diff options
Diffstat (limited to 'intern/cycles/kernel')
| -rw-r--r-- | intern/cycles/kernel/CMakeLists.txt | 6 | ||||
| -rw-r--r-- | intern/cycles/kernel/kernel.h | 13 | ||||
| -rw-r--r-- | intern/cycles/kernel/kernel_accumulate.h | 8 | ||||
| -rw-r--r-- | intern/cycles/kernel/kernel_bvh.h | 24 | ||||
| -rw-r--r-- | intern/cycles/kernel/kernel_emission.h | 13 | ||||
| -rw-r--r-- | intern/cycles/kernel/kernel_light.h | 42 | ||||
| -rw-r--r-- | intern/cycles/kernel/kernel_path.h | 30 | ||||
| -rw-r--r-- | intern/cycles/kernel/kernel_sse2.cpp | 60 | ||||
| -rw-r--r-- | intern/cycles/kernel/kernel_sse3.cpp (renamed from intern/cycles/kernel/kernel_optimized.cpp) | 6 | ||||
| -rw-r--r-- | intern/cycles/kernel/kernel_types.h | 10 | ||||
| -rw-r--r-- | intern/cycles/kernel/osl/SConscript | 1 | ||||
| -rw-r--r-- | intern/cycles/kernel/shaders/node_musgrave_texture.osl | 24 | ||||
| -rw-r--r-- | intern/cycles/kernel/shaders/node_texture.h | 14 |
13 files changed, 148 insertions, 103 deletions
diff --git a/intern/cycles/kernel/CMakeLists.txt b/intern/cycles/kernel/CMakeLists.txt index 6d5b9a063a0..e83756b7c8a 100644 --- a/intern/cycles/kernel/CMakeLists.txt +++ b/intern/cycles/kernel/CMakeLists.txt @@ -12,7 +12,8 @@ set(INC_SYS set(SRC kernel.cpp - kernel_optimized.cpp + kernel_sse2.cpp + kernel_sse3.cpp kernel.cl kernel.cu ) @@ -149,7 +150,8 @@ include_directories(SYSTEM ${INC_SYS}) add_library(cycles_kernel ${SRC} ${SRC_HEADERS} ${SRC_CLOSURE_HEADERS} ${SRC_SVM_HEADERS}) if(WITH_CYCLES_OPTIMIZED_KERNEL) - set_source_files_properties(kernel_optimized.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_OPTIMIZED_KERNEL_FLAGS}") + set_source_files_properties(kernel_sse2.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE2_KERNEL_FLAGS}") + set_source_files_properties(kernel_sse3.cpp PROPERTIES COMPILE_FLAGS "${CYCLES_SSE3_KERNEL_FLAGS}") endif() if(WITH_CYCLES_CUDA) diff --git a/intern/cycles/kernel/kernel.h b/intern/cycles/kernel/kernel.h index 26c0bcd6d1a..20ea5a61906 100644 --- a/intern/cycles/kernel/kernel.h +++ b/intern/cycles/kernel/kernel.h @@ -44,11 +44,18 @@ void kernel_cpu_shader(KernelGlobals *kg, uint4 *input, float4 *output, int type, int i); #ifdef WITH_OPTIMIZED_KERNEL -void kernel_cpu_optimized_path_trace(KernelGlobals *kg, float *buffer, unsigned int *rng_state, +void kernel_cpu_sse2_path_trace(KernelGlobals *kg, float *buffer, unsigned int *rng_state, int sample, int x, int y, int offset, int stride); -void kernel_cpu_optimized_tonemap(KernelGlobals *kg, uchar4 *rgba, float *buffer, +void kernel_cpu_sse2_tonemap(KernelGlobals *kg, uchar4 *rgba, float *buffer, int sample, int resolution, int x, int y, int offset, int stride); -void kernel_cpu_optimized_shader(KernelGlobals *kg, uint4 *input, float4 *output, +void kernel_cpu_sse2_shader(KernelGlobals *kg, uint4 *input, float4 *output, + int type, int i); + +void kernel_cpu_sse3_path_trace(KernelGlobals *kg, float *buffer, unsigned int *rng_state, + int sample, int x, int y, int offset, int stride); +void