Cycles: Introduce postfix for kernel body definition

Increases flexibility of code-generation for kernel entry points.

Currently no functional changes, preparing for integration with oneAPI.

4 files changed, 42 insertions, 1 deletions

diff --git a/intern/cycles/kernel/device/cuda/config.h b/intern/cycles/kernel/device/cuda/config.h
index 1f66bb0175a..88149e92ec9 100644
--- a/intern/cycles/kernel/device/cuda/config.h
+++ b/intern/cycles/kernel/device/cuda/config.h
@@ -88,6 +88,7 @@
   extern "C" __global__ void __launch_bounds__(block_num_threads)
 
 #define ccl_gpu_kernel_signature(name, ...) kernel_gpu_##name(__VA_ARGS__)
+#define ccl_gpu_kernel_postfix
 
 #define ccl_gpu_kernel_call(x) x
 
diff --git a/intern/cycles/kernel/device/gpu/kernel.h b/intern/cycles/kernel/device/gpu/kernel.h
index 26ab99766ad..82b51843864 100644
--- a/intern/cycles/kernel/device/gpu/kernel.h
+++ b/intern/cycles/kernel/device/gpu/kernel.h
@@ -58,6 +58,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     INTEGRATOR_STATE_WRITE(state, shadow_path, queued_kernel) = 0;
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_init_from_camera,
@@ -89,6 +90,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
   ccl_gpu_kernel_call(
       integrator_init_from_camera(nullptr, state, tile, render_buffer, x, y, sample));
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_init_from_bake,
@@ -120,6 +122,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
   ccl_gpu_kernel_call(
       integrator_init_from_bake(nullptr, state, tile, render_buffer, x, y, sample));
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_intersect_closest,
@@ -134,6 +137,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(integrator_intersect_closest(NULL, state, render_buffer));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_intersect_shadow,
@@ -147,6 +151,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(integrator_intersect_shadow(NULL, state));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_intersect_subsurface,
@@ -160,6 +165,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(integrator_intersect_subsurface(NULL, state));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_intersect_volume_stack,
@@ -173,6 +179,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(integrator_intersect_volume_stack(NULL, state));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_shade_background,
@@ -187,6 +194,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(integrator_shade_background(NULL, state, render_buffer));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_shade_light,
@@ -201,6 +209,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(integrator_shade_light(NULL, state, render_buffer));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_shade_shadow,
@@ -215,6 +224,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(integrator_shade_shadow(NULL, state, render_buffer));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_shade_surface,
@@ -229,6 +239,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(integrator_shade_surface(NULL, state, render_buffer));
   }
 }
+ccl_gpu_kernel_postfix
 
 #ifdef __KERNEL_METAL__
 constant int __dummy_constant [[function_constant(0)]];
@@ -256,6 +267,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
 #endif
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(integrator_shade_volume,
@@ -270,6 +282,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(integrator_shade_volume(NULL, state, render_buffer));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_queued_paths_array,
@@ -288,6 +301,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
                                   num_indices,
                                   ccl_gpu_kernel_lambda_pass);
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_queued_shadow_paths_array,
@@ -306,6 +320,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
                                   num_indices,
                                   ccl_gpu_kernel_lambda_pass);
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_active_paths_array,
@@ -321,6 +336,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
                                   num_indices,
                                   ccl_gpu_kernel_lambda_pass);
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_terminated_paths_array,
@@ -337,6 +353,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
                                   num_indices,
                                   ccl_gpu_kernel_lambda_pass);
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_terminated_shadow_paths_array,
@@ -353,6 +370,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
                                   num_indices,
                                   ccl_gpu_kernel_lambda_pass);
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_sorted_paths_array,
@@ -380,6 +398,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)
                                   key_prefix_sum,
                                   ccl_gpu_kernel_lambda_pass);
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_compact_paths_array,
@@ -399,6 +418,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
                                   num_indices,
                                   ccl_gpu_kernel_lambda_pass);
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_compact_states,
@@ -416,6 +436,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_call(integrator_state_move(NULL, to_state, from_state));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_compact_shadow_paths_array,
@@ -435,6 +456,7 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_ACTIVE_INDEX_DEFAULT_BLOCK_SIZE)
                                   num_indices,
                                   ccl_gpu_kernel_lambda_pass);
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_signature(integrator_compact_shadow_states,
@@ -452,12 +474,14 @@ ccl_gpu_kernel_threads(GPU_PARALLEL_SORTED_INDEX_DEFAULT_BLOCK_SIZE)
     ccl_gpu_kernel_call(integrator_shadow_state_move(NULL, to_state, from_state));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel_threads(GPU_PARALLEL_PREFIX_SUM_DEFAULT_BLOCK_SIZE) ccl_gpu_kernel_signature(
     prefix_sum, ccl_global int *counter, ccl_global int *prefix_sum, int num_values)
 {
   gpu_parallel_prefix_sum(ccl_gpu_global_id_x(), counter, prefix_sum, num_values);
 }
+ccl_gpu_kernel_postfix
 
