D = INTEGRATOR_STATE(state, ray, D);
  const float ray_time = INTEGRATOR_STATE(state, ray, time);
+ const uint32_t path_flag = INTEGRATOR_STATE(state, path, flag);
+ const float3 N = INTEGRATOR_STATE(state, path, mis_origin_n);
 
  /* Advance ray to new start distance. */
  INTEGRATOR_STATE_WRITE(state, ray, tmin) = intersection_t_offset(isect.t);
 
  LightSample ls ccl_optional_struct_init;
- const bool use_light_sample = light_sample_from_intersection(kg, &isect, ray_P, ray_D, &ls);
+ const bool use_light_sample = light_sample_from_intersection(
+ kg, &isect, ray_P, ray_D, N, path_flag, &ls);
 
  if (!use_light_sample) {
  return;
  }
 
  /* Use visibility flag to skip lights. */
 #ifdef __PASSES__
- const uint32_t path_flag = INTEGRATOR_STATE(state, path, flag);
  if (!is_light_shader_visible_to_path(ls.shader, path_flag)) {
  return;
  }

intern/cycles/kernel/light/distribution.h

@@ -46,7 +46,9 @@ ccl_device_noinline bool light_distribution_sample(KernelGlobals kg,
  const float3 rand,
  const float time,
  const float3 P,
+ const float3 N,
  const int object_receiver,
+ const int shader_flags,
  const int bounce,
  const uint32_t path_flag,
  ccl_private LightSample *ls)
@@ -56,8 +58,19 @@ ccl_device_noinline bool light_distribution_sample(KernelGlobals kg,
  const int index = light_distribution_sample(kg, rand.z);
  const float pdf_selection = kernel_data.integrator.distribution_pdf_lights;
  const float2 rand_uv = float3_to_float2(rand);
- return light_sample<in_volume_segment>(
- kg, rand_uv, time, P, object_receiver, bounce, path_flag, index, 0, pdf_selection, ls);
+ return light_sample<in_volume_segment>(kg,
+ rand_uv,
+ time,
+ P,
+ N,
+ object_receiver,
+ shader_flags,
+ bounce,
+ path_flag,
+ index,
+ 0,
+ pdf_selection,
+ ls);
 }
 
 ccl_device_inline float light_distribution_pdf_lamp(KernelGlobals kg)

intern/cycles/kernel/light/light.h

@@ -96,6 +96,8 @@ ccl_device_inline bool light_sample(KernelGlobals kg,
  const int lamp,
  const float2 rand,
  const float3 P,
+ const float3 N,
+ const int shader_flags,
  const uint32_t path_flag,
  ccl_private LightSample *ls)
 {
@@ -150,7 +152,7 @@ ccl_device_inline bool light_sample(KernelGlobals kg,
  }
  }
  else if (type == LIGHT_POINT) {
- if (!point_light_sample<in_volume_segment>(klight, rand, P, ls)) {
+ if (!point_light_sample(klight, rand, P, N, shader_flags, ls)) {
  return false;
  }
  }
@@ -171,7 +173,9 @@ ccl_device_noinline bool light_sample(KernelGlobals kg,
  const float2 rand,
  const float time,
  const float3 P,
+ const float3 N,
  const int object_receiver,
+ const int shader_flags,
  const int bounce,
  const uint32_t path_flag,
  const int emitter_index,
@@ -233,7 +237,7 @@ ccl_device_noinline bool light_sample(KernelGlobals kg,
  return false;
  }
 
- if (!light_sample<in_volume_segment>(kg, light, rand, P, path_flag, ls)) {
+ if (!light_sample<in_volume_segment>(kg, light, rand, P, N, shader_flags, path_flag, ls)) {
  return false;
  }
  }
@@ -446,6 +450,8 @@ ccl_device bool light_sample_from_intersection(KernelGlobals kg,
  ccl_private const Intersection *ccl_restrict isect,
  const float3 ray_P,
  const float3 ray_D,
+ const float3 N,
+ const uint32_t path_flag,
  ccl_private LightSample *ccl_restrict ls)
 {
  const int lamp = isect->prim;
@@ -468,7 +474,7 @@ ccl_device bool light_sample_from_intersection(KernelGlobals kg,
  }
  }
  else if (type == LIGHT_POINT) {
- if (!point_light_sample_from_intersection(klight, isect, ray_P, ray_D, ls)) {
+ if (!point_light_sample_from_intersection(klight, isect, ray_P, ray_D, N, path_flag, ls)) {
  return false;
  }
  }