kernel_cpu_sse3_tonemap(KernelGlobals *kg, uchar4 *rgba, float *buffer, + int sample, int resolution, int x, int y, int offset, int stride); +void kernel_cpu_sse3_shader(KernelGlobals *kg, uint4 *input, float4 *output, int type, int i); #endif diff --git a/intern/cycles/kernel/kernel_accumulate.h b/intern/cycles/kernel/kernel_accumulate.h index 86177301357..1cf230634fc 100644 --- a/intern/cycles/kernel/kernel_accumulate.h +++ b/intern/cycles/kernel/kernel_accumulate.h @@ -220,7 +220,7 @@ __device_inline void path_radiance_accum_ao(PathRadiance *L, float3 throughput, #endif } -__device_inline void path_radiance_accum_light(PathRadiance *L, float3 throughput, BsdfEval *bsdf_eval, float3 shadow, int bounce, bool is_lamp) +__device_inline void path_radiance_accum_light(PathRadiance *L, float3 throughput, BsdfEval *bsdf_eval, float3 shadow, float shadow_fac, int bounce, bool is_lamp) { #ifdef __PASSES__ if(L->use_light_pass) { @@ -233,9 +233,9 @@ __device_inline void path_radiance_accum_light(PathRadiance *L, float3 throughpu if(is_lamp) { float3 sum = throughput*(bsdf_eval->diffuse + bsdf_eval->glossy + bsdf_eval->transmission); - L->shadow.x += shadow.x; - L->shadow.y += shadow.y; - L->shadow.z += shadow.z; + L->shadow.x += shadow.x*shadow_fac; + L->shadow.y += shadow.y*shadow_fac; + L->shadow.z += shadow.z*shadow_fac; L->shadow.w += average(sum); } } diff --git a/intern/cycles/kernel/kernel_bvh.h b/intern/cycles/kernel/kernel_bvh.h index 1a85b5bbefd..2b9ebf35d0c 100644 --- a/intern/cycles/kernel/kernel_bvh.h +++ b/intern/cycles/kernel/kernel_bvh.h @@ -126,21 +126,21 @@ __device_inline void bvh_node_intersect(KernelGlobals *kg, /* intersect ray against child nodes */ float3 ood = P * idir; - float c0lox = n0xy.x * idir.x - ood.x; - float c0hix = n0xy.y * idir.x - ood.x; - float c0loy = n0xy.z * idir.y - ood.y; - float c0hiy = n0xy.w * idir.y - ood.y; - float c0loz = nz.x * idir.z - ood.z; - float c0hiz = nz.y * idir.z - ood.z; + NO_EXTENDED_PRECISION float c0lox = n0xy.x * idir.x - ood.x; + NO_EXTENDED_PRECISION float c0hix = n0xy.y * idir.x - ood.x; + NO_EXTENDED_PRECISION float c0loy = n0xy.z * idir.y - ood.y; + NO_EXTENDED_PRECISION float c0hiy = n0xy.w * idir.y - ood.y; + NO_EXTENDED_PRECISION float c0loz = nz.x * idir.z - ood.z; + NO_EXTENDED_PRECISION float c0hiz = nz.y * idir.z - ood.z; NO_EXTENDED_PRECISION float c0min = max4(min(c0lox, c0hix), min(c0loy, c0hiy), min(c0loz, c0hiz), 0.0f); NO_EXTENDED_PRECISION float c0max = min4(max(c0lox, c0hix), max(c0loy, c0hiy), max(c0loz, c0hiz), t); - float c1loz = nz.z * idir.z - ood.z; - float c1hiz = nz.w * idir.z - ood.z; - float c1lox = n1xy.x * idir.x - ood.x; - float c1hix = n1xy.y * idir.x - ood.x; - float c1loy = n1xy.z * idir.y - ood.y; - float c1hiy = n1xy.w * idir.y - ood.y; + NO_EXTENDED_PRECISION float c1loz = nz.z * idir.z - ood.z; + NO_EXTENDED_PRECISION float c1hiz = nz.w * idir.z - ood.z; + NO_EXTENDED_PRECISION float c1lox = n1xy.x * idir.x - ood.x; + NO_EXTENDED_PRECISION float c1hix = n1xy.y * idir.x - ood.x; + NO_EXTENDED_PRECISION float c1loy = n1xy.z * idir.y - ood.y; + NO_EXTENDED_PRECISION float c1hiy = n1xy.w * idir.y - ood.y; NO_EXTENDED_PRECISION float c1min = max4(min(c1lox, c1hix), min(c1loy, c1hiy), min(c1loz, c1hiz), 0.0f); NO_EXTENDED_PRECISION float c1max = min4(max(c1lox, c1hix), max(c1loy, c1hiy), max(c1loz, c1hiz), t); diff --git a/intern/cycles/kernel/kernel_emission.h b/intern/cycles/kernel/kernel_emission.h index 54bc0717b60..4048bbd9dfc 100644 --- a/intern/cycles/kernel/kernel_emission.h +++ b/intern/cycles/kernel/kernel_emission.h @@ -76,7 +76,7 @@ __device float3 direct_emissive_eval(KernelGlobals *kg, float rando, __device bool direct_emission(KernelGlobals *kg, ShaderData *sd, int lindex, float randt, float rando, float randu, float randv, Ray *ray, BsdfEval *eval, - int *lamp) + bool *is_lamp) { LightSample ls; @@ -92,12 +92,6 @@ __device bool direct_emission(KernelGlobals *kg, ShaderData *sd, int lindex, light_sample(kg, randt, randu, randv, sd->time, sd->P, &ls); } - /* return lamp index for MIS */ - if(ls.use_mis) - *lamp = ls.lamp; - else - *lamp= ~0; - if(ls.pdf == 0.0f) return false; @@ -114,7 +108,7 @@ __device bool direct_emission(KernelGlobals *kg, ShaderData *sd, int lindex, shader_bsdf_eval(kg, sd, ls.D, eval, &bsdf_pdf); - if(ls.use_mis) { + if(ls.shader & SHADER_USE_MIS) { /* multiple importance sampling */ float mis_weight = power_heuristic(ls.pdf, bsdf_pdf); light_eval *= mis_weight; @@ -146,6 +140,9 @@ __device bool direct_emission(KernelGlobals *kg, ShaderData *sd, int lindex, ray->t = 0.0f; } + /* return if it's a lamp for shadow pass */ + *is_lamp = (ls.prim == ~0); + return true; } diff --git a/intern/cycles/kernel/kernel_light.h b/intern/cycles/kernel/kernel_light.h index 6ba3e439329..10a32226f17 100644 --- a/intern/cycles/kernel/kernel_light.h +++ b/intern/cycles/kernel/kernel_light.h @@ -31,7 +31,6 @@ typedef struct LightSample { int prim; /* primitive id for triangle/curve ligths */ int shader; /* shader id */ int lamp; /* lamp id */ - int use_mis; /* for lamps with size zero */ LightType type; /* type of light */ } LightSample; @@ -218,11 +217,10 @@ __device void lamp_light_sample(KernelGlobals *kg, int lamp, LightType type = (LightType)__float_as_int(data0.x); ls->type = type; -#ifdef __LAMP_MIS__ - ls->use_mis = true; -#else - ls->use_mis = false; -#endif + ls->shader = __float_as_int(data1.x); + ls->object = ~0; + ls->prim = ~0; + ls->lamp = lamp; if(type == LIGHT_DISTANT) { /* distant light */ @@ -233,10 +231,6 @@ __device void lamp_light_sample(KernelGlobals *kg, int lamp, if(radius > 0.0f) D = distant_light_sample(D, radius, randu, randv); -#ifdef __LAMP_MIS__ - else - ls->use_mis = false; -#endif ls->P = D; ls->Ng = D; @@ -257,9 +251,6 @@ __device void lamp_light_sample(KernelGlobals *kg, int lamp, ls->D = -D; ls->t = FLT_MAX; ls->eval_fac = 1.0f; -#ifndef __LAMP_MIS__ - ls->use_mis = true; -#endif } #endif else { @@ -271,10 +262,6 @@ __device void lamp_light_sample(KernelGlobals *kg, int lamp, if(radius > 