diff options
63 files changed, 3338 insertions, 1571 deletions
diff --git a/build_files/cmake/config/blender_full.cmake b/build_files/cmake/config/blender_full.cmake index bd8a2dbaf2c..87b8ed5f921 100644 --- a/build_files/cmake/config/blender_full.cmake +++ b/build_files/cmake/config/blender_full.cmake @@ -77,6 +77,6 @@ elseif(APPLE) set(WITH_OPENSUBDIV OFF CACHE BOOL "" FORCE) set(WITH_CODEC_QUICKTIME ON CACHE BOOL "" FORCE) - include("${CMAKE_SOURCE_DIR}/build_files/cmake/platform/platform_apple_xcode.cmake") + include("${CMAKE_CURRENT_SOURCE_DIR}/../platform/platform_apple_xcode.cmake") apple_check_quicktime() endif() diff --git a/build_files/cmake/config/blender_release.cmake b/build_files/cmake/config/blender_release.cmake index 42e8c111714..b60c8c7380c 100644 --- a/build_files/cmake/config/blender_release.cmake +++ b/build_files/cmake/config/blender_release.cmake @@ -78,6 +78,6 @@ elseif(APPLE) set(WITH_OPENSUBDIV OFF CACHE BOOL "" FORCE) set(WITH_CODEC_QUICKTIME ON CACHE BOOL "" FORCE) - include("${CMAKE_SOURCE_DIR}/build_files/cmake/platform/platform_apple_xcode.cmake") + include("${CMAKE_CURRENT_SOURCE_DIR}/../platform/platform_apple_xcode.cmake") apple_check_quicktime() endif() diff --git a/intern/cycles/kernel/CMakeLists.txt b/intern/cycles/kernel/CMakeLists.txt index c3772dfa2d8..3750225571d 100644 --- a/intern/cycles/kernel/CMakeLists.txt +++ b/intern/cycles/kernel/CMakeLists.txt @@ -196,10 +196,42 @@ set(SRC_UTIL_HEADERS ../util/util_math.h ../util/util_math_fast.h ../util/util_math_intersect.h + ../util/util_math_float2.h + ../util/util_math_float3.h + ../util/util_math_float4.h + ../util/util_math_int2.h + ../util/util_math_int3.h + ../util/util_math_int4.h ../util/util_static_assert.h ../util/util_transform.h ../util/util_texture.h ../util/util_types.h + ../util/util_types_float2.h + ../util/util_types_float2_impl.h + ../util/util_types_float3.h + ../util/util_types_float3_impl.h + ../util/util_types_float4.h + ../util/util_types_float4_impl.h + ../util/util_types_int2.h + ../util/util_types_int2_impl.h + ../util/util_types_int3.h + ../util/util_types_int3_impl.h + ../util/util_types_int4.h + ../util/util_types_int4_impl.h + ../util/util_types_uchar2.h + ../util/util_types_uchar2_impl.h + ../util/util_types_uchar3.h + ../util/util_types_uchar3_impl.h + ../util/util_types_uchar4.h + ../util/util_types_uchar4_impl.h + ../util/util_types_uint2.h + ../util/util_types_uint2_impl.h + ../util/util_types_uint3.h + ../util/util_types_uint3_impl.h + ../util/util_types_uint4.h + ../util/util_types_uint4_impl.h + ../util/util_types_vector3.h + ../util/util_types_vector3_impl.h ) set(SRC_SPLIT_HEADERS diff --git a/intern/cycles/kernel/kernel_path_branched.h b/intern/cycles/kernel/kernel_path_branched.h index 36fd6c95fe7..085eb42325d 100644 --- a/intern/cycles/kernel/kernel_path_branched.h +++ b/intern/cycles/kernel/kernel_path_branched.h @@ -337,11 +337,6 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in float num_samples_inv = 1.0f/num_samples; for(int j = 0; j < num_samples; j++) { - /* workaround to fix correlation bug in T38710, can find better solution - * in random number generator later, for now this is done here to not impact - * performance of rendering without volumes */ - RNG tmp_rng = cmj_hash(*rng, state.rng_offset); - PathState ps = state; Ray pray = ray; float3 tp = throughput; @@ -352,8 +347,8 @@ ccl_device float4 kernel_branched_path_integrate(KernelGlobals *kg, RNG *rng, in /* scatter sample. if we use distance sampling and take just one * sample for direct and indirect light, we could share this * computation, but makes code a bit complex */ - float rphase = path_state_rng_1D_for_decision(kg, &tmp_rng, &ps, PRNG_PHASE); - float rscatter = path_state_rng_1D_for_decision(kg, &tmp_rng, &ps, PRNG_SCATTER_DISTANCE); + float rphase = path_state_rng_1D_for_decision(kg, rng, &ps, PRNG_PHASE); + float rscatter = path_state_rng_1D_for_decision(kg, rng, &ps, PRNG_SCATTER_DISTANCE); VolumeIntegrateResult result = kernel_volume_decoupled_scatter(kg, &ps, &pray, &sd, &tp, rphase, rscatter, &volume_segment, NULL, false); diff --git a/intern/cycles/kernel/kernel_random.h b/intern/cycles/kernel/kernel_random.h index d4f0caff5de..e8a912ccc0b 100644 --- a/intern/cycles/kernel/kernel_random.h +++ b/intern/cycles/kernel/kernel_random.h @@ -20,14 +20,15 @@ CCL_NAMESPACE_BEGIN #ifdef __SOBOL__ -/* skip initial numbers that are not as well distributed, especially the +/* Skip initial numbers that are not as well distributed, especially the * first sequence is just 0 everywhere, which can be problematic for e.g. - * path termination */ + * path termination. + */ #define SOBOL_SKIP 64 -/* High Dimensional Sobol */ +/* High Dimensional Sobol. */ -/* van der corput radical inverse */ +/* Van der Corput radical inverse. */ ccl_device uint van_der_corput(uint bits) { bits = (bits << 16) | (bits >> 16); @@ -38,58 +39,63 @@ ccl_device uint van_der_corput(uint bits) return bits; } -/* sobol radical inverse */ +/* Sobol radical inverse. */ ccl_device uint sobol(uint i) { uint r = 0; - - for(uint v = 1U << 31; i; i >>= 1, v ^= v >> 1) - if(i & 1) + for(uint v = 1U << 31; i; i >>= 1, v ^= v >> 1) { + if(i & 1) { r ^= v; - + } + } return r; } -/* inverse of sobol radical inverse */ +/* Inverse of sobol radical inverse. */ ccl_device uint sobol_inverse(uint i) { const uint msb = 1U << 31; uint r = 0; - - for(uint v = 1; i; i <<= 1, v ^= v << 1) - if(i & msb) + for(uint v = 1; i; i <<= 1, v ^= v << 1) { + if(i & msb) { r ^= v; - + } + } return r; } -/* multidimensional sobol with generator matrices - * dimension 0 and 1 are equal to van_der_corput() and sobol() respectively */ +/* Multidimensional sobol with generator matrices + * dimension 0 and 1 are equal to van_der_corput() and sobol() respectively. + */ ccl_device uint sobol_dimension(KernelGlobals *kg, int index, int dimension) { uint result = 0; uint i = index; - - for(uint j = 0; i; i >>= 1, j++) - if(i & 1) + for(uint j = 0; i; i >>= 1, j++) { + if(i & 1) { result ^= kernel_tex_fetch(__sobol_directions, 32*dimension + j); - + } + } return result; } -/* lookup index and x/y coordinate, assumes m is a power of two */ -ccl_device uint sobol_lookup(const uint m, const uint frame, const uint ex, const uint ey, uint *x, uint *y) +/* Lookup index and x/y coordinate, assumes m is a power of two. */ +ccl_device uint sobol_lookup(const uint m, + const uint frame, + const uint ex, + const uint ey, + uint *x, uint *y) { - /* shift is constant per frame */ + /* Shift is constant per frame. */ const uint shift = frame << (m << 1); const uint sobol_shift = sobol(shift); - /* van der Corput is its own inverse */ + /* Van der Corput is its own inverse. */ const uint lower = van_der_corput(ex << (32 - m)); - /* need to compensate for ey difference and shift */ + /* Need to compensate for ey difference and shift. */ const uint sobol_lower = sobol(lower); - const uint mask = ~-(1 << m) << (32 - m); /* only m upper bits */ + const uint mask = ~-(1 << m) << (32 - m); /* Only m upper bits. */ const uint delta = ((ey << (32 - m)) ^ sobol_lower ^ sobol_shift) & mask; - /* only use m upper bits for the index (m is a power of two) */ + /* Only use m upper bits for the index (m is a power of two). */ const uint sobol_result = delta | (delta >> m); const uint upper = sobol_inverse(sobol_result); const uint index = shift | upper | lower; @@ -98,11 +104,14 @@ ccl_device uint sobol_lookup(const uint m, const uint frame, const uint ex, cons return index; } -ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension) +ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, + RNG *rng, + int sample, int num_samples, + int dimension) { #ifdef __CMJ__ if(kernel_data.integrator.sampling_pattern == SAMPLING_PATTERN_CMJ) { - /* correlated multi-jittered */ + /* Correlated multi-jitter. */ int p = *rng + dimension; return cmj_sample_1D(sample, num_samples, p); } @@ -113,7 +122,7 @@ ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, RNG *rng, int sample float r = (float)result * (1.0f/(float)0xFFFFFFFF); return r; #else - /* compute sobol sequence value using direction vectors */ + /* Compute sobol sequence value using direction vectors. */ uint result = sobol_dimension(kg, sample + SOBOL_SKIP, dimension); float r = (float)result * (1.0f/(float)0xFFFFFFFF); @@ -130,24 +139,33 @@ ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, RNG *rng, int sample #endif } -ccl_device_forceinline void path_rng_2D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension, float *fx, float *fy) +ccl_device_forceinline void path_rng_2D(KernelGlobals *kg, + RNG *rng, + int sample, int num_samples, + int dimension, + float *fx, float *fy) { #ifdef __CMJ__ if(kernel_data.integrator.sampling_pattern == SAMPLING_PATTERN_CMJ) { - /* correlated multi-jittered */ + /* Correlated multi-jitter. */ int p = *rng + dimension; cmj_sample_2D(sample, num_samples, p, fx, fy); } else #endif { - /* sobol */ + /* Sobol. */ *fx = path_rng_1D(kg, rng, sample, num_samples, dimension); *fy = path_rng_1D(kg, rng, sample, num_samples, dimension + 1); } } -ccl_device_inline void path_rng_init(KernelGlobals *kg, ccl_global uint *rng_state, int sample, int num_samples, RNG *rng, int x, int y, float *fx, float *fy) +ccl_device_inline void path_rng_init(KernelGlobals *kg, + ccl_global uint *rng_state, + int sample, int num_samples, + RNG *rng, + int x, int y, + float *fx, float *fy) { #ifdef __SOBOL_FULL_SCREEN__ uint px, py; @@ -182,29 +200,43 @@ ccl_device_inline void path_rng_init(KernelGlobals *kg, ccl_global uint *rng_sta #endif } -ccl_device void path_rng_end(KernelGlobals *kg, ccl_global uint *rng_state, RNG rng) +ccl_device void path_rng_end(KernelGlobals *kg, + ccl_global uint *rng_state, + RNG rng) { /* nothing to do */ } -#else +#else /* __SOBOL__ */ /* Linear Congruential Generator */ -ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension) +ccl_device_forceinline float path_rng_1D(KernelGlobals *kg, + RNG *rng, + int sample, int num_samples, + int dimension) { /* implicit mod 2^32 */ *rng = (1103515245*(*rng) + 12345); return (float)*rng * (1.0f/(float)0xFFFFFFFF); } -ccl_device_inline void path_rng_2D(KernelGlobals *kg, RNG *rng, int sample, int num_samples, int dimension, float *fx, float *fy) +ccl_device_inline void path_rng_2D(KernelGlobals *kg, + RNG *rng, + int sample, int num_samples, + int dimension, + float *fx, float *fy) { *fx = path_rng_1D(kg, rng, sample, num_samples, dimension); *fy = path_rng_1D(kg, rng, sample, num_samples, dimension + 1); } -ccl_device void path_rng_init(KernelGlobals *kg, ccl_global uint *rng_state, int sample, int num_samples, RNG *rng, int x, int y, float *fx, float *fy) +ccl_device void path_rng_init(KernelGlobals *kg, + ccl_global uint *rng_state, + int sample, int num_samples, + RNG *rng, + int x, int y, + float *fx, float *fy) { /* load state */ *rng = *rng_state; @@ -220,13 +252,15 @@ ccl_device void path_rng_init(KernelGlobals *kg, ccl_global uint *rng_state, int } } -ccl_device void path_rng_end(KernelGlobals *kg, ccl_global uint *rng_state, RNG rng) +ccl_device void path_rng_end(KernelGlobals *kg, + ccl_global uint *rng_state, + RNG rng) { /* store state for next sample */ *rng_state = rng; } -#endif +#endif /* __SOBOL__ */ /* Linear Congruential Generator */ @@ -257,49 +291,108 @@ ccl_device uint lcg_init(uint seed) * dimension to avoid using the same sequence twice. * * For branches in the path we must be careful not to reuse the same number - * in a sequence and offset accordingly. */ + * in a sequence and offset accordingly. + */ -ccl_device_inline float path_state_rng_1D(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int dimension) +ccl_device_inline float path_state_rng_1D(KernelGlobals *kg, + RNG *rng, + const ccl_addr_space PathState *state, + int dimension) { - return path_rng_1D(kg, rng, state->sample, state->num_samples, state->rng_offset + dimension); + return path_rng_1D(kg, + rng, + state->sample, state->num_samples, + state->rng_offset + dimension); } -ccl_device_inline float path_state_rng_1D_for_decision(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int dimension) +ccl_device_inline float path_state_rng_1D_for_decision( + KernelGlobals *kg, + RNG *rng, + const ccl_addr_space PathState *state, + int dimension) { - /* the rng_offset is not increased for transparent bounces. if we do then + /* The rng_offset is not increased for transparent bounces. if we do then * fully transparent objects can become subtly visible by the different * sampling patterns used where the transparent object is. * * however for some random numbers that will determine if we next bounce * is transparent we do need to increase the offset to avoid always making - * the same decision */ - int rng_offset = state->rng_offset + state->transparent_bounce*PRNG_BOUNCE_NUM; - return path_rng_1D(kg, rng, state->sample, state->num_samples, rng_offset + dimension); + * the same decision. */ + const int rng_offset = state->rng_offset + state->transparent_bounce * PRNG_BOUNCE_NUM; + return path_rng_1D(kg, + rng, + state->sample, state->num_samples, + rng_offset + dimension); } -ccl_device_inline void path_state_rng_2D(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int dimension, float *fx, float *fy) +ccl_device_inline void path_state_rng_2D(KernelGlobals *kg, + RNG *rng, + const ccl_addr_space PathState *state, + int dimension, + float *fx, float *fy) { - path_rng_2D(kg, rng, state->sample, state->num_samples, state->rng_offset + dimension, fx, fy); + path_rng_2D(kg, + rng, + state->sample, state->num_samples, + state->rng_offset + dimension, + fx, fy); } -ccl_device_inline float path_branched_rng_1D(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int branch, int num_branches, int dimension) +ccl_device_inline float path_branched_rng_1D( + KernelGlobals *kg, + RNG *rng, + const ccl_addr_space PathState *state, + int branch, + int num_branches, + int dimension) { - return path_rng_1D(kg, rng, state->sample*num_branches + branch, state->num_samples*num_branches, state->rng_offset + dimension); + return path_rng_1D(kg, + rng, + state->sample * num_branches + branch, + state->num_samples * num_branches, + state->rng_offset + dimension); } -ccl_device_inline float path_branched_rng_1D_for_decision(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int branch, int num_branches, int dimension) +ccl_device_inline float path_branched_rng_1D_for_decision( + KernelGlobals *kg, + RNG *rng, + const ccl_addr_space PathState *state, + int branch, + int num_branches, + int dimension) { - int rng_offset = state->rng_offset + state->transparent_bounce*PRNG_BOUNCE_NUM; - return path_rng_1D(kg, rng, state->sample*num_branches + branch, state->num_samples*num_branches, rng_offset + dimension); + const int rng_offset = state->rng_offset + state->transparent_bounce * PRNG_BOUNCE_NUM; + return path_rng_1D(kg, + rng, + state->sample * num_branches + branch, + state->num_samples * num_branches, + rng_offset + dimension); } -ccl_device_inline void path_branched_rng_2D(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int branch, int num_branches, int dimension, float *fx, float *fy) +ccl_device_inline void path_branched_rng_2D( + KernelGlobals *kg, + RNG *rng, + const ccl_addr_space PathState *state, + int branch, + int num_branches, + int dimension, + float *fx, float *fy) { - path_rng_2D(kg, rng, state->sample*num_branches + branch, state->num_samples*num_branches, state->rng_offset + dimension, fx, fy); + path_rng_2D(kg, + rng, + state->sample * num_branches + branch, + state->num_samples * num_branches, + state->rng_offset + dimension, + fx, fy); } -/* Utitility functions to get light termination value, since it might not be needed in many cases. */ -ccl_device_inline float path_state_rng_light_termination(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state) +/* Utitility functions to get light termination value, + * since it might not be needed in many cases. + */ +ccl_device_inline float path_state_rng_light_termination( + KernelGlobals *kg, + RNG *rng, + const ccl_addr_space PathState *state) { if(kernel_data.integrator.light_inv_rr_threshold > 0.0f) { return path_state_rng_1D_for_decision(kg, rng, state, PRNG_LIGHT_TERMINATE); @@ -307,15 +400,27 @@ ccl_device_inline float path_state_rng_light_termination(KernelGlobals *kg, RNG return 0.0f; } -ccl_device_inline float path_branched_rng_light_termination(KernelGlobals *kg, RNG *rng, const ccl_addr_space PathState *state, int branch, int num_branches) +ccl_device_inline float path_branched_rng_light_termination( + KernelGlobals *kg, + RNG *rng, + const ccl_addr_space PathState *state, + int branch, + int num_branches) { if(kernel_data.integrator.light_inv_rr_threshold > 0.0f) { - return path_branched_rng_1D_for_decision(kg, rng, state, branch, num_branches, PRNG_LIGHT_TERMINATE); + return path_branched_rng_1D_for_decision(kg, + rng, + state, + branch, + num_branches, + PRNG_LIGHT_TERMINATE); } return 0.0f; } -ccl_device_inline void path_state_branch(ccl_addr_space PathState *state, int branch, int num_branches) +ccl_device_inline void path_state_branch(ccl_addr_space PathState *state, + int branch, + int num_branches) { /* path is splitting into a branch, adjust so that each branch * still gets a unique sample from the same sequence */ @@ -324,14 +429,17 @@ ccl_device_inline void path_state_branch(ccl_addr_space PathState *state, int br state->num_samples = state->num_samples*num_branches; } -ccl_device_inline uint lcg_state_init(RNG *rng, int rng_offset, int sample, uint scramble) +ccl_device_inline uint lcg_state_init(RNG *rng, + int rng_offset, + int sample, + uint scramble) { return lcg_init(*rng + rng_offset + sample*scramble); } ccl_device float lcg_step_float_addrspace(ccl_addr_space uint *rng) { - /* implicit mod 2^32 */ + /* Implicit mod 2^32 */ *rng = (1103515245*(*rng) + 12345); return (float)*rng * (1.0f/(float)0xFFFFFFFF); } diff --git a/intern/cycles/util/CMakeLists.txt b/intern/cycles/util/CMakeLists.txt index a015fef8284..388aba65460 100644 --- a/intern/cycles/util/CMakeLists.txt +++ b/intern/cycles/util/CMakeLists.txt @@ -53,6 +53,12 @@ set(SRC_HEADERS util_math_cdf.h util_math_fast.h util_math_intersect.h + util_math_float2.h + util_math_float3.h + util_math_float4.h + util_math_int2.h + util_math_int3.h + util_math_int4.h util_md5.h util_opengl.h util_optimization.h @@ -80,6 +86,32 @@ set(SRC_HEADERS util_time.h util_transform.h util_types.h + util_types_float2.h + util_types_float2_impl.h + util_types_float3.h + util_types_float3_impl.h + util_types_float4.h + util_types_float4_impl.h + util_types_int2.h + util_types_int2_impl.h + util_types_int3.h + util_types_int3_impl.h + util_types_int4.h + util_types_int4_impl.h + util_types_uchar2.h + util_types_uchar2_impl.h + util_types_uchar3.h + util_types_uchar3_impl.h + util_types_uchar4.h + util_types_uchar4_impl.h + util_types_uint2.h + util_types_uint2_impl.h + util_types_uint3.h + util_types_uint3_impl.h + util_types_uint4.h + util_types_uint4_impl.h + util_types_vector3.h + util_types_vector3_impl.h util_vector.h util_version.h util_view.h diff --git a/intern/cycles/util/util_math.h b/intern/cycles/util/util_math.h index 77781ed4574..52b4fa859b7 100644 --- a/intern/cycles/util/util_math.h +++ b/intern/cycles/util/util_math.h @@ -28,12 +28,10 @@ #ifndef __KERNEL_OPENCL__ - -#include <float.h> -#include <math.h> -#include <stdio.h> - -#endif +# include <float.h> +# include <math.h> +# include <stdio.h> +#endif /* __KERNEL_OPENCL__ */ #include "util/util_types.h" @@ -43,49 +41,44 @@ CCL_NAMESPACE_BEGIN /* Division */ #ifndef M_PI_F -#define M_PI_F (3.1415926535897932f) /* pi */ +# define M_PI_F (3.1415926535897932f) /* pi */ #endif #ifndef M_PI_2_F -#define M_PI_2_F (1.5707963267948966f) /* pi/2 */ +# define M_PI_2_F (1.5707963267948966f) /* pi/2 */ #endif #ifndef M_PI_4_F -#define M_PI_4_F (0.7853981633974830f) /* pi/4 */ +# define M_PI_4_F (0.7853981633974830f) /* pi/4 */ #endif #ifndef M_1_PI_F -#define M_1_PI_F (0.3183098861837067f) /* 1/pi */ +# define M_1_PI_F (0.3183098861837067f) /* 1/pi */ #endif #ifndef M_2_PI_F -#define M_2_PI_F (0.6366197723675813f) /* 2/pi */ +# define M_2_PI_F (0.6366197723675813f) /* 2/pi */ #endif /* Multiplication */ #ifndef M_2PI_F -#define M_2PI_F (6.2831853071795864f) /* 2*pi */ +# define M_2PI_F (6.2831853071795864f) /* 2*pi */ #endif #ifndef M_4PI_F -#define M_4PI_F (12.566370614359172f) /* 4*pi */ +# define M_4PI_F (12.566370614359172f) /* 4*pi */ #endif /* Float sqrt variations */ - #ifndef M_SQRT2_F -#define M_SQRT2_F (1.4142135623730950f) /* sqrt(2) */ +# define M_SQRT2_F (1.4142135623730950f) /* sqrt(2) */ #endif - #ifndef M_LN2_F -#define M_LN2_F (0.6931471805599453f) /* ln(2) */ +# define M_LN2_F (0.6931471805599453f) /* ln(2) */ #endif - #ifndef M_LN10_F -#define M_LN10_F (2.3025850929940457f) /* ln(10) */ +# define M_LN10_F (2.3025850929940457f) /* ln(10) */ #endif /* Scalar */ #ifdef _WIN32 - -#ifndef __KERNEL_OPENCL__ - +# ifndef __KERNEL_OPENCL__ ccl_device_inline float fmaxf(float a, float b) { return (a > b)? a: b; @@ -95,13 +88,10 @@ ccl_device_inline float fminf(float a, float b) { return (a < b)? a: b; } - -#endif - -#endif +# endif /* !__KERNEL_OPENCL__ */ +#endif /* _WIN32 */ #ifndef __KERNEL_GPU__ - using std::isfinite; using std::isnan; @@ -157,8 +147,7 @@ ccl_device_inline T max4(const T& a, const T& b, const T& c, const T& d) { return max(max(a,b),max(c,d)); } - -#endif +#endif /* __KERNEL_GPU__ */ ccl_device_inline float min4(float a, float b, float c, float d) { @@ -170,13 +159,7 @@ ccl_device_inline float max4(float a, float b, float c, float d) return max(max(a, b), max(c, d)); } -ccl_device_inline float max3(float3 a) -{ - return max(max(a.x, a.y), a.z); -} - #ifndef __KERNEL_OPENCL__ - ccl_device_inline int clamp(int a, int mn, int mx) { return min(max(a, mn), mx); @@ -191,17 +174,14 @@ ccl_device_inline float mix(float a, float b, float t) { return a + t*(b - a); } - -#endif +#endif /* __KERNEL_OPENCL__ */ #ifndef __KERNEL_CUDA__ - ccl_device_inline float saturate(float a) { return clamp(a, 0.0f, 1.0f); } - -#endif +#endif /* __KERNEL_CUDA__ */ ccl_device_inline int float_to_int(float f) { @@ -242,453 +222,6 @@ ccl_device_inline int mod(int x, int m) return (x % m + m) % m; } -/* Float2 Vector */ - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline bool is_zero(const float2& a) -{ - return (a.x == 0.0f && a.y == 0.0f); -} - -#endif - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline float average(const float2& a) -{ - return (a.x + a.y)*(1.0f/2.0f); -} - -#endif - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline float2 operator-(const float2& a) -{ - return make_float2(-a.x, -a.y); -} - -ccl_device_inline float2 operator*(const float2& a, const float2& b) -{ - return make_float2(a.x*b.x, a.y*b.y); -} - -ccl_device_inline float2 operator*(const float2& a, float f) -{ - return make_float2(a.x*f, a.y*f); -} - -ccl_device_inline float2 operator*(float f, const float2& a) -{ - return make_float2(a.x*f, a.y*f); -} - -ccl_device_inline float2 operator/(float f, const float2& a) -{ - return make_float2(f/a.x, f/a.y); -} - -ccl_device_inline float2 operator/(const float2& a, float f) -{ - float invf = 1.0f/f; - return make_float2(a.x*invf, a.y*invf); -} - -ccl_device_inline float2 operator/(const float2& a, const float2& b) -{ - return make_float2(a.x/b.x, a.y/b.y); -} - -ccl_device_inline float2 operator+(const float2& a, const float2& b) -{ - return make_float2(a.x+b.x, a.y+b.y); -} - -ccl_device_inline float2 operator-(const float2& a, const float2& b) -{ - return make_float2(a.x-b.x, a.y-b.y); -} - -ccl_device_inline float2 operator+=(float2& a, const float2& b) -{ - return a = a + b; -} - -ccl_device_inline float2 operator*=(float2& a, const float2& b) -{ - return a = a * b; -} - -ccl_device_inline float2 operator*=(float2& a, float f) -{ - return a = a * f; -} - -ccl_device_inline float2 operator/=(float2& a, const float2& b) -{ - return a = a / b; -} - -ccl_device_inline float2 operator/=(float2& a, float f) -{ - float invf = 1.0f/f; - return a = a * invf; -} - - -ccl_device_inline float dot(const float2& a, const float2& b) -{ - return a.x*b.x + a.y*b.y; -} - -ccl_device_inline float cross(const float2& a, const float2& b) -{ - return (a.x*b.y - a.y*b.x); -} - -#endif - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline bool operator==(const int2 a, const int2 b) -{ - return (a.x == b.x && a.y == b.y); -} - -ccl_device_inline float len(const float2& a) -{ - return sqrtf(dot(a, a)); -} - -ccl_device_inline float2 normalize(const float2& a) -{ - return a/len(a); -} - -ccl_device_inline float2 normalize_len(const float2& a, float *t) -{ - *t = len(a); - return a/(*t); -} - -ccl_device_inline float2 safe_normalize(const float2& a) -{ - float t = len(a); - return (t != 0.0f)? a/t: a; -} - -ccl_device_inline bool operator==(const float2& a, const float2& b) -{ - return (a.x == b.x && a.y == b.y); -} - -ccl_device_inline bool operator!=(const float2& a, const float2& b) -{ - return !(a == b); -} - -ccl_device_inline float2 min(const float2& a, const float2& b) -{ - return make_float2(min(a.x, b.x), min(a.y, b.y)); -} - -ccl_device_inline float2 max(const float2& a, const float2& b) -{ - return make_float2(max(a.x, b.x), max(a.y, b.y)); -} - -ccl_device_inline float2 clamp(const float2& a, const float2& mn, const float2& mx) -{ - return min(max(a, mn), mx); -} - -ccl_device_inline float2 fabs(const float2& a) -{ - return make_float2(fabsf(a.x), fabsf(a.y)); -} - -ccl_device_inline float2 as_float2(const float4& a) -{ - return make_float2(a.x, a.y); -} - -#endif - -#ifndef __KERNEL_GPU__ - -ccl_device_inline void print_float2(const char *label, const float2& a) -{ - printf("%s: %.8f %.8f\n", label, (double)a.x, (double)a.y); -} - -#endif - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline float2 interp(const float2& a, const float2& b, float t) -{ - return a + t*(b - a); -} - -#endif - -/* Float3 Vector */ - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline float3 operator-(const float3& a) -{ -#ifdef __KERNEL_SSE__ - return float3(_mm_xor_ps(a.m128, _mm_castsi128_ps(_mm_set1_epi32(0x80000000)))); -#else - return make_float3(-a.x, -a.y, -a.z); -#endif -} - -ccl_device_inline float3 operator*(const float3& a, const float3& b) -{ -#ifdef __KERNEL_SSE__ - return float3(_mm_mul_ps(a.m128,b.m128)); -#else - return make_float3(a.x*b.x, a.y*b.y, a.z*b.z); -#endif -} - -ccl_device_inline float3 operator*(const float3& a, const float f) -{ -#ifdef __KERNEL_SSE__ - return float3(_mm_mul_ps(a.m128,_mm_set1_ps(f))); -#else - return make_float3(a.x*f, a.y*f, a.z*f); -#endif -} - -ccl_device_inline float3 operator*(const float f, const float3& a) -{ - /* TODO(sergey): Currently disabled, gives speedup but causes precision issues. */ -#if defined(__KERNEL_SSE__) && 0 - return float3(_mm_mul_ps(_mm_set1_ps(f), a.m128)); -#else - return make_float3(a.x*f, a.y*f, a.z*f); -#endif -} - -ccl_device_inline float3 operator/(const float f, const float3& a) -{ - /* TODO(sergey): Currently disabled, gives speedup but causes precision issues. */ -#if defined(__KERNEL_SSE__) && 0 - __m128 rc = _mm_rcp_ps(a.m128); - return float3(_mm_mul_ps(_mm_set1_ps(f),rc)); -#else - return make_float3(f / a.x, f / a.y, f / a.z); -#endif -} - -ccl_device_inline float3 operator/(const float3& a, const float f) -{ - float invf = 1.0f/f; - return a * invf; -} - -ccl_device_inline float3 operator/(const float3& a, const float3& b) -{ - /* TODO(sergey): Currently disabled, gives speedup but causes precision issues. */ -#if defined(__KERNEL_SSE__) && 0 - __m128 rc = _mm_rcp_ps(b.m128); - return float3(_mm_mul_ps(a, rc)); -#else - return make_float3(a.x / b.x, a.y / b.y, a.z / b.z); -#endif -} - -ccl_device_inline float3 operator+(const float3& a, const float3& b) -{ -#ifdef __KERNEL_SSE__ - return float3(_mm_add_ps(a.m128, b.m128)); -#else - return make_float3(a.x + b.x, a.y + b.y, a.z + b.z); -#endif -} - -ccl_device_inline float3 operator-(const float3& a, const float3& b) -{ -#ifdef __KERNEL_SSE__ - return float3(_mm_sub_ps(a.m128, b.m128)); -#else - return make_float3(a.x - b.x, a.y - b.y, a.z - b.z); -#endif -} - -ccl_device_inline float3 operator+=(float3& a, const float3& b) -{ - return a = a + b; -} - -ccl_device_inline float3 operator*=(float3& a, const float3& b) -{ - return a = a * b; -} - -ccl_device_inline float3 operator*=(float3& a, float f) -{ - return a = a * f; -} - -ccl_device_inline float3 operator/=(float3& a, const float3& b) -{ - return a = a / b; -} - -ccl_device_inline float3 operator/=(float3& a, float f) -{ - float invf = 1.0f/f; - return a = a * invf; -} - -ccl_device_inline float dot(const float3& a, const float3& b) -{ -#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) - return _mm_cvtss_f32(_mm_dp_ps(a, b, 0x7F)); -#else - return a.x*b.x + a.y*b.y + a.z*b.z; -#endif -} - -ccl_device_inline float dot_xy(const float3& a, const float3& b) -{ -#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) - return _mm_cvtss_f32(_mm_hadd_ps(_mm_mul_ps(a,b),b)); -#else - return a.x*b.x + a.y*b.y; -#endif -} - -ccl_device_inline float dot(const float4& a, const float4& b) -{ -#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) - return _mm_cvtss_f32(_mm_dp_ps(a, b, 0xFF)); -#else - return (a.x*b.x + a.y*b.y) + (a.z*b.z + a.w*b.w); -#endif -} - -ccl_device_inline float3 cross(const float3& a, const float3& b) -{ - float3 r = make_float3(a.y*b.z - a.z*b.y, a.z*b.x - a.x*b.z, a.x*b.y - a.y*b.x); - return r; -} - -#endif - -ccl_device_inline float len(const float3 a) -{ -#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) - return _mm_cvtss_f32(_mm_sqrt_ss(_mm_dp_ps(a.m128, a.m128, 0x7F))); -#else - return sqrtf(dot(a, a)); -#endif -} - -ccl_device_inline float len_squared(const float3 a) -{ - return dot(a, a); -} - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline float len_squared(const float4& a) -{ - return dot(a, a); -} - -ccl_device_inline float3 normalize(const float3& a) -{ -#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) - __m128 norm = _mm_sqrt_ps(_mm_dp_ps(a.m128, a.m128, 0x7F)); - return float3(_mm_div_ps(a.m128, norm)); -#else - return a/len(a); -#endif -} - -#endif - -ccl_device_inline float3 saturate3(float3 a) -{ - return make_float3(saturate(a.x), saturate(a.y), saturate(a.z)); -} - -ccl_device_inline float3 normalize_len(const float3 a, float *t) -{ - *t = len(a); - float x = 1.0f / *t; - return a*x; -} - -ccl_device_inline float3 safe_normalize(const float3 a) -{ - float t = len(a); - return (t != 0.0f)? a * (1.0f/t) : a; -} - -ccl_device_inline float3 safe_normalize_len(const float3 a, float *t) -{ - *t = len(a); - return (*t != 0.0f)? a/(*t): a; -} - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline bool operator==(const float3& a, const float3& b) -{ -#ifdef __KERNEL_SSE__ - return (_mm_movemask_ps(_mm_cmpeq_ps(a.m128, b.m128)) & 7) == 7; -#else - return (a.x == b.x && a.y == b.y && a.z == b.z); -#endif -} - -ccl_device_inline bool operator!=(const float3& a, const float3& b) -{ - return !(a == b); -} - -ccl_device_inline float3 min(const float3& a, const float3& b) -{ -#ifdef __KERNEL_SSE__ - return float3(_mm_min_ps(a.m128, b.m128)); -#else - return make_float3(min(a.x, b.x), min(a.y, b.y), min(a.z, b.z)); -#endif -} - -ccl_device_inline float3 max(const float3& a, const float3& b) -{ -#ifdef __KERNEL_SSE__ - return float3(_mm_max_ps(a.m128, b.m128)); -#else - return make_float3(max(a.x, b.x), max(a.y, b.y), max(a.z, b.z)); -#endif -} - -ccl_device_inline float3 clamp(const float3& a, const float3& mn, const float3& mx) -{ - return min(max(a, mn), mx); -} - -ccl_device_inline float3 fabs(const float3& a) -{ -#ifdef __KERNEL_SSE__ - __m128 mask = _mm_castsi128_ps(_mm_set1_epi32(0x7fffffff)); - return float3(_mm_and_ps(a.m128, mask)); -#else - return make_float3(fabsf(a.x), fabsf(a.y), fabsf(a.z)); -#endif -} - -#endif - ccl_device_inline float3 float2_to_float3(const float2 a) { return make_float3(a.x, a.y, 0.0f); @@ -704,509 +237,21 @@ ccl_device_inline float4 float3_to_float4(const float3 a) return make_float4(a.x, a.y, a.z, 1.0f); } -#ifndef __KERNEL_GPU__ - -ccl_device_inline void print_float3(const char *label, const float3& a) -{ - printf("%s: %.8f %.8f %.8f\n", label, (double)a.x, (double)a.y, (double)a.z); -} - -ccl_device_inline float3 rcp(const float3& a) -{ -#ifdef __KERNEL_SSE__ - const float4 r(_mm_rcp_ps(a.m128)); - return float3(_mm_sub_ps(_mm_add_ps(r, r), - _mm_mul_ps(_mm_mul_ps(r, r), a))); -#else - return make_float3(1.0f/a.x, 1.0f/a.y, 1.0f/a.z); -#endif -} - -#endif - -ccl_device_inline float3 interp(float3 a, float3 b, float t) -{ - return a + t*(b - a); -} - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline float3 mix(const float3& a, const float3& b, float t) -{ - return a + t*(b - a); -} - -#endif - -ccl_device_inline bool is_zero(const float3 a) -{ -#ifdef __KERNEL_SSE__ - return a == make_float3(0.0f); -#else - return (a.x == 0.0f && a.y == 0.0f && a.z == 0.0f); -#endif -} - -ccl_device_inline float reduce_add(const float3 a) -{ - return (a.x + a.y + a.z); -} - -ccl_device_inline float average(const float3 a) -{ - return reduce_add(a)*(1.0f/3.0f); -} - -ccl_device_inline bool isequal_float3(const float3 a, const float3 b) -{ -#ifdef __KERNEL_OPENCL__ - return all(a == b); -#else - return a == b; -#endif -} - -/* Float4 Vector */ - -#ifdef __KERNEL_SSE__ - -template<size_t index_0, size_t index_1, size_t index_2, size_t index_3> -__forceinline const float4 shuffle(const float4& b) -{ - return float4(_mm_castsi128_ps( - _mm_shuffle_epi32(_mm_castps_si128(b), - _MM_SHUFFLE(index_3, index_2, index_1, index_0)))); -} - -#if defined(__KERNEL_SSE3__) -template<> __forceinline const float4 shuffle<0, 0, 2, 2>(const float4& b) -{ - return float4(_mm_moveldup_ps(b)); -} - -template<> __forceinline const float4 shuffle<1, 1, 3, 3>(const float4& b) -{ - return float4(_mm_movehdup_ps(b)); -} -#endif - -template<> __forceinline const float4 shuffle<0, 1, 0, 1>(const float4& b) -{ - return float4(_mm_castpd_ps(_mm_movedup_pd(_mm_castps_pd(b)))); -} - -#endif - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline float4 operator-(const float4& a) -{ -#ifdef __KERNEL_SSE__ - __m128 mask = _mm_castsi128_ps(_mm_set1_epi32(0x80000000)); - return float4(_mm_xor_ps(a.m128, mask)); -#else - return make_float4(-a.x, -a.y, -a.z, -a.w); -#endif -} - -ccl_device_inline float4 operator*(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - return float4(_mm_mul_ps(a.m128, b.m128)); -#else - return make_float4(a.x*b.x, a.y*b.y, a.z*b.z, a.w*b.w); -#endif -} - -ccl_device_inline float4 operator*(const float4& a, float f) -{ -#if defined(__KERNEL_SSE__) - return a * make_float4(f); -#else - return make_float4(a.x*f, a.y*f, a.z*f, a.w*f); -#endif -} - -ccl_device_inline float4 operator*(float f, const float4& a) -{ - return a * f; -} - -ccl_device_inline float4 rcp(const float4& a) -{ -#ifdef __KERNEL_SSE__ - float4 r(_mm_rcp_ps(a.m128)); - return float4(_mm_sub_ps(_mm_add_ps(r, r), - _mm_mul_ps(_mm_mul_ps(r, r), a))); -#else - return make_float4(1.0f/a.x, 1.0f/a.y, 1.0f/a.z, 1.0f/a.w); -#endif -} - -ccl_device_inline float4 operator/(const float4& a, float f) -{ - return a * (1.0f/f); -} - -ccl_device_inline float4 operator/(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - return a * rcp(b); -#else - return make_float4(a.x/b.x, a.y/b.y, a.z/b.z, a.w/b.w); -#endif - -} - -ccl_device_inline float4 operator+(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - return float4(_mm_add_ps(a.m128, b.m128)); -#else - return make_float4(a.x+b.x, a.y+b.y, a.z+b.z, a.w+b.w); -#endif -} - -ccl_device_inline float4 operator-(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - return float4(_mm_sub_ps(a.m128, b.m128)); -#else - return make_float4(a.x-b.x, a.y-b.y, a.z-b.z, a.w-b.w); -#endif -} - -ccl_device_inline float4 operator+=(float4& a, const float4& b) -{ - return a = a + b; -} - -ccl_device_inline float4 operator*=(float4& a, const float4& b) -{ - return a = a * b; -} - -ccl_device_inline float4 operator/=(float4& a, float f) -{ - return a = a / f; -} - -ccl_device_inline int4 