 /* --------------------------------------------------------------------
  * Adaptive sampling.
@@ -494,6 +518,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     atomic_fetch_and_add_uint32(num_active_pixels, popcount(num_active_pixels_mask));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(adaptive_sampling_filter_x,
@@ -512,6 +537,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
         kernel_adaptive_sampling_filter_x(NULL, render_buffer, sy + y, sx, sw, offset, stride));
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(adaptive_sampling_filter_y,
@@ -530,6 +556,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
         kernel_adaptive_sampling_filter_y(NULL, render_buffer, sx + x, sy, sh, offset, stride));
   }
 }
+ccl_gpu_kernel_postfix
 
 /* --------------------------------------------------------------------
  * Cryptomatte.
@@ -546,6 +573,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(kernel_cryptomatte_post(nullptr, render_buffer, pixel_index));
   }
 }
+ccl_gpu_kernel_postfix
 
 /* --------------------------------------------------------------------
  * Film.
@@ -627,6 +655,7 @@ ccl_device_inline void kernel_gpu_film_convert_half_write(ccl_global uchar4 *rgb
 \
     FILM_GET_PASS_PIXEL_F32(variant, input_channel_count); \
   } \
+  ccl_gpu_kernel_postfix \
 \
   ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS) \
       ccl_gpu_kernel_signature(film_convert_##variant##_half_rgba, \
@@ -666,7 +695,8 @@ ccl_device_inline void kernel_gpu_film_convert_half_write(ccl_global uchar4 *rgb
     const half4 half_pixel = float4_to_half4_display( \
         make_float4(pixel[0], pixel[1], pixel[2], pixel[3])); \
     kernel_gpu_film_convert_half_write(rgba, rgba_offset, rgba_stride, x, y, half_pixel); \
-  }
+  } \
+  ccl_gpu_kernel_postfix
 
 /* 1 channel inputs */
 KERNEL_FILM_CONVERT_VARIANT(depth, 1)
@@ -706,6 +736,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(kernel_displace_evaluate(NULL, input, output, offset + i));
   }
 }
+ccl_gpu_kernel_postfix
 
 /* Background */
 
@@ -721,6 +752,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_call(kernel_background_evaluate(NULL, input, output, offset + i));
   }
 }
+ccl_gpu_kernel_postfix
 
 /* Curve Shadow Transparency */
 
@@ -737,6 +769,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
         kernel_curve_shadow_transparency_evaluate(NULL, input, output, offset + i));
   }
 }
+ccl_gpu_kernel_postfix
 
 /* --------------------------------------------------------------------
  * Denoising.
@@ -770,6 +803,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
   color_out[1] = clamp(color_out[1], 0.0f, 10000.0f);
   color_out[2] = clamp(color_out[2], 0.0f, 10000.0f);
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(filter_guiding_preprocess,
@@ -849,6 +883,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     flow_out[1] = -motion_in[1] * pixel_scale;
   }
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(filter_guiding_set_fake_albedo,
@@ -877,6 +912,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
   albedo_out[1] = 0.5f;
   albedo_out[2] = 0.5f;
 }
+ccl_gpu_kernel_postfix
 
 ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     ccl_gpu_kernel_signature(filter_color_postprocess,
@@ -936,6 +972,7 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     denoised_pixel[3] = 0;
   }
 }
+ccl_gpu_kernel_postfix
 
 /* --------------------------------------------------------------------
  * Shadow catcher.
@@ -961,3 +998,4 @@ ccl_gpu_kernel(GPU_KERNEL_BLOCK_NUM_THREADS, GPU_KERNEL_MAX_REGISTERS)
     atomic_fetch_and_add_uint32(num_possible_splits, popcount(can_split_mask));
   }
 }
+ccl_gpu_kernel_postfix
diff --git a/intern/cycles/kernel/device/hip/config.h b/intern/cycles/kernel/device/hip/config.h
index a5a5924d5e0..c7e7306d628 100644
--- a/intern/cycles/kernel/device/hip/config.h
+++ b/intern/cycles/kernel/device/hip/config.h
@@ -31,6 +31,7 @@
   extern "C" __global__ void __launch_bounds__(block_num_threads)
 
 #define ccl_gpu_kernel_signature(name, ...) kernel_gpu_##name(__VA_ARGS__)
+#define ccl_gpu_kernel_postfix
 
 #define ccl_gpu_kernel_call(x) x
 
diff --git a/intern/cycles/kernel/device/metal/compat.h b/intern/cycles/kernel/device/metal/compat.h
index c12987c0a91..4e309f16c08 100644
--- a/intern/cycles/kernel/device/metal/compat.h
+++ b/intern/cycles/kernel/device/metal/compat.h
@@ -132,6 +132,7 @@ void kernel_gpu_##name::run(thread MetalKernelContext& context, \
                   uint simd_group_index, \
                   uint num_simd_groups) ccl_global const
 
+#define ccl_gpu_kernel_postfix
 #define ccl_gpu_kernel_call(x) context.x
 
 /* define a function object where "func" is the lambda body, and additional parameters are used to specify captured state  */