0.0f) /* sphere light */ ls->P += sphere_light_sample(P, ls->P, radius, randu, randv); -#ifdef __LAMP_MIS__ - else - ls->use_mis = false; -#endif ls->D = normalize_len(ls->P - P, &ls->t); ls->Ng = -ls->D; @@ -304,13 +291,6 @@ __device void lamp_light_sample(KernelGlobals *kg, int lamp, float invarea = data2.x; - if(invarea == 0.0f) { -#ifdef __LAMP_MIS__ - ls->use_mis = false; -#endif - invarea = 1.0f; - } - ls->eval_fac = 0.25f*invarea; ls->pdf = invarea; } @@ -318,11 +298,6 @@ __device void lamp_light_sample(KernelGlobals *kg, int lamp, ls->eval_fac *= kernel_data.integrator.inv_pdf_lights; ls->pdf *= lamp_light_pdf(kg, ls->Ng, -ls->D, ls->t); } - - ls->shader = __float_as_int(data1.x); - ls->object = ~0; - ls->prim = ~0; - ls->lamp = lamp; } __device bool lamp_light_eval(KernelGlobals *kg, int lamp, float3 P, float3 D, float t, LightSample *ls) @@ -336,7 +311,9 @@ __device bool lamp_light_eval(KernelGlobals *kg, int lamp, float3 P, float3 D, f ls->object = ~0; ls->prim = ~0; ls->lamp = lamp; - ls->use_mis = false; /* flag not used for eval */ + + if(!(ls->shader & SHADER_USE_MIS)) + return false; if(type == LIGHT_DISTANT) { /* distant light */ @@ -475,7 +452,7 @@ __device void triangle_light_sample(KernelGlobals *kg, int prim, int object, ls->object = object; ls->prim = prim; ls->lamp = ~0; - ls->use_mis = true; + ls->shader |= SHADER_USE_MIS; ls->t = 0.0f; ls->type = LIGHT_AREA; ls->eval_fac = 1.0f; @@ -529,11 +506,10 @@ __device void curve_segment_light_sample(KernelGlobals *kg, int prim, int object ls->object = object; ls->prim = prim; ls->lamp = ~0; - ls->use_mis = true; ls->t = 0.0f; ls->type = LIGHT_STRAND; ls->eval_fac = 1.0f; - ls->shader = __float_as_int(v00.z); + ls->shader = __float_as_int(v00.z) | SHADER_USE_MIS; object_transform_light_sample(kg, ls, object, time); } diff --git a/intern/cycles/kernel/kernel_path.h b/intern/cycles/kernel/kernel_path.h index 1a5df66e6c2..9b13f4d2a42 100644 --- a/intern/cycles/kernel/kernel_path.h +++ b/intern/cycles/kernel/kernel_path.h @@ -254,7 +254,7 @@ __device float4 kernel_path_progressive(KernelGlobals *kg, RNG *rng, int sample, bool hit = scene_intersect(kg, &ray, visibility, &isect); #ifdef __LAMP_MIS__ - if(kernel_data.integrator.pdf_lights > 0.0f && !(state.flag & PATH_RAY_CAMERA)) { + if(kernel_data.integrator.use_lamp_mis && !(state.flag & PATH_RAY_CAMERA)) { /* ray starting from previous non-transparent bounce */ Ray light_ray; @@ -399,20 +399,19 @@ __device float4 kernel_path_progressive(KernelGlobals *kg, RNG *rng, int sample, Ray light_ray; BsdfEval L_light; - int lamp; + bool is_lamp; #ifdef __OBJECT_MOTION__ light_ray.time = sd.time; #endif - if(direct_emission(kg, &sd, -1, light_t, light_o, light_u, light_v, &light_ray, &L_light, &lamp)) { + if(direct_emission(kg, &sd, -1, light_t, light_o, light_u, light_v, &light_ray, &L_light, &is_lamp)) { /* trace shadow ray */ float3 shadow; if(!shadow_blocked(kg, &state, &light_ray, &shadow)) { /* accumulate */ - bool is_lamp = (lamp != ~0); - path_radiance_accum_light(&L, throughput, &L_light, shadow, state.bounce, is_lamp); + path_radiance_accum_light(&L, throughput, &L_light, shadow, 1.0f, state.bounce, is_lamp); } } } @@ -501,7 +500,7 @@ __device void kernel_path_indirect(KernelGlobals *kg, RNG *rng, int sample, Ray bool hit = scene_intersect(kg, &ray, visibility, &isect); #ifdef __LAMP_MIS__ - if(kernel_data.integrator.pdf_lights > 0.0f && !