operator<(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - /* TODO(sergey): avoid cvt. */ - return int4(_mm_cvtps_epi32(_mm_cmplt_ps(a.m128, b.m128))); -#else - return make_int4(a.x < b.x, a.y < b.y, a.z < b.z, a.w < b.w); -#endif -} - -ccl_device_inline int4 operator>=(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - /* TODO(sergey): avoid cvt. */ - return int4(_mm_cvtps_epi32(_mm_cmpge_ps(a.m128, b.m128))); -#else - return make_int4(a.x >= b.x, a.y >= b.y, a.z >= b.z, a.w >= b.w); -#endif -} - -ccl_device_inline int4 operator<=(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - /* TODO(sergey): avoid cvt. */ - return int4(_mm_cvtps_epi32(_mm_cmple_ps(a.m128, b.m128))); -#else - return make_int4(a.x <= b.x, a.y <= b.y, a.z <= b.z, a.w <= b.w); -#endif -} - -ccl_device_inline bool operator==(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - return (_mm_movemask_ps(_mm_cmpeq_ps(a.m128, b.m128)) & 15) == 15; -#else - return (a.x == b.x && a.y == b.y && a.z == b.z && a.w == b.w); -#endif -} - -ccl_device_inline float4 cross(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - return (shuffle<1,2,0,0>(a)*shuffle<2,0,1,0>(b)) - (shuffle<2,0,1,0>(a)*shuffle<1,2,0,0>(b)); -#else - return make_float4(a.y*b.z - a.z*b.y, a.z*b.x - a.x*b.z, a.x*b.y - a.y*b.x, 0.0f); -#endif -} - -ccl_device_inline bool is_zero(const float4& a) -{ -#ifdef __KERNEL_SSE__ - return a == make_float4(0.0f); -#else - return (a.x == 0.0f && a.y == 0.0f && a.z == 0.0f && a.w == 0.0f); -#endif -} - -ccl_device_inline float reduce_add(const float4& a) -{ -#ifdef __KERNEL_SSE__ - float4 h(shuffle<1,0,3,2>(a) + a); - /* TODO(sergey): Investigate efficiency. */ - return _mm_cvtss_f32(shuffle<2,3,0,1>(h) + h); -#else - return ((a.x + a.y) + (a.z + a.w)); -#endif -} - -ccl_device_inline float average(const float4& a) -{ - return reduce_add(a) * 0.25f; -} - -ccl_device_inline float len(const float4& a) -{ - return sqrtf(dot(a, a)); -} - -ccl_device_inline float4 normalize(const float4& a) -{ - return a/len(a); -} - -ccl_device_inline float4 safe_normalize(const float4& a) -{ - float t = len(a); - return (t != 0.0f)? a/t: a; -} - -ccl_device_inline float4 min(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - return float4(_mm_min_ps(a.m128, b.m128)); -#else - return make_float4(min(a.x, b.x), min(a.y, b.y), min(a.z, b.z), min(a.w, b.w)); -#endif -} - -ccl_device_inline float4 max(const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - return float4(_mm_max_ps(a.m128, b.m128)); -#else - return make_float4(max(a.x, b.x), max(a.y, b.y), max(a.z, b.z), max(a.w, b.w)); -#endif -} - -#endif - -#ifndef __KERNEL_GPU__ - -ccl_device_inline float4 select(const int4& mask, const float4& a, const float4& b) -{ -#ifdef __KERNEL_SSE__ - /* TODO(sergey): avoid cvt. */ - return float4(_mm_or_ps(_mm_and_ps(_mm_cvtepi32_ps(mask), a), - _mm_andnot_ps(_mm_cvtepi32_ps(mask), b))); -#else - return make_float4((mask.x)? a.x: b.x, (mask.y)? a.y: b.y, (mask.z)? a.z: b.z, (mask.w)? a.w: b.w); -#endif -} - -ccl_device_inline float4 reduce_min(const float4& a) -{ -#ifdef __KERNEL_SSE__ - float4 h = min(shuffle<1,0,3,2>(a), a); - return min(shuffle<2,3,0,1>(h), h); -#else - return make_float4(min(min(a.x, a.y), min(a.z, a.w))); -#endif -} - -ccl_device_inline float4 reduce_max(const float4& a) -{ -#ifdef __KERNEL_SSE__ - float4 h = max(shuffle<1,0,3,2>(a), a); - return max(shuffle<2,3,0,1>(h), h); -#else - return make_float4(max(max(a.x, a.y), max(a.z, a.w))); -#endif -} - -#if 0 -ccl_device_inline float4 reduce_add(const float4& a) -{ -#ifdef __KERNEL_SSE__ - float4 h = shuffle<1,0,3,2>(a) + a; - return shuffle<2,3,0,1>(h) + h; -#else - return make_float4((a.x + a.y) + (a.z + a.w)); -#endif -} -#endif - -ccl_device_inline void print_float4(const char *label, const float4& a) -{ - printf("%s: %.8f %.8f %.8f %.8f\n", label, (double)a.x, (double)a.y, (double)a.z, (double)a.w); -} - -#endif - -/* Int2 */ - -#ifndef __KERNEL_OPENCL__ - -ccl_device_inline int2 operator+(const int2 &a, const int2 &b) -{ - return make_int2(a.x + b.x, a.y + b.y); -} - -ccl_device_inline int2 operator+=(int2 &a, const int2 &b) -{ - return a = a + b; -} - -ccl_device_inline int2 operator-(const int2 &a, const int2 &b) -{ - return make_int2(a.x - b.x, a.y - b.y); -} - -ccl_device_inline int2 operator*(const int2 &a, const int2 &b) -{ - return make_int2(a.x * b.x, a.y * b.y); -} +CCL_NAMESPACE_END -ccl_device_inline int2 operator/(const int2 &a, const int2 &b) -{ - return make_int2(a.x / b.x, a.y / b.y); -} +#include "util/util_math_int2.h" +#include "util/util_math_int3.h" +#include "util/util_math_int4.h" -#endif +#include "util/util_math_float2.h" +#include "util/util_math_float3.h" +#include "util/util_math_float4.h" -/* Int3 */ +CCL_NAMESPACE_BEGIN #ifndef __KERNEL_OPENCL__ - -ccl_device_inline int3 min(int3 a, int3 b) -{ -#if defined(__KERNEL_SSE__) && defined(__KERNEL_SSE41__) - return int3(_mm_min_epi32(a.m128, b.m128)); -#else - return make_int3(min(a.x, b.x), min(a.y, b.y), min(a.z, b.z)); -#endif -} - -ccl_device_inline int3 max(int3 a, int3 b) -{ -#if defined(__KERNEL_SSE__) && defined(__KERNEL_SSE41__) - return int3(_mm_max_epi32(a.m128, b.m128)); -#else - return make_int3(max(a.x, b.x), max(a.y, b.y), max(a.z, b.z)); -#endif -} - -ccl_device_inline int3 clamp(const int3& a, int mn, int mx) -{ -#ifdef __KERNEL_SSE__ - return min(max(a, make_int3(mn)), make_int3(mx)); -#else - return make_int3(clamp(a.x, mn, mx), clamp(a.y, mn, mx), clamp(a.z, mn, mx)); -#endif -} - -ccl_device_inline int3 clamp(const int3& a, int3& mn, int mx) -{ -#ifdef __KERNEL_SSE__ - return min(max(a, mn), make_int3(mx)); -#else - return make_int3(clamp(a.x, mn.x, mx), clamp(a.y, mn.y, mx), clamp(a.z, mn.z, mx)); -#endif -} - -#endif - -#ifndef __KERNEL_GPU__ - -ccl_device_inline void print_int3(const char *label, const int3& a) -{ - printf("%s: %d %d %d\n", label, a.x, a.y, a.z); -} - -#endif - -/* Int4 */ - -#ifndef __KERNEL_GPU__ - -ccl_device_inline int4 operator+(const int4& a, const int4& b) -{ -#ifdef __KERNEL_SSE__ - return int4(_mm_add_epi32(a.m128, b.m128)); -#else - return make_int4(a.x+b.x, a.y+b.y, a.z+b.z, a.w+b.w); -#endif -} - -ccl_device_inline int4 operator+=(int4& a, const int4& b) -{ - return a = a + b; -} - -ccl_device_inline int4 operator>>(const int4& a, int i) -{ -#ifdef __KERNEL_SSE__ - return int4(_mm_srai_epi32(a.m128, i)); -#else - return make_int4(a.x >> i, a.y >> i, a.z >> i, a.w >> i); -#endif -} - -ccl_device_inline int4 min(int4 a, int4 b) -{ -#if defined(__KERNEL_SSE__) && defined(__KERNEL_SSE41__) - return int4(_mm_min_epi32(a.m128, b.m128)); -#else - return make_int4(min(a.x, b.x), min(a.y, b.y), min(a.z, b.z), min(a.w, b.w)); -#endif -} - -ccl_device_inline int4 max(int4 a, int4 b) -{ -#if defined(__KERNEL_SSE__) && defined(__KERNEL_SSE41__) - return int4(_mm_max_epi32(a.m128, b.m128)); -#else - return make_int4(max(a.x, b.x), max(a.y, b.y), max(a.z, b.z), max(a.w, b.w)); -#endif -} - -ccl_device_inline int4 clamp(const int4& a, const int4& mn, const int4& mx) -{ - return min(max(a, mn), mx); -} - -ccl_device_inline int4 select(const int4& mask, const int4& a, const int4& b) -{ -#ifdef __KERNEL_SSE__ - const __m128 m = _mm_cvtepi32_ps(mask); - /* TODO(sergey): avoid cvt. */ - return int4(_mm_castps_si128(_mm_or_ps(_mm_and_ps(m, _mm_castsi128_ps(a)), - _mm_andnot_ps(m, _mm_castsi128_ps(b))))); -#else - return make_int4((mask.x)? a.x: b.x, (mask.y)? a.y: b.y, (mask.z)? a.z: b.z, (mask.w)? a.w: b.w); -#endif -} - -ccl_device_inline void print_int4(const char *label, const int4& a) -{ - printf("%s: %d %d %d %d\n", label, a.x, a.y, a.z, a.w); -} - -#endif - /* Int/Float conversion */ -#ifndef __KERNEL_OPENCL__ - ccl_device_inline int as_int(uint i) { union { uint ui; int i; } u; @@ -1256,7 +301,6 @@ ccl_device_inline float __uint_as_float(uint i) return u.f; } - /* Interpolation */ template<class A, class B> A lerp(const A& a, const A& b, const B& t) @@ -1266,12 +310,13 @@ template<class A, class B> A lerp(const A& a, const A& b, const B& t) /* Triangle */ -ccl_device_inline float triangle_area(const float3& v1, const float3& v2, const float3& v3) +ccl_device_inline float triangle_area(const float3& v1, + const float3& v2, + const float3& v3) { return len(cross(v3 - v2, v1 - v2))*0.5f; } - -#endif +#endif /* __KERNEL_OPENCL__ */ /* Versions of functions which are safe for fast math. */ ccl_device_inline bool isnan_safe(float f) @@ -1382,16 +427,16 @@ ccl_device_inline float3 rotate_around_axis(float3 p, float3 axis, float angle) float3 r; r.x = ((costheta + (1 - costheta) * axis.x * axis.x) * p.x) + - (((1 - costheta) * axis.x * axis.y - axis.z * sintheta) * p.y) + - (((1 - costheta) * axis.x * axis.z + axis.y * sintheta) * p.z); + (((1 - costheta) * axis.x * axis.y - axis.z * sintheta) * p.y) + + (((1 - costheta) * axis.x * axis.z + axis.y * sintheta) * p.z); r.y = (((1 - costheta) * axis.x * axis.y + axis.z * sintheta) * p.x) + - ((costheta + (1 - costheta) * axis.y * axis.y) * p.y) + - (((1 - costheta) * axis.y * axis.z - axis.x * sintheta) * p.z); + ((costheta + (1 - costheta) * axis.y * axis.y) * p.y) + + (((1 - costheta) * axis.y * axis.z - axis.x * sintheta) * p.z); r.z = (((1 - costheta) * axis.x * axis.z - axis.y * sintheta) * p.x) + - (((1 - costheta) * axis.y * axis.z + axis.x * sintheta) * p.y) + - ((costheta + (1 - costheta) * axis.z * axis.z) * p.z); + (((1 - costheta) * axis.y * axis.z + axis.x * sintheta) * p.y) + + ((costheta + (1 - costheta) * axis.z * axis.z) * p.z); return r; } diff --git a/intern/cycles/util/util_math_float2.h b/intern/cycles/util/util_math_float2.h new file mode 100644 index 00000000000..6f9d0855d50 --- /dev/null +++ b/intern/cycles/util/util_math_float2.h @@ -0,0 +1,227 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_MATH_FLOAT2_H__ +#define __UTIL_MATH_FLOAT2_H__ + +#ifndef __UTIL_MATH_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +/******************************************************************************* + * Declaration. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline float2 operator-(const float2& a); +ccl_device_inline float2 operator*(const float2& a, const float2& b); +ccl_device_inline float2 operator*(const float2& a, float f); +ccl_device_inline float2 operator*(float f, const float2& a); +ccl_device_inline float2 operator/(float f, const float2& a); +ccl_device_inline float2 operator/(const float2& a, float f); +ccl_device_inline float2 operator/(const float2& a, const float2& b); +ccl_device_inline float2 operator+(const float2& a, const float2& b); +ccl_device_inline float2 operator-(const float2& a, const float2& b); +ccl_device_inline float2 operator+=(float2& a, const float2& b); +ccl_device_inline float2 operator*=(float2& a, const float2& b); +ccl_device_inline float2 operator*=(float2& a, float f); +ccl_device_inline float2 operator/=(float2& a, const float2& b); +ccl_device_inline float2 operator/=(float2& a, float f); + +ccl_device_inline bool operator==(const float2& a, const float2& b); +ccl_device_inline bool operator!=(const float2& a, const float2& b); + +ccl_device_inline bool is_zero(const float2& a); +ccl_device_inline float average(const float2& a); +ccl_device_inline float dot(const float2& a, const float2& b); +ccl_device_inline float cross(const float2& a, const float2& b); +ccl_device_inline float len(const float2& a); +ccl_device_inline float2 normalize(const float2& a); +ccl_device_inline float2 normalize_len(const float2& a, float *t); +ccl_device_inline float2 safe_normalize(const float2& a); +ccl_device_inline float2 min(const float2& a, const float2& b); +ccl_device_inline float2 max(const float2& a, const float2& b); +ccl_device_inline float2 clamp(const float2& a, const float2& mn, const float2& mx); +ccl_device_inline float2 fabs(const float2& a); +ccl_device_inline float2 as_float2(const float4& a); +ccl_device_inline float2 interp(const float2& a, const float2& b, float t); +#endif /* !__KERNEL_OPENCL__ */ + +/******************************************************************************* + * Definition. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline float2 operator-(const float2& a) +{ + return make_float2(-a.x, -a.y); +} + +ccl_device_inline float2 operator*(const float2& a, const float2& b) +{ + return make_float2(a.x*b.x, a.y*b.y); +} + +ccl_device_inline float2 operator*(const float2& a, float f) +{ + return make_float2(a.x*f, a.y*f); +} + +ccl_device_inline float2 operator*(float f, const float2& a) +{ + return make_float2(a.x*f, a.y*f); +} + +ccl_device_inline float2 operator/(float f, const float2& a) +{ + return make_float2(f/a.x, f/a.y); +} + +ccl_device_inline float2 operator/(const float2& a, float f) +{ + float invf = 1.0f/f; + return make_float2(a.x*invf, a.y*invf); +} + +ccl_device_inline float2 operator/(const float2& a, const float2& b) +{ + return make_float2(a.x/b.x, a.y/b.y); +} + +ccl_device_inline float2 operator+(const float2& a, const float2& b) +{ + return make_float2(a.x+b.x, a.y+b.y); +} + +ccl_device_inline float2 operator-(const float2& a, const float2& b) +{ + return make_float2(a.x-b.x, a.y-b.y); +} + +ccl_device_inline float2 operator+=(float2& a, const float2& b) +{ + return a = a + b; +} + +ccl_device_inline float2 operator*=(float2& a, const float2& b) +{ + return a = a * b; +} + +ccl_device_inline float2 operator*=(float2& a, float f) +{ + return a = a * f; +} + +ccl_device_inline float2 operator/=(float2& a, const float2& b) +{ + return a = a / b; +} + +ccl_device_inline float2 operator/=(float2& a, float f) +{ + float invf = 1.0f/f; + return a = a * invf; +} + +ccl_device_inline bool operator==(const float2& a, const float2& b) +{ + return (a.x == b.x && a.y == b.y); +} + +ccl_device_inline bool operator!=(const float2& a, const float2& b) +{ + return !(a == b); +} + +ccl_device_inline bool is_zero(const float2& a) +{ + return (a.x == 0.0f && a.y == 0.0f); +} + +ccl_device_inline float average(const float2& a) +{ + return (a.x + a.y)*(1.0f/2.0f); +} + +ccl_device_inline float dot(const float2& a, const float2& b) +{ + return a.x*b.x + a.y*b.y; +} + +ccl_device_inline float cross(const float2& a, const float2& b) +{ + return (a.x*b.y - a.y*b.x); +} + +ccl_device_inline float len(const float2& a) +{ + return sqrtf(dot(a, a)); +} + +ccl_device_inline float2 normalize(const float2& a) +{ + return a/len(a); +} + +ccl_device_inline float2 normalize_len(const float2& a, float *t) +{ + *t = len(a); + return a/(*t); +} + +ccl_device_inline float2 safe_normalize(const float2& a) +{ + float t = len(a); + return (t != 0.0f)? a/t: a; +} + +ccl_device_inline float2 min(const float2& a, const float2& b) +{ + return make_float2(min(a.x, b.x), min(a.y, b.y)); +} + +ccl_device_inline float2 max(const float2& a, const float2& b) +{ + return make_float2(max(a.x, b.x), max(a.y, b.y)); +} + +ccl_device_inline float2 clamp(const float2& a, const float2& mn, const float2& mx) +{ + return min(max(a, mn), mx); +} + +ccl_device_inline float2 fabs(const float2& a) +{ + return make_float2(fabsf(a.x), fabsf(a.y)); +} + +ccl_device_inline float2 as_float2(const float4& a) +{ + return make_float2(a.x, a.y); +} + +ccl_device_inline float2 interp(const float2& a, const float2& b, float t) +{ + return a + t*(b - a); +} +#endif /* !__KERNEL_OPENCL__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_MATH_FLOAT2_H__ */ diff --git a/intern/cycles/util/util_math_float3.h b/intern/cycles/util/util_math_float3.h new file mode 100644 index 00000000000..e0c6b551040 --- /dev/null +++ b/intern/cycles/util/util_math_float3.h @@ -0,0 +1,365 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_MATH_FLOAT3_H__ +#define __UTIL_MATH_FLOAT3_H__ + +#ifndef __UTIL_MATH_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +/******************************************************************************* + * Declaration. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline float3 operator-(const float3& a); +ccl_device_inline float3 operator*(const float3& a, const float3& b); +ccl_device_inline float3 operator*(const float3& a, const float f); +ccl_device_inline float3 operator*(const float f, const float3& a); +ccl_device_inline float3 operator/(const float f, const float3& a); +ccl_device_inline float3 operator/(const float3& a, const float f); +ccl_device_inline float3 operator/(const float3& a, const float3& b); +ccl_device_inline float3 operator+(const float3& a, const float3& b); +ccl_device_inline float3 operator-(const float3& a, const float3& b); +ccl_device_inline float3 operator+=(float3& a, const float3& b); +ccl_device_inline float3 operator*=(float3& a, const float3& b); +ccl_device_inline float3 operator*=(float3& a, float f); +ccl_device_inline float3 operator/=(float3& a, const float3& b); +ccl_device_inline float3 operator/=(float3& a, float f); + +ccl_device_inline bool operator==(const float3& a, const float3& b); +ccl_device_inline bool operator!=(const float3& a, const float3& b); + +ccl_device_inline float dot(const float3& a, const float3& b); +ccl_device_inline float dot_xy(const float3& a, const float3& b); +ccl_device_inline float3 cross(const float3& a, const float3& b); +ccl_device_inline float3 normalize(const float3& a); +ccl_device_inline float3 min(const float3& a, const float3& b); +ccl_device_inline float3 max(const float3& a, const float3& b); +ccl_device_inline float3 clamp(const float3& a, const float3& mn, const float3& mx); +ccl_device_inline float3 fabs(const float3& a); +ccl_device_inline float3 mix(const float3& a, const float3& b, float t); +ccl_device_inline float3 rcp(const float3& a); +#endif /* !__KERNEL_OPENCL__ */ + +ccl_device_inline float max3(float3 a); +ccl_device_inline float len(const float3 a); +ccl_device_inline float len_squared(const float3 a); + +ccl_device_inline float3 saturate3(float3 a); +ccl_device_inline float3 safe_normalize(const float3 a); +ccl_device_inline float3 normalize_len(const float3 a, float *t);; +ccl_device_inline float3 safe_normalize_len(const float3 a, float *t); +ccl_device_inline float3 interp(float3 a, float3 b, float t); + +ccl_device_inline bool is_zero(const float3 a); +ccl_device_inline float reduce_add(const float3 a); +ccl_device_inline float average(const float3 a); +ccl_device_inline bool isequal_float3(const float3 a, const float3 b); + +/******************************************************************************* + * Definition. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline float3 operator-(const float3& a) +{ +#ifdef __KERNEL_SSE__ + return float3(_mm_xor_ps(a.m128, _mm_castsi128_ps(_mm_set1_epi32(0x80000000)))); +#else + return make_float3(-a.x, -a.y, -a.z); +#endif +} + +ccl_device_inline float3 operator*(const float3& a, const float3& b) +{ +#ifdef __KERNEL_SSE__ + return float3(_mm_mul_ps(a.m128,b.m128)); +#else + return make_float3(a.x*b.x, a.y*b.y, a.z*b.z); +#endif +} + +ccl_device_inline float3 operator*(const float3& a, const float f) +{ +#ifdef __KERNEL_SSE__ + return float3(_mm_mul_ps(a.m128,_mm_set1_ps(f))); +#else + return make_float3(a.x*f, a.y*f, a.z*f); +#endif +} + +ccl_device_inline float3 operator*(const float f, const float3& a) +{ + /* TODO(sergey): Currently disabled, gives speedup but causes precision issues. */ +#if defined(__KERNEL_SSE__) && 0 + return float3(_mm_mul_ps(_mm_set1_ps(f), a.m128)); +#else + return make_float3(a.x*f, a.y*f, a.z*f); +#endif +} + +ccl_device_inline float3 operator/(const float f, const float3& a) +{ + /* TODO(sergey): Currently disabled, gives speedup but causes precision issues. */ +#if defined(__KERNEL_SSE__) && 0 + __m128 rc = _mm_rcp_ps(a.m128); + return float3(_mm_mul_ps(_mm_set1_ps(f),rc)); +#else + return make_float3(f / a.x, f / a.y, f / a.z); +#endif +} + +ccl_device_inline float3 operator/(const float3& a, const float f) +{ + float invf = 1.0f/f; + return a * invf; +} + +ccl_device_inline float3 operator/(const float3& a, const float3& b) +{ + /* TODO(sergey): Currently disabled, gives speedup but causes precision issues. */ +#if defined(__KERNEL_SSE__) && 0 + __m128 rc = _mm_rcp_ps(b.m128); + return float3(_mm_mul_ps(a, rc)); +#else + return make_float3(a.x / b.x, a.y / b.y, a.z / b.z); +#endif +} + +ccl_device_inline float3 operator+(const float3& a, const float3& b) +{ +#ifdef __KERNEL_SSE__ + return float3(_mm_add_ps(a.m128, b.m128)); +#else + return make_float3(a.x + b.x, a.y + b.y, a.z + b.z); +#endif +} + +ccl_device_inline float3 operator-(const float3& a, const float3& b) +{ +#ifdef __KERNEL_SSE__ + return float3(_mm_sub_ps(a.m128, b.m128)); +#else + return make_float3(a.x - b.x, a.y - b.y, a.z - b.z); +#endif +} + +ccl_device_inline float3 operator+=(float3& a, const float3& b) +{ + return a = a + b; +} + +ccl_device_inline float3 operator*=(float3& a, const float3& b) +{ + return a = a * b; +} + +ccl_device_inline float3 operator*=(float3& a, float f) +{ + return a = a * f; +} + +ccl_device_inline float3 operator/=(float3& a, const float3& b) +{ + return a = a / b; +} + +ccl_device_inline float3 operator/=(float3& a, float f) +{ + float invf = 1.0f/f; + return a = a * invf; +} + +ccl_device_inline bool operator==(const float3& a, const float3& b) +{ +#ifdef __KERNEL_SSE__ + return (_mm_movemask_ps(_mm_cmpeq_ps(a.m128, b.m128)) & 7) == 7; +#else + return (a.x == b.x && a.y == b.y && a.z == b.z); +#endif +} + +ccl_device_inline bool operator!=(const float3& a, const float3& b) +{ + return !(a == b); +} + +ccl_device_inline float dot(const float3& a, const float3& b) +{ +#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) + return _mm_cvtss_f32(_mm_dp_ps(a, b, 0x7F)); +#else + return a.x*b.x + a.y*b.y + a.z*b.z; +#endif +} + +ccl_device_inline float dot_xy(const float3& a, const float3& b) +{ +#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) + return _mm_cvtss_f32(_mm_hadd_ps(_mm_mul_ps(a,b),b)); +#else + return a.x*b.x + a.y*b.y; +#endif +} + +ccl_device_inline float3 cross(const float3& a, const float3& b) +{ + float3 r = make_float3(a.y*b.z - a.z*b.y, a.z*b.x - a.x*b.z, a.x*b.y - a.y*b.x); + return r; +} + +ccl_device_inline float3 normalize(const float3& a) +{ +#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) + __m128 norm = _mm_sqrt_ps(_mm_dp_ps(a.m128, a.m128, 0x7F)); + return float3(_mm_div_ps(a.m128, norm)); +#else + return a/len(a); +#endif +} + +ccl_device_inline float3 min(const float3& a, const float3& b) +{ +#ifdef __KERNEL_SSE__ + return float3(_mm_min_ps(a.m128, b.m128)); +#else + return make_float3(min(a.x, b.x), min(a.y, b.y), min(a.z, b.z)); +#endif +} + +ccl_device_inline float3 max(const float3& a, const float3& b) +{ +#ifdef __KERNEL_SSE__ + return float3(_mm_max_ps(a.m128, b.m128)); +#else + return make_float3(max(a.x, b.x), max(a.y, b.y), max(a.z, b.z)); +#endif +} + +ccl_device_inline float3 clamp(const float3& a, const float3& mn, const float3& mx) +{ + return min(max(a, mn), mx); +} + +ccl_device_inline float3 fabs(const float3& a) +{ +#ifdef __KERNEL_SSE__ + __m128 mask = _mm_castsi128_ps(_mm_set1_epi32(0x7fffffff)); + return float3(_mm_and_ps(a.m128, mask)); +#else + return make_float3(fabsf(a.x), fabsf(a.y), fabsf(a.z)); +#endif +} + +ccl_device_inline float3 mix(const float3& a, const float3& b, float t) +{ + return a + t*(b - a); +} + +ccl_device_inline float3 rcp(const float3& a) +{ +#ifdef __KERNEL_SSE__ + const float4 r(_mm_rcp_ps(a.m128)); + return float3(_mm_sub_ps(_mm_add_ps(r, r), + _mm_mul_ps(_mm_mul_ps(r, r), a))); +#else + return make_float3(1.0f/a.x, 1.0f/a.y, 1.0f/a.z); +#endif +} +#endif /* !__KERNEL_OPENCL__ */ + +ccl_device_inline float max3(float3 a) +{ + return max(max(a.x, a.y), a.z); +} + +ccl_device_inline float len(const float3 a) +{ +#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) + return _mm_cvtss_f32(_mm_sqrt_ss(_mm_dp_ps(a.m128, a.m128, 0x7F))); +#else + return sqrtf(dot(a, a)); +#endif +} + +ccl_device_inline float len_squared(const float3 a) +{ + return dot(a, a); +} + +ccl_device_inline float3 saturate3(float3 a) +{ + return make_float3(saturate(a.x), saturate(a.y), saturate(a.z)); +} + +ccl_device_inline float3 normalize_len(const float3 a, float *t) +{ + *t = len(a); + float x = 1.0f / *t; + return a*x; +} + +ccl_device_inline float3 safe_normalize(const float3 a) +{ + float t = len(a); + return (t != 0.0f)? a * (1.0f/t) : a; +} + +ccl_device_inline float3 safe_normalize_len(const float3 a, float *t) +{ + *t = len(a); + return (*t != 0.0f)? a/(*t): a; +} + +ccl_device_inline float3 interp(float3 a, float3 b, float t) +{ + return a + t*(b - a); +} + +ccl_device_inline bool is_zero(const float3 a) +{ +#ifdef __KERNEL_SSE__ + return a == make_float3(0.0f); +#else + return (a.x == 0.0f && a.y == 0.0f && a.z == 0.0f); +#endif +} + +ccl_device_inline float reduce_add(const float3 a) +{ + return (a.x + a.y + a.z); +} + +ccl_device_inline float average(const float3 a) +{ + return reduce_add(a)*(1.0f/3.0f); +} + +ccl_device_inline bool isequal_float3(const float3 a, const float3 b) +{ +#ifdef __KERNEL_OPENCL__ + return all(a == b); +#else + return a == b; +#endif +} +CCL_NAMESPACE_END + +#endif /* __UTIL_MATH_FLOAT3_H__ */ diff --git a/intern/cycles/util/util_math_float4.h b/intern/cycles/util/util_math_float4.h new file mode 100644 index 00000000000..d89121b3a1d --- /dev/null +++ b/intern/cycles/util/util_math_float4.h @@ -0,0 +1,393 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_MATH_FLOAT4_H__ +#define __UTIL_MATH_FLOAT4_H__ + +#ifndef __UTIL_MATH_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +/******************************************************************************* + * Declaration. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline float4 operator-(const float4& a); +ccl_device_inline float4 operator*(const float4& a, const float4& b); +ccl_device_inline float4 operator*(const float4& a, float f); +ccl_device_inline float4 operator*(float f, const float4& a); +ccl_device_inline float4 operator/(const float4& a, float f); +ccl_device_inline float4 operator/(const float4& a, const float4& b); +ccl_device_inline float4 operator+(const float4& a, const float4& b); +ccl_device_inline float4 operator-(const float4& a, const float4& b); +ccl_device_inline float4 operator+=(float4& a, const float4& b); +ccl_device_inline float4 operator*=(float4& a, const float4& b); +ccl_device_inline float4 operator/=(float4& a, float f); + +ccl_device_inline int4 operator<(const float4& a, const float4& b); +ccl_device_inline int4 operator>=(const float4& a, const float4& b); +ccl_device_inline int4 operator<=(const float4& a, const float4& b); +ccl_device_inline bool operator==(const float4& a, const float4& b); + +ccl_device_inline float dot(const float4& a, const float4& b); +ccl_device_inline float len_squared(const float4& a); +ccl_device_inline float4 rcp(const float4& a); +ccl_device_inline float4 cross(const float4& a, const float4& b); +ccl_device_inline bool is_zero(const float4& a); +ccl_device_inline float reduce_add(const float4& a); +ccl_device_inline float average(const float4& a); +ccl_device_inline float len(const float4& a); +ccl_device_inline float4 normalize(const float4& a); +ccl_device_inline float4 safe_normalize(const float4& a); +ccl_device_inline float4 min(const float4& a, const float4& b); +ccl_device_inline float4 max(const float4& a, const float4& b); +#endif /* !__KERNEL_OPENCL__*/ + +#ifdef __KERNEL_SSE__ +template<size_t index_0, size_t index_1, size_t index_2, size_t index_3> +__forceinline const float4 shuffle(const float4& b); + +template<> __forceinline const float4 shuffle<0, 1, 0, 1>(const float4& b); + +# ifdef __KERNEL_SSE3__ +template<> __forceinline const float4 shuffle<0, 0, 2, 2>(const float4& b); +template<> __forceinline const float4 shuffle<1, 1, 3, 3>(const float4& b); +# endif +#endif /* __KERNEL_SSE__ */ + +#ifndef __KERNEL_GPU__ +ccl_device_inline float4 select(const int4& mask, + const float4& a, + const float4& b); +ccl_device_inline float4 reduce_min(const float4& a); +ccl_device_inline float4 reduce_max(const float4& a); +# if 0 +ccl_device_inline float4 reduce_add(const float4& a); +# endif +#endif /* !__KERNEL_GPU__ */ + +/******************************************************************************* + * Definition. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline float4 operator-(const float4& a) +{ +#ifdef __KERNEL_SSE__ + __m128 mask = _mm_castsi128_ps(_mm_set1_epi32(0x80000000)); + return float4(_mm_xor_ps(a.m128, mask)); +#else + return make_float4(-a.x, -a.y, -a.z, -a.w); +#endif +} + +ccl_device_inline float4 operator*(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + return float4(_mm_mul_ps(a.m128, b.m128)); +#else + return make_float4(a.x*b.x, a.y*b.y, a.z*b.z, a.w*b.w); +#endif +} + +ccl_device_inline float4 operator*(const float4& a, float f) +{ +#if defined(__KERNEL_SSE__) + return a * make_float4(f); +#else + return make_float4(a.x*f, a.y*f, a.z*f, a.w*f); +#endif +} + +ccl_device_inline float4 operator*(float f, const float4& a) +{ + return a * f; +} + +ccl_device_inline float4 operator/(const float4& a, float f) +{ + return a * (1.0f/f); +} + +ccl_device_inline float4 operator/(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + return a * rcp(b); +#else + return make_float4(a.x/b.x, a.y/b.y, a.z/b.z, a.w/b.w); +#endif + +} + +ccl_device_inline float4 operator+(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + return float4(_mm_add_ps(a.m128, b.m128)); +#else + return make_float4(a.x+b.x, a.y+b.y, a.z+b.z, a.w+b.w); +#endif +} + +ccl_device_inline float4 operator-(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + return float4(_mm_sub_ps(a.m128, b.m128)); +#else + return make_float4(a.x-b.x, a.y-b.y, a.z-b.z, a.w-b.w); +#endif +} + +ccl_device_inline float4 operator+=(float4& a, const float4& b) +{ + return a = a + b; +} + +ccl_device_inline float4 operator*=(float4& a, const float4& b) +{ + return a = a * b; +} + +ccl_device_inline float4 operator/=(float4& a, float f) +{ + return a = a / f; +} + +ccl_device_inline int4 operator<(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + /* TODO(sergey): avoid cvt. */ + return int4(_mm_cvtps_epi32(_mm_cmplt_ps(a.m128, b.m128))); +#else + return make_int4(a.x < b.x, a.y < b.y, a.z < b.z, a.w < b.w); +#endif +} + +ccl_device_inline int4 operator>=(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + /* TODO(sergey): avoid cvt. */ + return int4(_mm_cvtps_epi32(_mm_cmpge_ps(a.m128, b.m128))); +#else + return make_int4(a.x >= b.x, a.y >= b.y, a.z >= b.z, a.w >= b.w); +#endif +} + +ccl_device_inline int4 operator<=(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + /* TODO(sergey): avoid cvt. */ + return int4(_mm_cvtps_epi32(_mm_cmple_ps(a.m128, b.m128))); +#else + return make_int4(a.x <= b.x, a.y <= b.y, a.z <= b.z, a.w <= b.w); +#endif +} + +ccl_device_inline bool operator==(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + return (_mm_movemask_ps(_mm_cmpeq_ps(a.m128, b.m128)) & 15) == 15; +#else + return (a.x == b.x && a.y == b.y && a.z == b.z && a.w == b.w); +#endif +} + +ccl_device_inline float dot(const float4& a, const float4& b) +{ +#if defined(__KERNEL_SSE41__) && defined(__KERNEL_SSE__) + return _mm_cvtss_f32(_mm_dp_ps(a, b, 0xFF)); +#else + return (a.x*b.x + a.y*b.y) + (a.z*b.z + a.w*b.w); +#endif +} + +ccl_device_inline float len_squared(const float4& a) +{ + return dot(a, a); +} + +ccl_device_inline float4 rcp(const float4& a) +{ +#ifdef __KERNEL_SSE__ + float4 r(_mm_rcp_ps(a.m128)); + return float4(_mm_sub_ps(_mm_add_ps(r, r), + _mm_mul_ps(_mm_mul_ps(r, r), a))); +#else + return make_float4(1.0f/a.x, 1.0f/a.y, 1.0f/a.z, 1.0f/a.w); +#endif +} + +ccl_device_inline float4 cross(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + return (shuffle<1,2,0,0>(a)*shuffle<2,0,1,0>(b)) - + (shuffle<2,0,1,0>(a)*shuffle<1,2,0,0>(b)); +#else + return make_float4(a.y*b.z - a.z*b.y, + a.z*b.x - a.x*b.z, + a.x*b.y - a.y*b.x, + 0.0f); +#endif +} + +ccl_device_inline bool is_zero(const float4& a) +{ +#ifdef __KERNEL_SSE__ + return a == make_float4(0.0f); +#else + return (a.x == 0.0f && a.y == 0.0f && a.z == 0.0f && a.w == 0.0f); +#endif +} + +ccl_device_inline float reduce_add(const float4& a) +{ +#ifdef __KERNEL_SSE__ + float4 h(shuffle<1,0,3,2>(a) + a); + /* TODO(sergey): Investigate efficiency. */ + return _mm_cvtss_f32(shuffle<2,3,0,1>(h) + h); +#else + return ((a.x + a.y) + (a.z + a.w)); +#endif +} + +ccl_device_inline float average(const float4& a) +{ + return reduce_add(a) * 0.25f; +} + +ccl_device_inline float len(const float4& a) +{ + return sqrtf(dot(a, a)); +} + +ccl_device_inline float4 normalize(const float4& a) +{ + return a/len(a); +} + +ccl_device_inline float4 safe_normalize(const float4& a) +{ + float t = len(a); + return (t != 0.0f)? a/t: a; +} + +ccl_device_inline float4 min(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + return float4(_mm_min_ps(a.m128, b.m128)); +#else + return make_float4(min(a.x, b.x), + min(a.y, b.y), + min(a.z, b.z), + min(a.w, b.w)); +#endif +} + +ccl_device_inline float4 max(const float4& a, const float4& b) +{ +#ifdef __KERNEL_SSE__ + return float4(_mm_max_ps(a.m128, b.m128)); +#else + return make_float4(max(a.x, b.x), + max(a.y, b.y), + max(a.z, b.z), + max(a.w, b.w)); +#endif +} +#endif /* !__KERNEL_OPENCL__*/ + +#ifdef __KERNEL_SSE__ +template<size_t index_0, size_t index_1, size_t index_2, size_t index_3> +__forceinline const float4 shuffle(const float4& b) +{ + return float4(_mm_castsi128_ps( + _mm_shuffle_epi32(_mm_castps_si128(b), + _MM_SHUFFLE(index_3, index_2, index_1, index_0)))); +} + +template<> __forceinline const float4 shuffle<0, 1, 0, 1>(const float4& b) +{ + return float4(_mm_castpd_ps(_mm_movedup_pd(_mm_castps_pd(b)))); +} + +# ifdef __KERNEL_SSE3__ +template<> __forceinline const float4 shuffle<0, 0, 2, 2>(const float4& b) +{ + return float4(_mm_moveldup_ps(b)); +} + +template<> __forceinline const float4 shuffle<1, 1, 3, 3>(const float4& b) +{ + return float4(_mm_movehdup_ps(b)); +} +# endif /* __KERNEL_SSE3__ */ +#endif /* __KERNEL_SSE__ */ + +#ifndef __KERNEL_GPU__ +ccl_device_inline float4 select(const int4& mask, + const float4& a, + const float4& b) +{ +#ifdef __KERNEL_SSE__ + /* TODO(sergey): avoid cvt. */ + return float4(_mm_or_ps(_mm_and_ps(_mm_cvtepi32_ps(mask), a), + _mm_andnot_ps(_mm_cvtepi32_ps(mask), b))); +#else + return make_float4((mask.x)? a.x: b.x, + (mask.y)? a.y: b.y, + (mask.z)? a.z: b.z, + (mask.w)? a.w: b.w); +#endif +} + +ccl_device_inline float4 reduce_min(const float4& a) +{ +#ifdef __KERNEL_SSE__ + float4 h = min(shuffle<1,0,3,2>(a), a); + return min(shuffle<2,3,0,1>(h), h); +#else + return make_float4(min(min(a.x, a.y), min(a.z, a.w))); +#endif +} + +ccl_device_inline float4 reduce_max(const float4& a) +{ +#ifdef __KERNEL_SSE__ + float4 h = max(shuffle<1,0,3,2>(a), a); + return max(shuffle<2,3,0,1>(h), h); +#else + return make_float4(max(max(a.x, a.y), max(a.z, a.w))); +#endif +} + +#if 0 +ccl_device_inline float4 reduce_add(const float4& a) +{ +#ifdef __KERNEL_SSE__ + float4 h = shuffle<1,0,3,2>(a) + a; + return shuffle<2,3,0,1>(h) + h; +#else + return make_float4((a.x + a.y) + (a.z + a.w)); +#endif +} +#endif +#endif /* !__KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_MATH_FLOAT4_H__ */ diff --git a/intern/cycles/util/util_math_int2.h b/intern/cycles/util/util_math_int2.h new file mode 100644 index 00000000000..828c49a131c --- /dev/null +++ b/intern/cycles/util/util_math_int2.h @@ -0,0 +1,77 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_MATH_INT2_H__ +#define __UTIL_MATH_INT2_H__ + +#ifndef __UTIL_MATH_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +/******************************************************************************* + * Declaration. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline bool operator==(const int2 a, const int2 b); +ccl_device_inline int2 operator+(const int2 &a, const int2 &b); +ccl_device_inline int2 operator+=(int2 &a, const int2 &b); +ccl_device_inline int2 operator-(const int2 &a, const int2 &b); +ccl_device_inline int2 operator*(const int2 &a, const int2 &b); +ccl_device_inline int2 operator/(const int2 &a, const int2 &b); +#endif /* !__KERNEL_OPENCL__ */ + +/******************************************************************************* + * Definition. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline bool operator==(const int2 a, const int2 b) +{ + return (a.x == b.x && a.y == b.y); +} + +ccl_device_inline int2 operator+(const int2 &a, const int2 &b) +{ + return make_int2(a.x + b.x, a.y + b.y); +} + +ccl_device_inline int2 operator+=(int2 &a, const int2 &b) +{ + return a = a + b; +} + +ccl_device_inline int2 operator-(const int2 &a, const int2 &b) +{ + return make_int2(a.x - b.x, a.y - b.y); +} + +ccl_device_inline int2 operator*(const int2 &a, const int2 &b) +{ + return make_int2(a.x * b.x, a.y * b.y); +} + +ccl_device_inline int2 operator/(const int2 &a, const int2 &b) +{ + return make_int2(a.x / b.x, a.y / b.y); +} +#endif /* !__KERNEL_OPENCL__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_MATH_INT2_H__ */ diff --git a/intern/cycles/util/util_math_int3.h b/intern/cycles/util/util_math_int3.h new file mode 100644 index 00000000000..fa7a02636de --- /dev/null +++ b/intern/cycles/util/util_math_int3.h @@ -0,0 +1,83 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_MATH_INT3_H__ +#define __UTIL_MATH_INT3_H__ + +#ifndef __UTIL_MATH_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +/******************************************************************************* + * Declaration. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline int3 min(int3 a, int3 b); +ccl_device_inline int3 max(int3 a, int3 b); +ccl_device_inline int3 clamp(const int3& a, int mn, int mx); +ccl_device_inline int3 clamp(const int3& a, int3& mn, int mx); +#endif /* !__KERNEL_OPENCL__ */ + +/******************************************************************************* + * Definition. + */ + +#ifndef __KERNEL_OPENCL__ +ccl_device_inline int3 min(int3 a, int3 b) +{ +#if defined(__KERNEL_SSE__) && defined(__KERNEL_SSE41__) + return int3(_mm_min_epi32(a.m128, b.m128)); +#else + return make_int3(min(a.x, b.x), min(a.y, b.y), min(a.z, b.z)); +#endif +} + +ccl_device_inline int3 max(int3 a, int3 b) +{ +#if defined(__KERNEL_SSE__) && defined(__KERNEL_SSE41__) + return int3(_mm_max_epi32(a.m128, b.m128)); +#else + return make_int3(max(a.x, b.x), max(a.y, b.y), max(a.z, b.z)); +#endif +} + +ccl_device_inline int3 clamp(const int3& a, int mn, int mx) +{ +#ifdef __KERNEL_SSE__ + return min(max(a, make_int3(mn)), make_int3(mx)); +#else + return make_int3(clamp(a.x, mn, mx), clamp(a.y, mn, mx), clamp(a.z, mn, mx)); +#endif +} + +ccl_device_inline int3 clamp(const int3& a, int3& mn, int mx) +{ +#ifdef __KERNEL_SSE__ + return min(max(a, mn), make_int3(mx)); +#else + return make_int3(clamp(a.x, mn.x, mx), + clamp(a.y, mn.y, mx), + clamp(a.z, mn.z, mx)); +#endif +} +#endif /* !__KERNEL_OPENCL__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_MATH_INT3_H__ */ diff --git a/intern/cycles/util/util_math_int4.h b/intern/cycles/util/util_math_int4.h new file mode 100644 index 00000000000..4b327c90c33 --- /dev/null +++ b/intern/cycles/util/util_math_int4.h @@ -0,0 +1,110 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_MATH_INT4_H__ +#define __UTIL_MATH_INT4_H__ + +#ifndef __UTIL_MATH_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +/******************************************************************************* + * Declaration. + */ + +#ifndef __KERNEL_GPU__ +ccl_device_inline int4 operator+(const int4& a, const int4& b); +ccl_device_inline int4 operator+=(int4& a, const int4& b); +ccl_device_inline int4 operator>>(const int4& a, int i); +ccl_device_inline int4 min(int4 a, int4 b); +ccl_device_inline int4 max(int4 a, int4 b); +ccl_device_inline int4 clamp(const int4& a, const int4& mn, const int4& mx); +ccl_device_inline int4 select(const int4& mask, const int4& a, const int4& b); +#endif /* __KERNEL_GPU__ */ + +/******************************************************************************* + * Definition. + */ + +#ifndef __KERNEL_GPU__ +ccl_device_inline int4 operator+(const int4& a, const int4& b) +{ +#ifdef __KERNEL_SSE__ + return int4(_mm_add_epi32(a.m128, b.m128)); +#else + return make_int4(a.x+b.x, a.y+b.y, a.z+b.z, a.w+b.w); +#endif +} + +ccl_device_inline int4 operator+=(int4& a, const int4& b) +{ + return a = a + b; +} + +ccl_device_inline int4 operator>>(const int4& a, int i) +{ +#ifdef __KERNEL_SSE__ + return int4(_mm_srai_epi32(a.m128, i)); +#else + return make_int4(a.x >> i, a.y >> i, a.z >> i, a.w >> i); +#endif +} + +ccl_device_inline int4 min(int4 a, int4 b) +{ +#if defined(__KERNEL_SSE__) && defined(__KERNEL_SSE41__) + return int4(_mm_min_epi32(a.m128, b.m128)); +#else + return make_int4(min(a.x, b.x), min(a.y, b.y), min(a.z, b.z), min(a.w, b.w)); +#endif +} + +ccl_device_inline int4 max(int4 a, int4 b) +{ +#if defined(__KERNEL_SSE__) && defined(__KERNEL_SSE41__) + return int4(_mm_max_epi32(a.m128, b.m128)); +#else + return make_int4(max(a.x, b.x), max(a.y, b.y), max(a.z, b.z), max(a.w, b.w)); +#endif +} + +ccl_device_inline int4 clamp(const int4& a, const int4& mn, const int4& mx) +{ + return min(max(a, mn), mx); +} + +ccl_device_inline int4 select(const int4& mask, const int4& a, const int4& b) +{ +#ifdef __KERNEL_SSE__ + const __m128 m = _mm_cvtepi32_ps(mask); + /* TODO(sergey): avoid cvt. */ + return int4(_mm_castps_si128( + _mm_or_ps(_mm_and_ps(m, _mm_castsi128_ps(a)), + _mm_andnot_ps(m, _mm_castsi128_ps(b))))); +#else + return make_int4((mask.x)? a.x: b.x, + (mask.y)? a.y: b.y, + (mask.z)? a.z: b.z, + (mask.w)? a.w: b.w); +#endif +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_MATH_INT4_H__ */ diff --git a/intern/cycles/util/util_types.h b/intern/cycles/util/util_types.h index 5a8f11694a7..137cc73b70b 100644 --- a/intern/cycles/util/util_types.h +++ b/intern/cycles/util/util_types.h @@ -18,78 +18,68 @@ #define __UTIL_TYPES_H__ #ifndef __KERNEL_OPENCL__ - -#include <stdlib.h> - +# include <stdlib.h> #endif /* Bitness */ #if defined(__ppc64__) || defined(__PPC64__) || defined(__x86_64__) || defined(__ia64__) || defined(_M_X64) -#define __KERNEL_64_BIT__ +# define __KERNEL_64_BIT__ #endif /* Qualifiers for kernel code shared by CPU and GPU */ #ifndef __KERNEL_GPU__ - -#define ccl_device static inline -#define ccl_device_noinline static -#define ccl_global -#define ccl_constant -#define ccl_local -#define ccl_local_param -#define ccl_private -#define ccl_restrict __restrict -#define __KERNEL_WITH_SSE_ALIGN__ - -#if defined(_WIN32) && !defined(FREE_WINDOWS) -#define ccl_device_inline static __forceinline -#define ccl_device_forceinline static __forceinline -#define ccl_align(...) __declspec(align(__VA_ARGS__)) -#ifdef __KERNEL_64_BIT__ -#define ccl_try_align(...) __declspec(align(__VA_ARGS__)) -#else -#undef __KERNEL_WITH_SSE_ALIGN__ -#define ccl_try_align(...) /* not support for function arguments (error C2719) */ -#endif -#define ccl_may_alias -#define ccl_always_inline __forceinline -#define ccl_never_inline __declspec(noinline) -#define ccl_maybe_unused - -#else - -#define ccl_device_inline static inline __attribute__((always_inline)) -#define ccl_device_forceinline static inline __attribute__((always_inline)) -#define ccl_align(...) __attribute__((aligned(__VA_ARGS__))) -#ifndef FREE_WINDOWS64 -#define __forceinline inline __attribute__((always_inline)) -#endif -#define ccl_try_align(...) __attribute__((aligned(__VA_ARGS__))) -#define ccl_may_alias __attribute__((__may_alias__)) -#define ccl_always_inline __attribute__((always_inline)) -#define ccl_never_inline __attribute__((noinline)) -#define ccl_maybe_unused __attribute__((used)) - -#endif - -#endif +# define ccl_device static inline +# define ccl_device_noinline static +# define ccl_global +# define ccl_constant +# define ccl_local +# define ccl_local_param +# define ccl_private +# define ccl_restrict __restrict +# define __KERNEL_WITH_SSE_ALIGN__ + +# if defined(_WIN32) && !defined(FREE_WINDOWS) +# define ccl_device_inline static __forceinline +# define ccl_device_forceinline static __forceinline +# define ccl_align(...) __declspec(align(__VA_ARGS__)) +# ifdef __KERNEL_64_BIT__ +# define ccl_try_align(...) __declspec(align(__VA_ARGS__)) +# else /* __KERNEL_64_BIT__ */ +# undef __KERNEL_WITH_SSE_ALIGN__ +/* No support for function arguments (error C2719). */ +# define ccl_try_align(...) +# endif /* __KERNEL_64_BIT__ */ +# define ccl_may_alias +# define ccl_always_inline __forceinline +# define ccl_never_inline __declspec(noinline) +# define ccl_maybe_unused +# else /* _WIN32 && !FREE_WINDOWS */ +# define ccl_device_inline static inline __attribute__((always_inline)) +# define ccl_device_forceinline static inline __attribute__((always_inline)) +# define ccl_align(...) __attribute__((aligned(__VA_ARGS__))) +# ifndef FREE_WINDOWS64 +# define __forceinline inline __attribute__((always_inline)) +# endif +# define ccl_try_align(...) __attribute__((aligned(__VA_ARGS__))) +# define ccl_may_alias __attribute__((__may_alias__)) +# define ccl_always_inline __attribute__((always_inline)) +# define ccl_never_inline __attribute__((noinline)) +# define ccl_maybe_unused __attribute__((used)) +# endif /* _WIN32 && !FREE_WINDOWS */ +#endif /* __KERNEL_GPU__ */ /* Standard Integer Types */ #ifndef __KERNEL_GPU__ - /* int8_t, uint16_t, and friends */ -#ifndef _WIN32 -#include <stdint.h> -#endif - +# ifndef _WIN32 +# include <stdint.h> +# endif /* SIMD Types */ - -#include "util/util_optimization.h" - -#endif +# include "util/util_optimization.h" +#endif /* __KERNEL_GPU__ */ CCL_NAMESPACE_BEGIN @@ -102,24 +92,18 @@ CCL_NAMESPACE_BEGIN /* Shorter Unsigned Names */ #ifndef __KERNEL_OPENCL__ - typedef unsigned char uchar; typedef unsigned int uint; - #endif /* Fixed Bits Types */ #ifdef __KERNEL_OPENCL__ - typedef ulong uint64_t; - #endif #ifndef __KERNEL_GPU__ - -#ifdef _WIN32 - +# ifdef _WIN32 typedef signed char int8_t; typedef unsigned char uint8_t; @@ -131,356 +115,17 @@ typedef unsigned int uint32_t; typedef long long int64_t; typedef unsigned long long uint64_t; - -#ifdef __KERNEL_64_BIT__ +# ifdef __KERNEL_64_BIT__ typedef int64_t ssize_t; -#else +# else typedef int32_t ssize_t; -#endif - -#endif +# endif +# endif /* _WIN32 */ /* Generic Memory Pointer */ typedef uint64_t device_ptr; - -/* Vector Types */ - -struct uchar2 { - uchar x, y; - - __forceinline uchar operator[](int i) const { return *(&x + i); } - __forceinline uchar& operator[](int i) { return *(&x + i); } -}; - -struct uchar3 { - uchar x, y, z; - - __forceinline uchar operator[](int i) const { return *(&x + i); } - __forceinline uchar& operator[](int i) { return *(&x + i); } -}; - -struct uchar4 { - uchar x, y, z, w; - - __forceinline uchar operator[](int i) const { return *(&x + i); } - __forceinline uchar& operator[](int i) { return *(&x + i); } -}; - -struct int2 { - int x, y; - - __forceinline int operator[](int i) const { return *(&x + i); } - __forceinline int& operator[](int i) { return *(&x + i); } -}; - -struct ccl_try_align(16) int3 { -#ifdef __KERNEL_SSE__ - union { - __m128i m128; - struct { int x, y, z, w; }; - }; - - __forceinline int3() {} - __forceinline explicit int3(const __m128i& a) : m128(a) {} - __forceinline operator const __m128i&(void) const { return m128; } - __forceinline operator __m128i&(void) { return m128; } - - int3(const int3& a) { m128 = a.m128; } - int3& operator =(const int3& a) { m128 = a.m128; return *this; } -#else - int x, y, z, w; -#endif - - __forceinline int operator[](int i) const { return *(&x + i); } - __forceinline int& operator[](int i) { return *(&x + i); } -}; - -struct ccl_try_align(16) int4 { -#ifdef __KERNEL_SSE__ - union { - __m128i m128; - struct { int x, y, z, w; }; - }; - - __forceinline int4() {} - __forceinline explicit int4(const __m128i& a) : m128(a) {} - __forceinline operator const __m128i&(void) const { return m128; } - __forceinline operator __m128i&(void) { return m128; } - - int4(const int4& a) : m128(a.m128) {} - int4& operator=(const int4& a) { m128 = a.m128; return *this; } -#else - int x, y, z, w; -#endif - - __forceinline int operator[](int i) const { return *(&x + i); } - __forceinline int& operator[](int i) { return *(&x + i); } -}; - -struct uint2 { - uint x, y; - - __forceinline uint operator[](uint i) const { return *(&x + i); } - __forceinline uint& operator[](uint i) { return *(&x + i); } -}; - -struct uint3 { - uint x, y, z; - - __forceinline uint operator[](uint i) const { return *(&x + i); } - __forceinline uint& operator[](uint i) { return *(&x + i); } -}; - -struct uint4 { - uint x, y, z, w; - - __forceinline uint operator[](uint i) const { return *(&x + i); } - __forceinline uint& operator[](uint i) { return *(&x + i); } -}; - -struct float2 { - float x, y; - - __forceinline float operator[](int i) const { return *(&x + i); } - __forceinline float& operator[](int i) { return *(&x + i); } -}; - -struct ccl_try_align(16) float3 { -#ifdef __KERNEL_SSE__ - union { - __m128 m128; - struct { float x, y, z, w; }; - }; - - __forceinline float3() {} - __forceinline explicit float3(const __m128& a) : m128(a) {} - __forceinline operator const __m128&(void) const { return m128; } - __forceinline operator __m128&(void) { return m128; } - - __forceinline float3(const float3& a) : m128(a.m128) {} - __forceinline float3& operator =(const float3& a) { m128 = a.m128; return *this; } -#else - float x, y, z, w; -#endif - - __forceinline float operator[](int i) const { return *(&x + i); } - __forceinline float& operator[](int i) { return *(&x + i); } -}; - -struct ccl_try_align(16) float4 { -#ifdef __KERNEL_SSE__ - union { - __m128 m128; - struct { float x, y, z, w; }; - }; - - __forceinline float4() {} - __forceinline explicit float4(const __m128& a) : m128(a) {} - __forceinline operator const __m128&(void) const { return m128; } - __forceinline operator __m128&(void) { return m128; } - - __forceinline float4(const float4& a) : m128(a.m128) {} - __forceinline float4& operator =(const float4& a) { m128 = a.m128; return *this; } - -#else - float x, y, z, w; -#endif - - __forceinline float operator[](int i) const { return *(&x + i); } - __forceinline float& operator[](int i) { return *(&x + i); } -}; - -template<typename T> -class vector3 -{ -public: - T x, y, z; - - ccl_always_inline vector3() {} - ccl_always_inline vector3(const T& a) - : x(a), y(a), z(a) {} - ccl_always_inline vector3(const T& x, const T& y, const T& z) - : x(x), y(y), z(z) {} -}; - -#endif - -#ifndef __KERNEL_GPU__ - -/* Vector Type Constructors - * - * OpenCL does not support C++ class, so we use these instead. */ - -ccl_device_inline uchar2 make_uchar2(uchar x, uchar y) -{ - uchar2 a = {x, y}; - return a; -} - -ccl_device_inline uchar3 make_uchar3(uchar x, uchar y, uchar z) -{ - uchar3 a = {x, y, z}; - return a; -} - -ccl_device_inline uchar4 make_uchar4(uchar x, uchar y, uchar z, uchar w) -{ - uchar4 a = {x, y, z, w}; - return a; -} - -ccl_device_inline int2 make_int2(int x, int y) -{ - int2 a = {x, y}; - return a; -} - -ccl_device_inline int3 make_int3(int x, int y, int z) -{ -#ifdef __KERNEL_SSE__ - int3 a; - a.m128 = _mm_set_epi32(0, z, y, x); -#else - int3 a = {x, y, z, 0}; -#endif - - return a; -} - -ccl_device_inline int4 make_int4(int x, int y, int z, int w) -{ -#ifdef __KERNEL_SSE__ - int4 a; - a.m128 = _mm_set_epi32(w, z, y, x); -#else - int4 a = {x, y, z, w}; -#endif - - return a; -} - -ccl_device_inline uint2 make_uint2(uint x, uint y) -{ - uint2 a = {x, y}; - return a; -} - -ccl_device_inline uint3 make_uint3(uint x, uint y, uint z) -{ - uint3 a = {x, y, z}; - return a; -} - -ccl_device_inline uint4 make_uint4(uint x, uint y, uint z, uint w) -{ - uint4 a = {x, y, z, w}; - return a; -} - -ccl_device_inline float2 make_float2(float x, float y) -{ - float2 a = {x, y}; - return a; -} - -ccl_device_inline float3 make_float3(float x, float y, float z) -{ -#ifdef __KERNEL_SSE__ - float3 a; - a.m128 = _mm_set_ps(0.0f, z, y, x); -#else - float3 a = {x, y, z, 0.0f}; -#endif - - return a; -} - -ccl_device_inline float4 make_float4(float x, float y, float z, float w) -{ -#ifdef __KERNEL_SSE__ - float4 a; - a.m128 = _mm_set_ps(w, z, y, x); -#else - float4 a = {x, y, z, w}; -#endif - - return a; -} - -ccl_device_inline int3 make_int3(int i) -{ -#ifdef __KERNEL_SSE__ - int3 a; - a.m128 = _mm_set1_epi32(i); -#else - int3 a = {i, i, i, i}; -#endif - - return a; -} - -ccl_device_inline int4 make_int4(int i) -{ -#ifdef __KERNEL_SSE__ - int4 a; - a.m128 = _mm_set1_epi32(i); -#else - int4 a = {i, i, i, i}; -#endif - - return a; -} - -ccl_device_inline float3 make_float3(float f) -{ -#ifdef __KERNEL_SSE__ - float3 a; - a.m128 = _mm_set1_ps(f); -#else - float3 a = {f, f, f, f}; -#endif - - return a; -} - -ccl_device_inline float4 make_float4(float f) -{ -#ifdef __KERNEL_SSE__ - float4 a; - a.m128 = _mm_set1_ps(f); -#else - float4 a = {f, f, f, f}; -#endif - - return a; -} - -ccl_device_inline float4 make_float4(const int4& i) -{ -#ifdef __KERNEL_SSE__ - float4 a; - a.m128 = _mm_cvtepi32_ps(i.m128); -#else - float4 a = {(float)i.x, (float)i.y, (float)i.z, (float)i.w}; -#endif - - return a; -} - -ccl_device_inline int4 make_int4(const float3& f) -{ -#ifdef __KERNEL_SSE__ - int4 a; - a.m128 = _mm_cvtps_epi32(f.m128); -#else - int4 a = {(int)f.x, (int)f.y, (int)f.z, (int)f.w}; -#endif - - return a; -} - -#endif +#endif /* __KERNEL_GPU__ */ ccl_device_inline size_t align_up(size_t offset, size_t alignment) { @@ -554,7 +199,7 @@ template<typename T> static inline T decltype_helper(T x) { return x; } * ... the compiler optimizes away the temp var */ #ifdef __GNUC__ #define CHECK_TYPE(var, type) { \ - TYPEOF(var) *__tmp; \ + TYPEOF(var) *__tmp; \ __tmp = (type *)NULL; \ (void)__tmp; \ } (void)0 @@ -576,5 +221,50 @@ template<typename T> static inline T decltype_helper(T x) { return x; } CCL_NAMESPACE_END +#ifndef __KERNEL_GPU__ +# include <cassert> +# define util_assert(statement) assert(statement) +#else +# define util_assert(statement) +#endif + +/* Vectorized types declaration. */ +#include "util/util_types_uchar2.h" +#include "util/util_types_uchar3.h" +#include "util/util_types_uchar4.h" + +#include "util/util_types_int2.h" +#include "util/util_types_int3.h" +#include "util/util_types_int4.h" + +#include "util/util_types_uint2.h" +#include "util/util_types_uint3.h" +#include "util/util_types_uint4.h" + +#include "util/util_types_float2.h" +#include "util/util_types_float3.h" +#include "util/util_types_float4.h" + +#include "util/util_types_vector3.h" + +/* Vectorized types implementation. */ +#include "util/util_types_uchar2_impl.h" +#include "util/util_types_uchar3_impl.h" +#include "util/util_types_uchar4_impl.h" + +#include "util/util_types_int2_impl.h" +#include "util/util_types_int3_impl.h" +#include "util/util_types_int4_impl.h" + +#include "util/util_types_uint2_impl.h" +#include "util/util_types_uint3_impl.h" +#include "util/util_types_uint4_impl.h" + +#include "util/util_types_float2_impl.h" +#include "util/util_types_float3_impl.h" +#include "util/util_types_float4_impl.h" + +#include "util/util_types_vector3_impl.h" + #endif /* __UTIL_TYPES_H__ */ diff --git a/intern/cycles/util/util_types_float2.h b/intern/cycles/util/util_types_float2.h new file mode 100644 index 00000000000..ec7a1f717a1 --- /dev/null +++ b/intern/cycles/util/util_types_float2.h @@ -0,0 +1,40 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_FLOAT2_H__ +#define __UTIL_TYPES_FLOAT2_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct float2 { + float x, y; + + __forceinline float operator[](int i) const; + __forceinline float& operator[](int i); +}; + +ccl_device_inline float2 make_float2(float x, float y); +ccl_device_inline void print_float2(const char *label, const float2& a); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_FLOAT2_H__ */ diff --git a/intern/cycles/util/util_types_float2_impl.h b/intern/cycles/util/util_types_float2_impl.h new file mode 100644 index 00000000000..782dda195eb --- /dev/null +++ b/intern/cycles/util/util_types_float2_impl.h @@ -0,0 +1,59 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_FLOAT2_IMPL_H__ +#define __UTIL_TYPES_FLOAT2_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +#ifndef __KERNEL_GPU__ +# include <cstdio> +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +__forceinline float float2::operator[](int i) const +{ + util_assert(i >= 0); + util_assert(i < 2); + return *(&x + i); +} + +__forceinline float& float2::operator[](int i) +{ + util_assert(i >= 0); + util_assert(i < 2); + return *(&x + i); +} + +ccl_device_inline float2 make_float2(float x, float y) +{ + float2 a = {x, y}; + return a; +} + +ccl_device_inline void print_float2(const char *label, const float2& a) +{ + printf("%s: %.8f %.8f\n", label, (double)a.x, (double)a.y); +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_FLOAT2_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_float3.h b/intern/cycles/util/util_types_float3.h new file mode 100644 index 00000000000..28146ad04f7 --- /dev/null +++ b/intern/cycles/util/util_types_float3.h @@ -0,0 +1,57 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_FLOAT3_H__ +#define __UTIL_TYPES_FLOAT3_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct ccl_try_align(16) float3 { +#ifdef __KERNEL_SSE__ + union { + __m128 m128; + struct { float x, y, z, w; }; + }; + + __forceinline float3(); + __forceinline float3(const float3& a); + __forceinline explicit float3(const __m128& a); + + __forceinline operator const __m128&(void) const; + __forceinline operator __m128&(void); + + __forceinline float3& operator =(const float3& a); +#else /* __KERNEL_SSE__ */ + float x, y, z, w; +#endif /* __KERNEL_SSE__ */ + + __forceinline float operator[](int i) const; + __forceinline float& operator[](int i); +}; + +ccl_device_inline float3 make_float3(float f); +ccl_device_inline float3 make_float3(float x, float y, float z); +ccl_device_inline void print_float3(const char *label, const float3& a); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_FLOAT3_H__ */ diff --git a/intern/cycles/util/util_types_float3_impl.h b/intern/cycles/util/util_types_float3_impl.h new file mode 100644 index 00000000000..45f61767d3f --- /dev/null +++ b/intern/cycles/util/util_types_float3_impl.h @@ -0,0 +1,105 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_FLOAT3_IMPL_H__ +#define __UTIL_TYPES_FLOAT3_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +#ifndef __KERNEL_GPU__ +# include <cstdio> +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +#ifdef __KERNEL_SSE__ +__forceinline float3::float3() +{ +} + +__forceinline float3::float3(const float3& a) + : m128(a.m128) +{ +} + +__forceinline float3::float3(const __m128& a) + : m128(a) +{ +} + +__forceinline float3::operator const __m128&(void) const +{ + return m128; +} + +__forceinline float3::operator __m128&(void) +{ + return m128; +} + +__forceinline float3& float3::operator =(const float3& a) +{ + m128 = a.m128; + return *this; +} +#endif /* __KERNEL_SSE__ */ + +__forceinline float float3::operator[](int i) const +{ + util_assert(i >= 0); + util_assert(i < 3); + return *(&x + i); +} + +__forceinline float& float3::operator[](int i) +{ + util_assert(i >= 0); + util_assert(i < 3); + return *(&x + i); +} + +ccl_device_inline float3 make_float3(float f) +{ +#ifdef __KERNEL_SSE__ + float3 a(_mm_set1_ps(f)); +#else + float3 a = {f, f, f, f}; +#endif + return a; +} + +ccl_device_inline float3 make_float3(float x, float y, float z) +{ +#ifdef __KERNEL_SSE__ + float3 a(_mm_set_ps(0.0f, z, y, x)); +#else + float3 a = {x, y, z, 0.0f}; +#endif + return a; +} + +ccl_device_inline void print_float3(const char *label, const float3& a) +{ + printf("%s: %.8f %.8f %.8f\n", label, (double)a.x, (double)a.y, (double)a.z); +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_FLOAT3_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_float4.h b/intern/cycles/util/util_types_float4.h new file mode 100644 index 00000000000..a7d9abe1b95 --- /dev/null +++ b/intern/cycles/util/util_types_float4.h @@ -0,0 +1,61 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_FLOAT4_H__ +#define __UTIL_TYPES_FLOAT4_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct int4; + +struct ccl_try_align(16) float4 { +#ifdef __KERNEL_SSE__ + union { + __m128 m128; + struct { float x, y, z, w; }; + }; + + __forceinline float4(); + __forceinline float4(const float4& a); + __forceinline explicit float4(const __m128& a); + + __forceinline operator const __m128&(void) const; + __forceinline operator __m128&(void); + + __forceinline float4& operator =(const float4& a); + +#else /* __KERNEL_SSE__ */ + float x, y, z, w; +#endif /* __KERNEL_SSE__ */ + + __forceinline float operator[](int i) const; + __forceinline float& operator[](int i); +}; + +ccl_device_inline float4 make_float4(float f); +ccl_device_inline float4 make_float4(float x, float y, float z, float w); +ccl_device_inline float4 make_float4(const int4& i); +ccl_device_inline void print_float4(const char *label, const float4& a); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_FLOAT4_H__ */ diff --git a/intern/cycles/util/util_types_float4_impl.h b/intern/cycles/util/util_types_float4_impl.h new file mode 100644 index 00000000000..ff3ec4d4ecf --- /dev/null +++ b/intern/cycles/util/util_types_float4_impl.h @@ -0,0 +1,117 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_FLOAT4_IMPL_H__ +#define __UTIL_TYPES_FLOAT4_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +#ifndef __KERNEL_GPU__ +# include <cstdio> +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +#ifdef __KERNEL_SSE__ +__forceinline float4::float4() +{ +} + +__forceinline float4::float4(const float4& a) + : m128(a.m128) +{ +} + +__forceinline float4::float4(const __m128& a) + : m128(a) +{ +} + +__forceinline float4::operator const __m128&(void) const +{ + return m128; +} + +__forceinline float4::operator __m128&(void) +{ + return m128; +} + +__forceinline float4& float4::operator =(const float4& a) +{ + m128 = a.m128; + return *this; +} +#endif /* __KERNEL_SSE__ */ + +__forceinline float float4::operator[](int i) const +{ + util_assert(i >= 0); + util_assert(i < 4); + return *(&x + i); +} + +__forceinline float& float4::operator[](int i) +{ + util_assert(i >= 0); + util_assert(i < 4); + return *(&x + i); +} + +ccl_device_inline float4 make_float4(float f) +{ +#ifdef __KERNEL_SSE__ + float4 a(_mm_set1_ps(f)); +#else + float4 a = {f, f, f, f}; +#endif + return a; +} + +ccl_device_inline float4 make_float4(float x, float y, float z, float w) +{ +#ifdef __KERNEL_SSE__ + float4 a(_mm_set_ps(w, z, y, x)); +#else + float4 a = {x, y, z, w}; +#endif + return a; +} + +ccl_device_inline float4 make_float4(const int4& i) +{ +#ifdef __KERNEL_SSE__ + float4 a(_mm_cvtepi32_ps(i.m128)); +#else + float4 a = {(float)i.x, (float)i.y, (float)i.z, (float)i.w}; +#endif + return a; +} + +ccl_device_inline void print_float4(const char *label, const float4& a) +{ + printf("%s: %.8f %.8f %.8f %.8f\n", + label, + (double)a.x, (double)a.y, (double)a.z, (double)a.w); +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_FLOAT4_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_int2.h b/intern/cycles/util/util_types_int2.h new file mode 100644 index 00000000000..82e860f89eb --- /dev/null +++ b/intern/cycles/util/util_types_int2.h @@ -0,0 +1,39 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_INT2_H__ +#define __UTIL_TYPES_INT2_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct int2 { + int x, y; + + __forceinline int operator[](int i) const; + __forceinline int& operator[](int i); +}; + +ccl_device_inline int2 make_int2(int x, int y); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_INT2_H__ */ diff --git a/intern/cycles/util/util_types_int2_impl.h b/intern/cycles/util/util_types_int2_impl.h new file mode 100644 index 00000000000..c7d3942e723 --- /dev/null +++ b/intern/cycles/util/util_types_int2_impl.h @@ -0,0 +1,50 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_INT2_IMPL_H__ +#define __UTIL_TYPES_INT2_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +int int2::operator[](int i) const +{ + util_assert(i >= 0); + util_assert(i < 2); + return *(&x + i); +} + +int& int2::operator[](int i) +{ + util_assert(i >= 0); + util_assert(i < 2); + return *(&x + i); +} + +ccl_device_inline int2 make_int2(int x, int y) +{ + int2 a = {x, y}; + return a; +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_INT2_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_int3.h b/intern/cycles/util/util_types_int3.h new file mode 100644 index 00000000000..9d43b201c02 --- /dev/null +++ b/intern/cycles/util/util_types_int3.h @@ -0,0 +1,57 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_INT3_H__ +#define __UTIL_TYPES_INT3_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct ccl_try_align(16) int3 { +#ifdef __KERNEL_SSE__ + union { + __m128i m128; + struct { int x, y, z, w; }; + }; + + __forceinline int3(); + __forceinline int3(const int3& a); + __forceinline explicit int3(const __m128i& a); + + __forceinline operator const __m128i&(void) const; + __forceinline operator __m128i&(void); + + __forceinline int3& operator =(const int3& a); +#else /* __KERNEL_SSE__ */ + int x, y, z, w; +#endif /* __KERNEL_SSE__ */ + + __forceinline int operator[](int i) const; + __forceinline int& operator[](int i); +}; + +ccl_device_inline int3 make_int3(int i); +ccl_device_inline int3 make_int3(int x, int y, int z); +ccl_device_inline void print_int3(const char *label, const int3& a); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_INT3_H__ */ diff --git a/intern/cycles/util/util_types_int3_impl.h b/intern/cycles/util/util_types_int3_impl.h new file mode 100644 index 00000000000..ada50c4812c --- /dev/null +++ b/intern/cycles/util/util_types_int3_impl.h @@ -0,0 +1,106 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_INT3_IMPL_H__ +#define __UTIL_TYPES_INT3_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +#ifndef __KERNEL_GPU__ +# include <cstdio> +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +#ifdef __KERNEL_SSE__ +__forceinline int3::int3() +{ +} + +__forceinline int3::int3(const __m128i& a) + : m128(a) +{ +} + +__forceinline int3::int3(const int3& a) + : m128(a.m128) +{ +} + +__forceinline int3::operator const __m128i&(void) const +{ + return m128; +} + +__forceinline int3::operator __m128i&(void) +{ + return m128; +} + +__forceinline int3& int3::operator =(const int3& a) +{ + m128 = a.m128; + return *this; +} +#endif /* __KERNEL_SSE__ */ + +__forceinline int int3::operator[](int i) const +{ + util_assert(i >= 0); + util_assert(i < 3); + return *(&x + i); +} + +__forceinline int& int3::operator[](int i) +{ + util_assert(i >= 0); + util_assert(i < 3); + return *(&x + i); +} + +ccl_device_inline int3 make_int3(int i) +{ +#ifdef __KERNEL_SSE__ + int3 a(_mm_set1_epi32(i)); +#else + int3 a = {i, i, i, i}; +#endif + return a; +} + +ccl_device_inline int3 make_int3(int x, int y, int z) +{ +#ifdef __KERNEL_SSE__ + int3 a(_mm_set_epi32(0, z, y, x)); +#else + int3 a = {x, y, z, 0}; +#endif + + return a; +} + +ccl_device_inline void print_int3(const char *label, const int3& a) +{ + printf("%s: %d %d %d\n", label, a.x, a.y, a.z); +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_INT3_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_int4.h b/intern/cycles/util/util_types_int4.h new file mode 100644 index 00000000000..cdd0ecbdae5 --- /dev/null +++ b/intern/cycles/util/util_types_int4.h @@ -0,0 +1,61 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_INT4_H__ +#define __UTIL_TYPES_INT4_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ + +struct float3; + +struct ccl_try_align(16) int4 { +#ifdef __KERNEL_SSE__ + union { + __m128i m128; + struct { int x, y, z, w; }; + }; + + __forceinline int4(); + __forceinline int4(const int4& a); + __forceinline explicit int4(const __m128i& a); + + __forceinline operator const __m128i&(void) const; + __forceinline operator __m128i&(void); + + __forceinline int4& operator=(const int4& a); +#else /* __KERNEL_SSE__ */ + int x, y, z, w; +#endif /* __KERNEL_SSE__ */ + + __forceinline int operator[](int i) const; + __forceinline int& operator[](int i); +}; + +ccl_device_inline int4 make_int4(int i); +ccl_device_inline int4 make_int4(int x, int y, int z, int w); +ccl_device_inline int4 make_int4(const float3& f); +ccl_device_inline void print_int4(const char *label, const int4& a); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_INT4_H__ */ diff --git a/intern/cycles/util/util_types_int4_impl.h b/intern/cycles/util/util_types_int4_impl.h new file mode 100644 index 00000000000..07cdc88f2dc --- /dev/null +++ b/intern/cycles/util/util_types_int4_impl.h @@ -0,0 +1,115 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_INT4_IMPL_H__ +#define __UTIL_TYPES_INT4_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +#ifndef __KERNEL_GPU__ +# include <cstdio> +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +#ifdef __KERNEL_SSE__ +__forceinline int4::int4() +{ +} + +__forceinline int4::int4(const int4& a) + : m128(a.m128) +{ +} + +__forceinline int4::int4(const __m128i& a) + : m128(a) +{ +} + +__forceinline int4::operator const __m128i&(void) const +{ + return m128; +} + +__forceinline int4::operator __m128i&(void) +{ + return m128; +} + +__forceinline int4& int4::operator=(const int4& a) +{ + m128 = a.m128; + return *this; +} +#endif /* __KERNEL_SSE__ */ + +__forceinline int int4::operator[](int i) const +{ + util_assert(i >= 0); + util_assert(i < 4); + return *(&x + i); +} + +__forceinline int& int4::operator[](int i) +{ + util_assert(i >= 0); + util_assert(i < 4); + return *(&x + i); +} + +ccl_device_inline int4 make_int4(int i) +{ +#ifdef __KERNEL_SSE__ + int4 a(_mm_set1_epi32(i)); +#else + int4 a = {i, i, i, i}; +#endif + return a; +} + +ccl_device_inline int4 make_int4(int x, int y, int z, int w) +{ +#ifdef __KERNEL_SSE__ + int4 a(_mm_set_epi32(w, z, y, x)); +#else + int4 a = {x, y, z, w}; +#endif + return a; +} + +ccl_device_inline int4 make_int4(const float3& f) +{ +#ifdef __KERNEL_SSE__ + int4 a(_mm_cvtps_epi32(f.m128)); +#else + int4 a = {(int)f.x, (int)f.y, (int)f.z, (int)f.w}; +#endif + return a; +} + +ccl_device_inline void print_int4(const char *label, const int4& a) +{ + printf("%s: %d %d %d %d\n", label, a.x, a.y, a.z, a.w); +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_INT4_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_uchar2.h b/intern/cycles/util/util_types_uchar2.h new file mode 100644 index 00000000000..f618a2234ca --- /dev/null +++ b/intern/cycles/util/util_types_uchar2.h @@ -0,0 +1,39 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UCHAR2_H__ +#define __UTIL_TYPES_UCHAR2_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct uchar2 { + uchar x, y; + + __forceinline uchar operator[](int i) const; + __forceinline uchar& operator[](int i); +}; + +ccl_device_inline uchar2 make_uchar2(uchar x, uchar y); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UCHAR2_H__ */ diff --git a/intern/cycles/util/util_types_uchar2_impl.h b/intern/cycles/util/util_types_uchar2_impl.h new file mode 100644 index 00000000000..d5f196d0ce0 --- /dev/null +++ b/intern/cycles/util/util_types_uchar2_impl.h @@ -0,0 +1,50 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UCHAR2_IMPL_H__ +#define __UTIL_TYPES_UCHAR2_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +uchar uchar2::operator[](int i) const +{ + util_assert(i >= 0); + util_assert(i < 2); + return *(&x + i); +} + +uchar& uchar2::operator[](int i) +{ + util_assert(i >= 0); + util_assert(i < 2); + return *(&x + i); +} + +ccl_device_inline uchar2 make_uchar2(uchar x, uchar y) +{ + uchar2 a = {x, y}; + return a; +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UCHAR2_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_uchar3.h b/intern/cycles/util/util_types_uchar3.h new file mode 100644 index 00000000000..1e3644e6fd6 --- /dev/null +++ b/intern/cycles/util/util_types_uchar3.h @@ -0,0 +1,39 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UCHAR3_H__ +#define __UTIL_TYPES_UCHAR3_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct uchar3 { + uchar x, y, z; + + __forceinline uchar operator[](int i) const; + __forceinline uchar& operator[](int i); +}; + +ccl_device_inline uchar3 make_uchar3(uchar x, uchar y, uchar z); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UCHAR3_H__ */ diff --git a/intern/cycles/util/util_types_uchar3_impl.h b/intern/cycles/util/util_types_uchar3_impl.h new file