(state.flag & PATH_RAY_CAMERA)) { + if(kernel_data.integrator.use_lamp_mis && !(state.flag & PATH_RAY_CAMERA)) { /* ray starting from previous non-transparent bounce */ Ray light_ray; @@ -612,21 +611,20 @@ __device void kernel_path_indirect(KernelGlobals *kg, RNG *rng, int sample, Ray Ray light_ray; BsdfEval L_light; - int lamp; + bool is_lamp; #ifdef __OBJECT_MOTION__ light_ray.time = sd.time; #endif /* sample random light */ - if(direct_emission(kg, &sd, -1, light_t, light_o, light_u, light_v, &light_ray, &L_light, &lamp)) { + if(direct_emission(kg, &sd, -1, light_t, light_o, light_u, light_v, &light_ray, &L_light, &is_lamp)) { /* trace shadow ray */ float3 shadow; if(!shadow_blocked(kg, &state, &light_ray, &shadow)) { /* accumulate */ - bool is_lamp = (lamp != ~0); - path_radiance_accum_light(L, throughput, &L_light, shadow, state.bounce, is_lamp); + path_radiance_accum_light(L, throughput, &L_light, shadow, 1.0f, state.bounce, is_lamp); } } } @@ -819,7 +817,7 @@ __device float4 kernel_path_non_progressive(KernelGlobals *kg, RNG *rng, int sam if(sd.flag & SD_BSDF_HAS_EVAL) { Ray light_ray; BsdfEval L_light; - int lamp; + bool is_lamp; #ifdef __OBJECT_MOTION__ light_ray.time = sd.time; @@ -837,14 +835,13 @@ __device float4 kernel_path_non_progressive(KernelGlobals *kg, RNG *rng, int sam float light_u = path_rng(kg, rng, sample*num_samples + j, rng_offset + PRNG_LIGHT_U); float light_v = path_rng(kg, rng, sample*num_samples + j, rng_offset + PRNG_LIGHT_V); - if(direct_emission(kg, &sd, i, 0.0f, 0.0f, light_u, light_v, &light_ray, &L_light, &lamp)) { + if(direct_emission(kg, &sd, i, 0.0f, 0.0f, light_u, light_v, &light_ray, &L_light, &is_lamp)) { /* trace shadow ray */ float3 shadow; if(!shadow_blocked(kg, &state, &light_ray, &shadow)) { /* accumulate */ - bool is_lamp = (lamp != ~0); - path_radiance_accum_light(&L, throughput*num_samples_inv, &L_light, shadow, state.bounce, is_lamp); + path_radiance_accum_light(&L, throughput*num_samples_inv, &L_light, shadow, num_samples_inv, state.bounce, is_lamp); } } } @@ -867,14 +864,13 @@ __device float4 kernel_path_non_progressive(KernelGlobals *kg, RNG *rng, int sam if(kernel_data.integrator.num_all_lights) light_t = 0.5f*light_t; - if(direct_emission(kg, &sd, -1, light_t, 0.0f, light_u, light_v, &light_ray, &L_light, &lamp)) { + if(direct_emission(kg, &sd, -1, light_t, 0.0f, light_u, light_v, &light_ray, &L_light, &is_lamp)) { /* trace shadow ray */ float3 shadow; if(!shadow_blocked(kg, &state, &light_ray, &shadow)) { /* accumulate */ - bool is_lamp = (lamp != ~0); - path_radiance_accum_light(&L, throughput*num_samples_inv, &L_light, shadow, state.bounce, is_lamp); + path_radiance_accum_light(&L, throughput*num_samples_inv, &L_light, shadow, num_samples_inv, state.bounce, is_lamp); } } } diff --git a/intern/cycles/kernel/kernel_sse2.cpp b/intern/cycles/kernel/kernel_sse2.cpp new file mode 100644 index 00000000000..e9d482ae5cf --- /dev/null +++ b/intern/cycles/kernel/kernel_sse2.cpp @@ -0,0 +1,60 @@ +/* + * Copyright 2011, Blender Foundation. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. + */ + +/* Optimized CPU kernel entry points. This file is compiled with SSE2 + * optimization flags and nearly all functions inlined, while kernel.cpp + * is compiled without for other CPU's. */ + +#ifdef WITH_OPTIMIZED_KERNEL + +#include "kernel.h" +#include "kernel_compat_cpu.h" +#include "kernel_math.h" +#include "kernel_types.h" +#include "kernel_globals.h" +#include "kernel_film.h" +#include "kernel_path.h" +#include "kernel_displace.h" + +CCL_NAMESPACE_BEGIN + +/* Path Tracing */ + +void kernel_cpu_sse2_path_trace(KernelGlobals *kg, float *buffer, unsigned int *rng_state, int sample, int x, int y, int offset, int stride) +{ + kernel_path_trace(kg, buffer, rng_state, sample, x, y, offset, stride); +} + +/* Tonemapping */ + +void kernel_cpu_sse2_tonemap(KernelGlobals *kg, uchar4 *rgba, float *buffer, int sample, int resolution, int x, int y, int offset, int stride) +{ + kernel_film_tonemap(kg, rgba, buffer, sample, resolution, x, y, offset, stride); +} + +/* Shader Evaluate */ + +void kernel_cpu_sse2_shader(KernelGlobals *kg, uint4 *input, float4 *output, int type, int i) +{ + kernel_shader_evaluate(kg, input, output, (ShaderEvalType)type, i); +} + +CCL_NAMESPACE_END + +#endif + diff --git a/intern/cycles/kernel/kernel_optimized.cpp b/intern/cycles/kernel/kernel_sse3.cpp index 0b662095133..9a8b389cf68 100644 --- a/intern/cycles/kernel/kernel_optimized.cpp +++ b/intern/cycles/kernel/kernel_sse3.cpp @@ -35,21 +35,21 @@ CCL_NAMESPACE_BEGIN /* Path Tracing */ -void kernel_cpu_optimized_path_trace(KernelGlobals *kg, float *buffer, unsigned int *rng_state, int sample, int x, int y, int offset, int stride) +void kernel_cpu_sse3_path_trace(KernelGlobals *kg, float *buffer, unsigned int *rng_state, int sample, int x, int y, int offset, int stride) { kernel_path_trace(kg, buffer, rng_state, sample, x, y, offset, stride); } /* Tonemapping */ -void kernel_cpu_optimized_tonemap(KernelGlobals *kg, uchar4 *rgba, float *buffer, int sample, int resolution, int x, int y, int offset, int stride) +void kernel_cpu_sse3_tonemap(KernelGlobals *kg, uchar4 *rgba, float *buffer, int sample, int resolution, int x, int y, int offset, int stride) { kernel_film_tonemap(kg, rgba, buffer, sample, resolution, x, y, offset, stride); } /* Shader Evaluate */ -void kernel_cpu_optimized_shader(KernelGlobals *kg, uint4 *input, float4 *output, int type, int i) +void kernel_cpu_sse3_shader(KernelGlobals *kg, uint4 *input, float4 *output, int type, int i) { kernel_shader_evaluate(kg, input, output, (ShaderEvalType)type, i); } diff --git a/intern/cycles/kernel/kernel_types.h b/intern/cycles/kernel/kernel_types.h index 1236f43e018..52bdddc72fb 100644 --- a/intern/cycles/kernel/kernel_types.h +++ b/intern/cycles/kernel/kernel_types.h @@ -44,7 +44,6 @@ CCL_NAMESPACE_BEGIN #define __KERNEL_SHADING__ #define __KERNEL_ADV_SHADING__ #define __NON_PROGRESSIVE__ -#define __LAMP_MIS__ #define __HAIR__ #ifdef WITH_OSL #define __OSL__ @@ -56,9 +55,6 @@ CCL_NAMESPACE_BEGIN #if __CUDA_ARCH__ >= 200 #define __KERNEL_ADV_SHADING__ #endif -#if __CUDA_ARCH__ >= 210 -#define __LAMP_MIS__ -#endif #endif #ifdef __KERNEL_OPENCL__ @@ -116,6 +112,7 @@ CCL_NAMESPACE_BEGIN #define __TRANSPARENT_SHADOWS__ #define __PASSES__ #define __BACKGROUND_MIS__ +#define __LAMP_MIS__ #define __AO__ #define __ANISOTROPIC__ #define __CAMERA_MOTION__ @@ -288,8 +285,9 @@ typedef enum ShaderFlag { SHADER_SMOOTH_NORMAL = (1 << 31), SHADER_CAST_SHADOW = (1 << 30), SHADER_AREA_LIGHT = (1 << 29), + SHADER_USE_MIS = (1 << 28), - SHADER_MASK = ~(SHADER_SMOOTH_NORMAL|SHADER_CAST_SHADOW|SHADER_AREA_LIGHT) + SHADER_MASK = ~(SHADER_SMOOTH_NORMAL|SHADER_CAST_SHADOW|SHADER_AREA_LIGHT|SHADER_USE_MIS) } ShaderFlag; /* Light Type */ @@ -680,7 +678,7 @@ typedef struct KernelIntegrator { int transmission_samples; int ao_samples; int mesh_light_samples; - int pad1; + int use_lamp_mis; } KernelIntegrator; typedef struct KernelBVH { diff --git a/intern/cycles/kernel/osl/SConscript b/intern/cycles/kernel/osl/SConscript index fe7fec463a7..09899567128 100644 --- a/intern/cycles/kernel/osl/SConscript +++ b/intern/cycles/kernel/osl/SConscript @@ -46,6 +46,7 @@ defs.append('WITH_OSL') if env['OURPLATFORM'] in ('win32-vc', 'win64-vc'): cxxflags.append('-DBOOST_NO_RTTI -DBOOST_NO_TYPEID'.split()) incs.append(env['BF_PTHREADS_INC']) + defs.append('OSL_STATIC_LIBRARY') else: cxxflags.append('-fno-rtti -DBOOST_NO_RTTI -DBOOST_NO_TYPEID'.split()) diff --git a/intern/cycles/kernel/shaders/node_musgrave_texture.osl b/intern/cycles/kernel/shaders/node_musgrave_texture.osl index 38232ea0aeb..8675c23d0d8 100644 --- a/intern/cycles/kernel/shaders/node_musgrave_texture.osl +++ b/intern/cycles/kernel/shaders/node_musgrave_texture.osl @@ -37,14 +37,14 @@ float noise_musgrave_fBm(point p, string basis, float H, float lacunarity, float int i; for (i = 0; i < (int)octaves; i++) { - value += safe_noise(p) * pwr; + value += safe_noise(p, 0) * pwr; pwr *= pwHL; p *= lacunarity; } rmd = octaves - floor(octaves); if (rmd != 0.0) - value += rmd * safe_noise(p) * pwr; + value += rmd * safe_noise(p, 0) * pwr; return value; } @@ -65,14 +65,14 @@ float noise_musgrave_multi_fractal(point p, string basis, float H, float lacunar int i; for (i = 0; i < (int)octaves; i++) { - value *= (pwr * safe_noise(p) + 1.0); + value *= (pwr * safe_noise(p, 0) + 1.0); pwr *= pwHL; p *= lacunarity; } rmd = octaves - floor(octaves); if (rmd != 0.0) - value *= (rmd * pwr * safe_noise(p) + 1.0); /* correct? */ + value *= (rmd * pwr * safe_noise(p, 0) + 1.0); /* correct? */ return value; } @@ -93,11 +93,11 @@ float noise_musgrave_hetero_terrain(point