mode 100644 index 00000000000..611021efb7f --- /dev/null +++ b/intern/cycles/util/util_types_uchar3_impl.h @@ -0,0 +1,50 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UCHAR3_IMPL_H__ +#define __UTIL_TYPES_UCHAR3_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +uchar uchar3::operator[](int i) const +{ + util_assert(i >= 0); + util_assert(i < 3); + return *(&x + i); +} + +uchar& uchar3::operator[](int i) +{ + util_assert(i >= 0); + util_assert(i < 3); + return *(&x + i); +} + +ccl_device_inline uchar3 make_uchar3(uchar x, uchar y, uchar z) +{ + uchar3 a = {x, y, z}; + return a; +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UCHAR3_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_uchar4.h b/intern/cycles/util/util_types_uchar4.h new file mode 100644 index 00000000000..3802cebbfb9 --- /dev/null +++ b/intern/cycles/util/util_types_uchar4.h @@ -0,0 +1,39 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UCHAR4_H__ +#define __UTIL_TYPES_UCHAR4_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct uchar4 { + uchar x, y, z, w; + + __forceinline uchar operator[](int i) const; + __forceinline uchar& operator[](int i); +}; + +ccl_device_inline uchar4 make_uchar4(uchar x, uchar y, uchar z, uchar w); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UCHAR4_H__ */ diff --git a/intern/cycles/util/util_types_uchar4_impl.h b/intern/cycles/util/util_types_uchar4_impl.h new file mode 100644 index 00000000000..03039f60c54 --- /dev/null +++ b/intern/cycles/util/util_types_uchar4_impl.h @@ -0,0 +1,50 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UCHAR4_IMPL_H__ +#define __UTIL_TYPES_UCHAR4_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +uchar uchar4::operator[](int i) const +{ + util_assert(i >= 0); + util_assert(i < 4); + return *(&x + i); +} + +uchar& uchar4::operator[](int i) +{ + util_assert(i >= 0); + util_assert(i < 4); + return *(&x + i); +} + +ccl_device_inline uchar4 make_uchar4(uchar x, uchar y, uchar z, uchar w) +{ + uchar4 a = {x, y, z, w}; + return a; +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UCHAR4_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_uint2.h b/intern/cycles/util/util_types_uint2.h new file mode 100644 index 00000000000..c4a31899614 --- /dev/null +++ b/intern/cycles/util/util_types_uint2.h @@ -0,0 +1,39 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UINT2_H__ +#define __UTIL_TYPES_UINT2_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct uint2 { + uint x, y; + + __forceinline uint operator[](uint i) const; + __forceinline uint& operator[](uint i); +}; + +ccl_device_inline uint2 make_uint2(uint x, uint y); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UINT2_H__ */ diff --git a/intern/cycles/util/util_types_uint2_impl.h b/intern/cycles/util/util_types_uint2_impl.h new file mode 100644 index 00000000000..b50ffa2667f --- /dev/null +++ b/intern/cycles/util/util_types_uint2_impl.h @@ -0,0 +1,48 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UINT2_IMPL_H__ +#define __UTIL_TYPES_UINT2_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +__forceinline uint uint2::operator[](uint i) const +{ + util_assert(i < 2); + return *(&x + i); +} + +__forceinline uint& uint2::operator[](uint i) +{ + util_assert(i < 2); + return *(&x + i); +} + +ccl_device_inline uint2 make_uint2(uint x, uint y) +{ + uint2 a = {x, y}; + return a; +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UINT2_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_uint3.h b/intern/cycles/util/util_types_uint3.h new file mode 100644 index 00000000000..aeeecd2df06 --- /dev/null +++ b/intern/cycles/util/util_types_uint3.h @@ -0,0 +1,39 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UINT3_H__ +#define __UTIL_TYPES_UINT3_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct uint3 { + uint x, y, z; + + __forceinline uint operator[](uint i) const; + __forceinline uint& operator[](uint i); +}; + +ccl_device_inline uint3 make_uint3(uint x, uint y, uint z); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UINT3_H__ */ diff --git a/intern/cycles/util/util_types_uint3_impl.h b/intern/cycles/util/util_types_uint3_impl.h new file mode 100644 index 00000000000..26005d5baff --- /dev/null +++ b/intern/cycles/util/util_types_uint3_impl.h @@ -0,0 +1,48 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UINT3_IMPL_H__ +#define __UTIL_TYPES_UINT3_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +__forceinline uint uint3::operator[](uint i) const +{ + util_assert(i < 3); + return *(&x + i); +} + +__forceinline uint& uint3::operator[](uint i) +{ + util_assert(i < 3); + return *(&x + i); +} + +ccl_device_inline uint3 make_uint3(uint x, uint y, uint z) +{ + uint3 a = {x, y, z}; + return a; +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UINT3_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_uint4.h b/intern/cycles/util/util_types_uint4.h new file mode 100644 index 00000000000..2d3a7bb85e4 --- /dev/null +++ b/intern/cycles/util/util_types_uint4.h @@ -0,0 +1,39 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UINT4_H__ +#define __UTIL_TYPES_UINT4_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +struct uint4 { + uint x, y, z, w; + + __forceinline uint operator[](uint i) const; + __forceinline uint& operator[](uint i); +}; + +ccl_device_inline uint4 make_uint4(uint x, uint y, uint z, uint w); +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UINT4_H__ */ diff --git a/intern/cycles/util/util_types_uint4_impl.h b/intern/cycles/util/util_types_uint4_impl.h new file mode 100644 index 00000000000..6d48131a446 --- /dev/null +++ b/intern/cycles/util/util_types_uint4_impl.h @@ -0,0 +1,48 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_UINT4_IMPL_H__ +#define __UTIL_TYPES_UINT4_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +__forceinline uint uint4::operator[](uint i) const +{ + util_assert(i < 3); + return *(&x + i); +} + +__forceinline uint& uint4::operator[](uint i) +{ + util_assert(i < 3); + return *(&x + i); +} + +ccl_device_inline uint4 make_uint4(uint x, uint y, uint z, uint w) +{ + uint4 a = {x, y, z, w}; + return a; +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_UINT4_IMPL_H__ */ diff --git a/intern/cycles/util/util_types_vector3.h b/intern/cycles/util/util_types_vector3.h new file mode 100644 index 00000000000..12acf9dc959 --- /dev/null +++ b/intern/cycles/util/util_types_vector3.h @@ -0,0 +1,41 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_VECTOR3_H__ +#define __UTIL_TYPES_VECTOR3_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +template<typename T> +class vector3 +{ +public: + T x, y, z; + + __forceinline vector3(); + __forceinline vector3(const T& a); + __forceinline vector3(const T& x, const T& y, const T& z); +}; +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_VECTOR3_H__ */ diff --git a/intern/cycles/util/util_types_vector3_impl.h b/intern/cycles/util/util_types_vector3_impl.h new file mode 100644 index 00000000000..2f6b8368540 --- /dev/null +++ b/intern/cycles/util/util_types_vector3_impl.h @@ -0,0 +1,47 @@ +/* + * Copyright 2011-2017 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. + */ + +#ifndef __UTIL_TYPES_VECTOR3_IMPL_H__ +#define __UTIL_TYPES_VECTOR3_IMPL_H__ + +#ifndef __UTIL_TYPES_H__ +# error "Do not include this file directly, include util_types.h instead." +#endif + +CCL_NAMESPACE_BEGIN + +#ifndef __KERNEL_GPU__ +template<typename T> +ccl_always_inline vector3<T>::vector3() +{ +} + +template<typename T> +ccl_always_inline vector3<T>::vector3(const T& a) + : x(a), y(a), z(a) +{ +} + +template<typename T> +ccl_always_inline vector3<T>::vector3(const T& x, const T& y, const T& z) + : x(x), y(y), z(z) +{ +} +#endif /* __KERNEL_GPU__ */ + +CCL_NAMESPACE_END + +#endif /* __UTIL_TYPES_VECTOR3_IMPL_H__ */ diff --git a/source/blender/alembic/intern/abc_archive.cc b/source/blender/alembic/intern/abc_archive.cc index 5f8fc1a3739..cdbda0ace69 100644 --- a/source/blender/alembic/intern/abc_archive.cc +++ b/source/blender/alembic/intern/abc_archive.cc @@ -28,6 +28,8 @@ # include "utfconv.h" #endif +#include <fstream> + using Alembic::Abc::Exception; using Alembic::Abc::ErrorHandler; using Alembic::Abc::IArchive; @@ -38,8 +40,9 @@ static IArchive open_archive(const std::string &filename, const std::vector<std::istream *> &input_streams, bool &is_hdf5) { + is_hdf5 = false; + try { - is_hdf5 = false; Alembic::AbcCoreOgawa::ReadArchive archive_reader(input_streams); return IArchive(archive_reader(filename), @@ -63,6 +66,27 @@ static IArchive open_archive(const std::string &filename, return IArchive(); } #else + /* Inspect the file to see whether it's really a HDF5 file. */ + char header[4]; /* char(0x89) + "HDF" */ + std::ifstream the_file(filename, std::ios::in | std::ios::binary); + if (!the_file) { + std::cerr << "Unable to open " << filename << std::endl; + } + else if (!the_file.read(header, sizeof(header))) { + std::cerr << "Unable to read from " << filename << std::endl; + } + else if (strncmp(header + 1, "HDF", 3)) { + std::cerr << filename << " has an unknown file format, unable to read." << std::endl; + } + else { + is_hdf5 = true; + std::cerr << filename << " is in the obsolete HDF5 format, unable to read." << std::endl; + } + + if (the_file.is_open()) { + the_file.close(); + } + return IArchive(); #endif } @@ -83,16 +107,20 @@ ArchiveReader::ArchiveReader(const char *filename) m_streams.push_back(&m_infile); - bool is_hdf5; - m_archive = open_archive(filename, m_streams, is_hdf5); + m_archive = open_archive(filename, m_streams, m_is_hdf5); /* We can't open an HDF5 file from a stream, so close it. */ - if (is_hdf5) { + if (m_is_hdf5) { m_infile.close(); m_streams.clear(); } } +bool ArchiveReader::is_hdf5() const +{ + return m_is_hdf5; +} + bool ArchiveReader::valid() const { return m_archive.valid(); diff --git a/source/blender/alembic/intern/abc_archive.h b/source/blender/alembic/intern/abc_archive.h index d412574b736..84309fbc9df 100644 --- a/source/blender/alembic/intern/abc_archive.h +++ b/source/blender/alembic/intern/abc_archive.h @@ -44,12 +44,21 @@ class ArchiveReader { Alembic::Abc::IArchive m_archive; std::ifstream m_infile; std::vector<std::istream *> m_streams; + bool m_is_hdf5; public: explicit ArchiveReader(const char *filename); bool valid() const; + /** + * Returns true when either Blender is compiled with HDF5 support and + * the archive was succesfully opened (valid() will also return true), + * or when Blender was built without HDF5 support but a HDF5 file was + * detected (valid() will return false). + */ + bool is_hdf5() const; + Alembic::Abc::IObject getTop(); }; diff --git a/source/blender/alembic/intern/abc_camera.cc b/source/blender/alembic/intern/abc_camera.cc index 33344f6c9c5..4f70b2a972c 100644 --- a/source/blender/alembic/intern/abc_camera.cc +++ b/source/blender/alembic/intern/abc_camera.cc @@ -117,10 +117,11 @@ bool AbcCameraReader::valid() const return m_schema.valid(); } -void AbcCameraReader::readObjectData(Main *bmain, const ISampleSelector &sample_sel) +void AbcCameraReader::readObjectData(Main *bmain, float time) { Camera *bcam = static_cast<Camera *>(BKE_camera_add(bmain, m_data_name.c_str())); + ISampleSelector sample_sel(time); CameraSample cam_sample; m_schema.get(cam_sample, sample_sel); diff --git a/source/blender/alembic/intern/abc_camera.h b/source/blender/alembic/intern/abc_camera.h index 0201dd9d6e1..1c4b7f77949 100644 --- a/source/blender/alembic/intern/abc_camera.h +++ b/source/blender/alembic/intern/abc_camera.h @@ -55,7 +55,7 @@ public: bool valid() const; - void readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel); + void readObjectData(Main *bmain, float time); }; #endif /* __ABC_CAMERA_H__ */ diff --git a/source/blender/alembic/intern/abc_curves.cc b/source/blender/alembic/intern/abc_curves.cc index d85d8df4d77..bf485c7ef61 100644 --- a/source/blender/alembic/intern/abc_curves.cc +++ b/source/blender/alembic/intern/abc_curves.cc @@ -211,7 +211,7 @@ bool AbcCurveReader::valid() const return m_curves_schema.valid(); } -void AbcCurveReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel) +void AbcCurveReader::readObjectData(Main *bmain, float time) { Curve *cu = BKE_curve_add(bmain, m_data_name.c_str(), OB_CURVE); @@ -219,6 +219,8 @@ void AbcCurveReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSele cu->actvert = CU_ACT_NONE; cu->resolu = 1; + const ISampleSelector sample_sel(time); + ICompoundProperty user_props = m_curves_schema.getUserProperties(); if (user_props) { const PropertyHeader *header = user_props.getPropertyHeader(ABC_CURVE_RESOLUTION_U_PROPNAME); @@ -387,11 +389,9 @@ void read_curve_sample(Curve *cu, const ICurvesSchema &schema, const ISampleSele * object directly and create a new DerivedMesh from that. Also we might need to * create new or delete existing NURBS in the curve. */ -DerivedMesh *AbcCurveReader::read_derivedmesh(DerivedMesh * /*dm*/, - const ISampleSelector &sample_sel, - int /*read_flag*/, - const char ** /*err_str*/) +DerivedMesh *AbcCurveReader::read_derivedmesh(DerivedMesh * /*dm*/, const float time, int /*read_flag*/, const char ** /*err_str*/) { + ISampleSelector sample_sel(time); const ICurvesSchema::Sample sample = m_curves_schema.getValue(sample_sel); const P3fArraySamplePtr &positions = sample.getPositions(); diff --git a/source/blender/alembic/intern/abc_curves.h b/source/blender/alembic/intern/abc_curves.h index 1ad6d0bf278..71b0d205820 100644 --- a/source/blender/alembic/intern/abc_curves.h +++ b/source/blender/alembic/intern/abc_curves.h @@ -55,11 +55,8 @@ public: bool valid() const; - void readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel); - DerivedMesh *read_derivedmesh(DerivedMesh *dm, - const Alembic::Abc::ISampleSelector &sample_sel, - int read_flag, - const char **err_str); + void readObjectData(Main *bmain, float time); + DerivedMesh *read_derivedmesh(DerivedMesh *, const float time, int read_flag, const char **err_str); }; /* ************************************************************************** */ diff --git a/source/blender/alembic/intern/abc_mesh.cc b/source/blender/alembic/intern/abc_mesh.cc index fc3ad239c5d..a0b175dbcdc 100644 --- a/source/blender/alembic/intern/abc_mesh.cc +++ b/source/blender/alembic/intern/abc_mesh.cc @@ -1020,15 +1020,17 @@ bool AbcMeshReader::valid() const return m_schema.valid(); } -void AbcMeshReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel) +void AbcMeshReader::readObjectData(Main *bmain, float time) { Mesh *mesh = BKE_mesh_add(bmain, m_data_name.c_str()); m_object = BKE_object_add_only_object(bmain, OB_MESH, m_object_name.c_str()); m_object->data = mesh; + const ISampleSelector sample_sel(time); + DerivedMesh *dm = CDDM_from_mesh(mesh); - DerivedMesh *ndm = this->read_derivedmesh(dm, sample_sel, MOD_MESHSEQ_READ_ALL, NULL); + DerivedMesh *ndm = this->read_derivedmesh(dm, time, MOD_MESHSEQ_READ_ALL, NULL); if (ndm != dm) { dm->release(dm); @@ -1047,11 +1049,9 @@ void AbcMeshReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSelec } } -DerivedMesh *AbcMeshReader::read_derivedmesh(DerivedMesh *dm, - const ISampleSelector &sample_sel, - int read_flag, - const char **err_str) +DerivedMesh *AbcMeshReader::read_derivedmesh(DerivedMesh *dm, const float time, int read_flag, const char **err_str) { + ISampleSelector sample_sel(time); const IPolyMeshSchema::Sample sample = m_schema.getValue(sample_sel); const P3fArraySamplePtr &positions = sample.getPositions(); @@ -1091,7 +1091,7 @@ DerivedMesh *AbcMeshReader::read_derivedmesh(DerivedMesh *dm, } CDStreamConfig config = get_config(new_dm ? new_dm : dm); - config.time = sample_sel.getRequestedTime(); + config.time = time; bool do_normals = false; read_mesh_sample(&settings, m_schema, sample_sel, config, do_normals); @@ -1238,7 +1238,7 @@ bool AbcSubDReader::valid() const return m_schema.valid(); } -void AbcSubDReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel) +void AbcSubDReader::readObjectData(Main *bmain, float time) { Mesh *mesh = BKE_mesh_add(bmain, m_data_name.c_str()); @@ -1246,7 +1246,7 @@ void AbcSubDReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSelec m_object->data = mesh; DerivedMesh *dm = CDDM_from_mesh(mesh); - DerivedMesh *ndm = this->read_derivedmesh(dm, sample_sel, MOD_MESHSEQ_READ_ALL, NULL); + DerivedMesh *ndm = this->read_derivedmesh(dm, time, MOD_MESHSEQ_READ_ALL, NULL); if (ndm != dm) { dm->release(dm); @@ -1254,6 +1254,7 @@ void AbcSubDReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSelec DM_to_mesh(ndm, mesh, m_object, CD_MASK_MESH, true); + const ISampleSelector sample_sel(time); const ISubDSchema::Sample sample = m_schema.getValue(sample_sel); Int32ArraySamplePtr indices = sample.getCreaseIndices(); Alembic::Abc::FloatArraySamplePtr sharpnesses = sample.getCreaseSharpnesses(); @@ -1284,11 +1285,9 @@ void AbcSubDReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSelec } } -DerivedMesh *AbcSubDReader::read_derivedmesh(DerivedMesh *dm, - const ISampleSelector &sample_sel, - int read_flag, - const char **err_str) +DerivedMesh *AbcSubDReader::read_derivedmesh(DerivedMesh *dm, const float time, int read_flag, const char **err_str) { + ISampleSelector sample_sel(time); const ISubDSchema::Sample sample = m_schema.getValue(sample_sel); const P3fArraySamplePtr &positions = sample.getPositions(); @@ -1328,7 +1327,7 @@ DerivedMesh *AbcSubDReader::read_derivedmesh(DerivedMesh *dm, /* Only read point data when streaming meshes, unless we need to create new ones. */ CDStreamConfig config = get_config(new_dm ? new_dm : dm); - config.time = sample_sel.getRequestedTime(); + config.time = time; read_subd_sample(&settings, m_schema, sample_sel, config); if (new_dm) { diff --git a/source/blender/alembic/intern/abc_mesh.h b/source/blender/alembic/intern/abc_mesh.h index 1b67c80cc72..64a3109232c 100644 --- a/source/blender/alembic/intern/abc_mesh.h +++ b/source/blender/alembic/intern/abc_mesh.h @@ -100,12 +100,9 @@ public: bool valid() const; - void readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel); + void readObjectData(Main *bmain, float time); - DerivedMesh *read_derivedmesh(DerivedMesh *dm, - const