p, string basis, float H, float lacuna int i; /* first unscaled octave of function; later octaves are scaled */ - value = offset + safe_noise(p); + value = offset + safe_noise(p, 0); p *= lacunarity; for (i = 1; i < (int)octaves; i++) { - increment = (safe_noise(p) + offset) * pwr * value; + increment = (safe_noise(p, 0) + offset) * pwr * value; value += increment; pwr *= pwHL; p *= lacunarity; @@ -105,7 +105,7 @@ float noise_musgrave_hetero_terrain(point p, string basis, float H, float lacuna rmd = octaves - floor(octaves); if (rmd != 0.0) { - increment = (safe_noise(p) + offset) * pwr * value; + increment = (safe_noise(p, 0) + offset) * pwr * value; value += rmd * increment; } @@ -128,7 +128,7 @@ float noise_musgrave_hybrid_multi_fractal(point p, string basis, float H, float pwr = pwHL; int i; - result = safe_noise(p) + offset; + result = safe_noise(p, 0) + offset; weight = gain * result; p *= lacunarity; @@ -136,7 +136,7 @@ float noise_musgrave_hybrid_multi_fractal(point p, string basis, float H, if (weight > 1.0) weight = 1.0; - signal = (safe_noise(p) + offset) * pwr; + signal = (safe_noise(p, 0) + offset) * pwr; pwr *= pwHL; result += weight * signal; weight *= gain * signal; @@ -145,7 +145,7 @@ float noise_musgrave_hybrid_multi_fractal(point p, string basis, float H, rmd = octaves - floor(octaves); if (rmd != 0.0) - result += rmd * ((safe_noise(p) + offset) * pwr); + result += rmd * ((safe_noise(p, 0) + offset) * pwr); return result; } @@ -166,7 +166,7 @@ float noise_musgrave_ridged_multi_fractal(point p, string basis, float H, float pwr = pwHL; int i; - signal = offset - fabs(safe_noise(p)); + signal = offset - fabs(safe_noise(p, 0)); signal *= signal; result = signal; weight = 1.0; @@ -174,7 +174,7 @@ float noise_musgrave_ridged_multi_fractal(point p, string basis, float H, for (i = 1; i < (int)octaves; i++) { p *= lacunarity; weight = clamp(signal * gain, 0.0, 1.0); - signal = offset - fabs(safe_noise(p)); + signal = offset - fabs(safe_noise(p, 0)); signal *= signal; signal *= weight; result += signal * pwr; diff --git a/intern/cycles/kernel/shaders/node_texture.h b/intern/cycles/kernel/shaders/node_texture.h index 2de0fc0ea57..463c68a6c27 100644 --- a/intern/cycles/kernel/shaders/node_texture.h +++ b/intern/cycles/kernel/shaders/node_texture.h @@ -151,9 +151,17 @@ float voronoi_CrS(point p) { return 2.0 * voronoi_Cr(p) - 1.0; } /* Noise Bases */ -float safe_noise(point p) +float safe_noise(point p, int type) { - float f = noise(p); + float f = 0.0; + + /* Perlin noise in range -1..1 */ + if (type == 0) + f = noise("perlin", p); + + /* Perlin noise in range 0..1 */ + else + f = noise(p); /* can happen for big coordinates, things even out to 0.5 then anyway */ if(!isfinite(f)) @@ -167,7 +175,7 @@ float noise_basis(point p, string basis) float result = 0.0; if (basis == "Perlin") - result = safe_noise(p); /* returns perlin noise in range 0..1 */ + result = safe_noise(p, 1); if (basis == "Voronoi F1") result = voronoi_F1S(p); if (basis == "Voronoi F2") |