Alembic::Abc::ISampleSelector &sample_sel, - int read_flag, - const char **err_str); + DerivedMesh *read_derivedmesh(DerivedMesh *dm, const float time, int read_flag, const char **err_str); private: void readFaceSetsSample(Main *bmain, Mesh *mesh, size_t poly_start, @@ -124,11 +121,8 @@ public: bool valid() const; - void readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel); - DerivedMesh *read_derivedmesh(DerivedMesh *dm, - const Alembic::Abc::ISampleSelector &sample_sel, - int read_flag, - const char **err_str); + void readObjectData(Main *bmain, float time); + DerivedMesh *read_derivedmesh(DerivedMesh *dm, const float time, int read_flag, const char **err_str); }; /* ************************************************************************** */ diff --git a/source/blender/alembic/intern/abc_nurbs.cc b/source/blender/alembic/intern/abc_nurbs.cc index eaef06fd6d1..d0b9561f679 100644 --- a/source/blender/alembic/intern/abc_nurbs.cc +++ b/source/blender/alembic/intern/abc_nurbs.cc @@ -239,7 +239,7 @@ static bool set_knots(const FloatArraySamplePtr &knots, float *&nu_knots) return true; } -void AbcNurbsReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel) +void AbcNurbsReader::readObjectData(Main *bmain, float time) { Curve *cu = static_cast<Curve *>(BKE_curve_add(bmain, "abc_curve", OB_SURF)); cu->actvert = CU_ACT_NONE; @@ -253,6 +253,7 @@ void AbcNurbsReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSele nu->resolu = cu->resolu; nu->resolv = cu->resolv; + const ISampleSelector sample_sel(time); const INuPatchSchema &schema = it->first; const INuPatchSchema::Sample smp = schema.getValue(sample_sel); diff --git a/source/blender/alembic/intern/abc_nurbs.h b/source/blender/alembic/intern/abc_nurbs.h index abe460a8988..1b2e7a8391f 100644 --- a/source/blender/alembic/intern/abc_nurbs.h +++ b/source/blender/alembic/intern/abc_nurbs.h @@ -54,7 +54,7 @@ public: bool valid() const; - void readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel); + void readObjectData(Main *bmain, float time); private: void getNurbsPatches(const Alembic::Abc::IObject &obj); diff --git a/source/blender/alembic/intern/abc_object.cc b/source/blender/alembic/intern/abc_object.cc index 972438ef3a6..ce0f9225228 100644 --- a/source/blender/alembic/intern/abc_object.cc +++ b/source/blender/alembic/intern/abc_object.cc @@ -159,15 +159,6 @@ void AbcObjectReader::object(Object *ob) m_object = ob; } -DerivedMesh *AbcObjectReader::read_derivedmesh(DerivedMesh *dm, - const Alembic::Abc::ISampleSelector &UNUSED(sample_sel), - int UNUSED(read_flag), - const char **UNUSED(err_str)) -{ - return dm; -} - - static Imath::M44d blend_matrices(const Imath::M44d &m0, const Imath::M44d &m1, const float weight) { float mat0[4][4], mat1[4][4], ret[4][4]; @@ -223,6 +214,14 @@ Imath::M44d get_matrix(const IXformSchema &schema, const float time) return s0.getMatrix(); } +DerivedMesh *AbcObjectReader::read_derivedmesh(DerivedMesh *dm, + const float UNUSED(time), + int UNUSED(read_flag), + const char **UNUSED(err_str)) +{ + return dm; +} + void AbcObjectReader::setupObjectTransform(const float time) { bool is_constant = false; diff --git a/source/blender/alembic/intern/abc_object.h b/source/blender/alembic/intern/abc_object.h index a9dbaa89c97..ade232b11e7 100644 --- a/source/blender/alembic/intern/abc_object.h +++ b/source/blender/alembic/intern/abc_object.h @@ -170,12 +170,12 @@ public: virtual bool valid() const = 0; - virtual void readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel) = 0; + virtual void readObjectData(Main *bmain, float time) = 0; - DerivedMesh *read_derivedmesh(DerivedMesh *dm, - const Alembic::Abc::ISampleSelector &sample_sel, - int read_flag, - const char **err_str); + virtual DerivedMesh *read_derivedmesh(DerivedMesh *dm, + const float time, + int read_flag, + const char **err_str); /** Reads the object matrix and sets up an object transform if animated. */ void setupObjectTransform(const float time); diff --git a/source/blender/alembic/intern/abc_points.cc b/source/blender/alembic/intern/abc_points.cc index d5ecf16b793..fc84759b1d9 100644 --- a/source/blender/alembic/intern/abc_points.cc +++ b/source/blender/alembic/intern/abc_points.cc @@ -151,12 +151,12 @@ bool AbcPointsReader::valid() const return m_schema.valid(); } -void AbcPointsReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel) +void AbcPointsReader::readObjectData(Main *bmain, float time) { Mesh *mesh = BKE_mesh_add(bmain, m_data_name.c_str()); DerivedMesh *dm = CDDM_from_mesh(mesh); - DerivedMesh *ndm = this->read_derivedmesh(dm, sample_sel, 0, NULL); + DerivedMesh *ndm = this->read_derivedmesh(dm, time, 0, NULL); if (ndm != dm) { dm->release(dm); @@ -178,7 +178,8 @@ void AbcPointsReader::readObjectData(Main *bmain, const Alembic::Abc::ISampleSel void read_points_sample(const IPointsSchema &schema, const ISampleSelector &selector, - CDStreamConfig &config) + CDStreamConfig &config, + float time) { Alembic::AbcGeom::IPointsSchema::Sample sample = schema.getValue(selector); @@ -188,7 +189,7 @@ void read_points_sample(const IPointsSchema &schema, N3fArraySamplePtr vnormals; if (has_property(prop, "N")) { - const Alembic::Util::uint32_t itime = static_cast<Alembic::Util::uint32_t>(selector.getRequestedTime()); + const Alembic::Util::uint32_t itime = static_cast<Alembic::Util::uint32_t>(time); const IN3fArrayProperty &normals_prop = IN3fArrayProperty(prop, "N", itime); if (normals_prop) { @@ -199,11 +200,9 @@ void read_points_sample(const IPointsSchema &schema, read_mverts(config.mvert, positions, vnormals); } -DerivedMesh *AbcPointsReader::read_derivedmesh(DerivedMesh *dm, - const ISampleSelector &sample_sel, - int /*read_flag*/, - const char ** /*err_str*/) +DerivedMesh *AbcPointsReader::read_derivedmesh(DerivedMesh *dm, const float time, int /*read_flag*/, const char ** /*err_str*/) { + ISampleSelector sample_sel(time); const IPointsSchema::Sample sample = m_schema.getValue(sample_sel); const P3fArraySamplePtr &positions = sample.getPositions(); @@ -215,7 +214,7 @@ DerivedMesh *AbcPointsReader::read_derivedmesh(DerivedMesh *dm, } CDStreamConfig config = get_config(new_dm ? new_dm : dm); - read_points_sample(m_schema, sample_sel, config); + read_points_sample(m_schema, sample_sel, config, time); return new_dm ? new_dm : dm; } diff --git a/source/blender/alembic/intern/abc_points.h b/source/blender/alembic/intern/abc_points.h index 1f74351d302..cb68dbca4d5 100644 --- a/source/blender/alembic/intern/abc_points.h +++ b/source/blender/alembic/intern/abc_points.h @@ -59,16 +59,14 @@ public: bool valid() const; - void readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel); + void readObjectData(Main *bmain, float time); - DerivedMesh *read_derivedmesh(DerivedMesh *dm, - const Alembic::Abc::ISampleSelector &sample_sel, - int read_flag, - const char **err_str); + DerivedMesh *read_derivedmesh(DerivedMesh *dm, const float time, int read_flag, const char **err_str); }; void read_points_sample(const Alembic::AbcGeom::IPointsSchema &schema, const Alembic::AbcGeom::ISampleSelector &selector, - CDStreamConfig &config); + CDStreamConfig &config, + float time); #endif /* __ABC_POINTS_H__ */ diff --git a/source/blender/alembic/intern/abc_transform.cc b/source/blender/alembic/intern/abc_transform.cc index c48be361877..6e218cac429 100644 --- a/source/blender/alembic/intern/abc_transform.cc +++ b/source/blender/alembic/intern/abc_transform.cc @@ -36,7 +36,6 @@ extern "C" { using Alembic::AbcGeom::OObject; using Alembic::AbcGeom::OXform; -using Alembic::Abc::ISampleSelector; /* ************************************************************************** */ @@ -154,7 +153,7 @@ bool AbcEmptyReader::valid() const return m_schema.valid(); } -void AbcEmptyReader::readObjectData(Main *bmain, const ISampleSelector &UNUSED(sample_sel)) +void AbcEmptyReader::readObjectData(Main *bmain, float /*time*/) { m_object = BKE_object_add_only_object(bmain, OB_EMPTY, m_object_name.c_str()); diff --git a/source/blender/alembic/intern/abc_transform.h b/source/blender/alembic/intern/abc_transform.h index f2e2edf8010..59388e155dc 100644 --- a/source/blender/alembic/intern/abc_transform.h +++ b/source/blender/alembic/intern/abc_transform.h @@ -69,7 +69,7 @@ public: bool valid() const; - void readObjectData(Main *bmain, const Alembic::Abc::ISampleSelector &sample_sel); + void readObjectData(Main *bmain, float time); }; #endif /* __ABC_TRANSFORM_H__ */ diff --git a/source/blender/alembic/intern/alembic_capi.cc b/source/blender/alembic/intern/alembic_capi.cc index 058c047e7d5..4aa4b8d548f 100644 --- a/source/blender/alembic/intern/alembic_capi.cc +++ b/source/blender/alembic/intern/alembic_capi.cc @@ -613,6 +613,7 @@ static std::pair<bool, AbcObjectReader *> visit_object( enum { ABC_NO_ERROR = 0, ABC_ARCHIVE_FAIL, + ABC_UNSUPPORTED_HDF5, }; struct ImportJobData { @@ -678,8 +679,12 @@ static void import_startjob(void *user_data, short *stop, short *do_update, floa ArchiveReader *archive = new ArchiveReader(data->filename); if (!archive->valid()) { - delete archive; +#ifndef WITH_ALEMBIC_HDF5 + data->error_code = archive->is_hdf5() ? ABC_UNSUPPORTED_HDF5 : ABC_ARCHIVE_FAIL; +#else data->error_code = ABC_ARCHIVE_FAIL; +#endif + delete archive; return; } @@ -723,13 +728,12 @@ static void import_startjob(void *user_data, short *stop, short *do_update, floa chrono_t min_time = std::numeric_limits<chrono_t>::max(); chrono_t max_time = std::numeric_limits<chrono_t>::min(); - ISampleSelector sample_sel(0.0f); std::vector<AbcObjectReader *>::iterator iter; for (iter = data->readers.begin(); iter != data->readers.end(); ++iter) { AbcObjectReader *reader = *iter; if (reader->valid()) { - reader->readObjectData(data->bmain, sample_sel); + reader->readObjectData(data->bmain, 0.0f); min_time = std::min(min_time, reader->minTime()); max_time = std::max(max_time, reader->maxTime()); @@ -861,6 +865,9 @@ static void import_endjob(void *user_data) case ABC_ARCHIVE_FAIL: WM_report(RPT_ERROR, "Could not open Alembic archive for reading! See console for detail."); break; + case ABC_UNSUPPORTED_HDF5: + WM_report(RPT_ERROR, "Alembic archive in obsolete HDF5 format is not supported."); + break; } WM_main_add_notifier(NC_SCENE | ND_FRAME, data->scene); @@ -955,7 +962,6 @@ DerivedMesh *ABC_read_mesh(CacheReader *reader, } const ObjectHeader &header = iobject.getHeader(); - ISampleSelector sample_sel(time); if (IPolyMesh::matches(header)) { if (ob->type != OB_MESH) { @@ -963,7 +969,7 @@ DerivedMesh *ABC_read_mesh(CacheReader *reader, return NULL; } - return abc_reader->read_derivedmesh(dm, sample_sel, read_flag, err_str); + return abc_reader->read_derivedmesh(dm, time, read_flag, err_str); } else if (ISubD::matches(header)) { if (ob->type != OB_MESH) { @@ -971,7 +977,7 @@ DerivedMesh *ABC_read_mesh(CacheReader *reader, return NULL; } - return abc_reader->read_derivedmesh(dm, sample_sel, read_flag, err_str); + return abc_reader->read_derivedmesh(dm, time, read_flag, err_str); } else if (IPoints::matches(header)) { if (ob->type != OB_MESH) { @@ -979,7 +985,7 @@ DerivedMesh *ABC_read_mesh(CacheReader *reader, return NULL; } - return abc_reader->read_derivedmesh(dm, sample_sel, read_flag, err_str); + return abc_reader->read_derivedmesh(dm, time, read_flag, err_str); } else if (ICurves::matches(header)) { if (ob->type != OB_CURVE) { @@ -987,7 +993,7 @@ DerivedMesh *ABC_read_mesh(CacheReader *reader, return NULL; } - return abc_reader->read_derivedmesh(dm, sample_sel, read_flag, err_str); + return abc_reader->read_derivedmesh(dm, time, read_flag, err_str); } *err_str = "Unsupported object type: verify object path"; // or poke developer diff --git a/source/blender/editors/mesh/editface.c b/source/blender/editors/mesh/editface.c index c4e87614732..a478526dee0 100644 --- a/source/blender/editors/mesh/editface.c +++ b/source/blender/editors/mesh/editface.c @@ -306,31 +306,30 @@ void paintface_deselect_all_visible(Object *ob, int action, bool flush_flags) bool paintface_minmax(Object *ob, float r_min[3], float r_max[3]) { - Mesh *me; - MPoly *mp; - MTexPoly *tf; - MLoop *ml; - MVert *mvert; + const Mesh *me; + const MPoly *mp; + const MLoop *ml; + const MVert *mvert; int a, b; bool ok = false; float vec[3], bmat[3][3]; me = BKE_mesh_from_object(ob); - if (!me || !me->mtpoly) return ok; + if (!me || !me->mloopuv) { + return ok; + } copy_m3_m4(bmat, ob->obmat); mvert = me->mvert; mp = me->mpoly; - tf = me->mtpoly; - for (a = me->totpoly; a > 0; a--, mp++, tf++) { + for (a = me->totpoly; a > 0; a--, mp++) { if (mp->flag & ME_HIDE || !(mp->flag & ME_FACE_SEL)) continue; ml = me->mloop + mp->totloop; for (b = 0; b < mp->totloop; b++, ml++) { - copy_v3_v3(vec, (mvert[ml->v].co)); - mul_m3_v3(bmat, vec); + mul_v3_m3v3(vec, bmat, mvert[ml->v].co); add_v3_v3v3(vec, vec, ob->obmat[3]); minmax_v3v3_v3(r_min, r_max, vec); } diff --git a/source/blender/editors/uvedit/uvedit_draw.c b/source/blender/editors/uvedit/uvedit_draw.c index 3149abc338b..16d430c9421 100644 --- a/source/blender/editors/uvedit/uvedit_draw.c +++ b/source/blender/editors/uvedit/uvedit_draw.c @@ -553,7 +553,7 @@ static void draw_uvs_texpaint(SpaceImage *sima, Scene *scene, SceneLayer *sl, Ob ma = give_current_material(ob, ob->actcol); - if (me->mtpoly) { + if (me->mloopuv) { MPoly *mpoly = me->mpoly; MLoopUV *mloopuv, *mloopuv_base; int a, b; diff --git a/source/blender/makesdna/DNA_cachefile_types.h b/source/blender/makesdna/DNA_cachefile_types.h index 46b1adf2725..e6fe8102b4f 100644 --- a/source/blender/makesdna/DNA_cachefile_types.h +++ b/source/blender/makesdna/DNA_cachefile_types.h @@ -50,7 +50,7 @@ enum { typedef struct AlembicObjectPath { struct AlembicObjectPath *next, *prev; - char path[1024]; /* 1024 = FILE_MAX, might use PATH_MAX in the future. */ + char path[4096]; /* 4096 = PATH_MAX */ } AlembicObjectPath; typedef struct CacheFile { @@ -64,7 +64,7 @@ typedef struct CacheFile { * CacheFile. */ ListBase object_paths; - char filepath[1024]; /* 1024 = FILE_MAX */ + char filepath[4096]; /* 4096 = PATH_MAX */ char is_sequence; char forward_axis; diff --git a/source/blender/windowmanager/intern/wm_files.c b/source/blender/windowmanager/intern/wm_files.c index f77b6c92029..99e0e28ab1b 100644 --- a/source/blender/windowmanager/intern/wm_files.c +++ b/source/blender/windowmanager/intern/wm_files.c @@ -725,7 +725,7 @@ int wm_homefile_read( if (filepath_startup_override != NULL) { /* pass */ } - else if (app_template_override) { + else if (app_template_override && app_template_override[0]) { app_template = app_template_override; } else if (!use_factory_settings && U.app_template[0]) { diff --git a/tests/python/alembic_tests.py b/tests/python/alembic_tests.py index 1cdfd75426a..2a2dd0230cf 100755 --- a/tests/python/alembic_tests.py +++ b/tests/python/alembic_tests.py @@ -372,6 +372,95 @@ class HairParticlesExportTest(AbstractAlembicTest): self.assertIn('.faceIndices', abcprop) +class LongNamesExportTest(AbstractAlembicTest): + @with_tempdir + def test_export_long_names(self, tempdir: pathlib.Path): + abc = tempdir / 'long-names.abc' + script = "import bpy; bpy.ops.wm.alembic_export(filepath='%s', start=1, end=1, " \ + "renderable_only=False, visible_layers_only=False, flatten=False)" % abc + self.run_blender('long-names.blend', script) + + name_parts = [ + 'foG9aeLahgoh5goacee1dah6Hethaghohjaich5pasizairuWigee1ahPeekiGh', + 'yoNgoisheedah2ua0eigh2AeCaiTee5bo0uphoo7Aixephah9racahvaingeeH4', + 'zuthohnoi1thooS3eezoo8seuph2Boo5aefacaethuvee1aequoonoox1sookie', + 'wugh4ciTh3dipiepeequait5uug7thiseek5ca7Eijei5ietaizokohhaecieto', + 'up9aeheenein9oteiX6fohP3thiez6Ahvah0oohah1ep2Eesho4Beboechaipoh', + 'coh4aehiacheTh0ue0eegho9oku1lohl4loht9ohPoongoow7dasiego6yimuis', + 'lohtho8eigahfeipohviepajaix4it2peeQu6Iefee1nevihaes4cee2soh4noy', + 'kaht9ahv0ieXaiyih7ohxe8bah7eeyicahjoa2ohbu7Choxua7oongah6sei4bu', + 'deif0iPaechohkee5nahx6oi2uJeeN7ze3seunohJibe4shai0mah5Iesh3Quai', + 'ChohDahshooNee0NeNohthah0eiDeese3Vu6ohShil1Iey9ja0uebi2quiShae6', + 'Dee1kai7eiph2ahh2nufah3zai3eexeengohQue1caj0eeW0xeghi3eshuadoot', + 'aeshiup3aengajoog0AhCoo5tiu3ieghaeGhie4Tu1ohh1thee8aepheingah1E', + 'ooRa6ahciolohshaifoopeo9ZeiGhae2aech4raisheiWah9AaNga0uas9ahquo', + 'thaepheip2aip6shief4EaXopei8ohPo0ighuiXah2ashowai9nohp4uach6Mei', + 'ohph4yaev3quieji3phophiem3OoNuisheepahng4waithae3Naichai7aw3noo', + 'aibeawaneBahmieyuph8ieng8iopheereeD2uu9Uyee5bei2phahXeir8eeJ8oo', + 'ooshahphei2hoh3uth5chaen7ohsai6uutiesucheichai8ungah9Gie1Aiphie', + 'eiwohchoo7ere2iebohn4Aapheichaelooriiyaoxaik7ooqua7aezahx0aeJei', + 'Vah0ohgohphiefohTheshieghichaichahch5moshoo0zai5eeva7eisi4yae8T', + 'EibeeN0fee0Gohnguz8iec6yeigh7shuNg4eingu3siph9joucahpeidoom4ree', + 'iejiu3shohheeZahHusheimeefaihoh5eecachu5eeZie9ceisugu9taidohT3U', + 'eex6dilakaix5Eetai7xiCh5Jaa8aiD4Ag3tuij1aijohv5fo0heevah8hohs3m', + 'ohqueeNgahraew6uraemohtoo5qua3oojiex6ohqu6Aideibaithaiphuriquie', + 'cei0eiN4Shiey7Aeluy3unohboo5choiphahc2mahbei5paephaiKeso1thoog1', + 'ieghif4ohKequ7ong0jah5ooBah0eiGh1caechahnahThae9Shoo0phopashoo4', + 'roh9er3thohwi5am8iequeequuSh3aic0voocai3ihi5nie2abahphupiegh7vu', + 'uv3Quei7wujoo5beingei2aish5op4VaiX0aebai7iwoaPee5pei8ko9IepaPig', + 'co7aegh5beitheesi9lu7jeeQu3johgeiphee9cheichi8aithuDehu2gaeNein', + 'thai3Tiewoo4nuir1ohy4aithiuZ7shae1luuwei5phibohriepe2paeci1Ach8', + 'phoi3ribah7ufuvoh8eigh1oB6deeBaiPohphaghiPieshahfah5EiCi3toogoo', + 'aiM8geil7ooreinee4Cheiwea4yeec8eeshi7Sei4Shoo3wu6ohkaNgooQu1mai', + 'agoo3faciewah9ZeesiXeereek7am0eigaeShie3Tisu8haReeNgoo0ci2Hae5u', + 'Aesatheewiedohshaephaenohbooshee8eu7EiJ8isal1laech2eiHo0noaV3ta', + 'liunguep3ooChoo4eir8ahSie8eenee0oo1TooXu8Cais8Aimo4eir6Phoo3xei', + 'toe9heepeobein3teequachemei0Cejoomef9ujie3ohwae9AiNgiephi3ep0de', + 'ua6xooY9uzaeB3of6sheiyaedohoiS5Eev0Aequ9ahm1zoa5Aegh3ooz9ChahDa', + 'eevasah6Bu9wi7EiwiequumahkaeCheegh6lui8xoh4eeY4ieneavah8phaibun', + 'AhNgei2sioZeeng6phaecheemeehiShie5eFeiTh6ooV8iiphabud0die4siep4', + 'kushe6Xieg6ahQuoo9aex3aipheefiec1esa7OhBuG0ueziep9phai5eegh1vie', + 'Jie5yu8aafuQuoh9shaep3moboh3Pooy7och8oC6obeik6jaew2aiLooweib3ch', + 'ohohjajaivaiRail3odaimei6aekohVaicheip2wu7phieg5Gohsaing2ahxaiy', + 'hahzaht6yaiYu9re9jah9loisiit4ahtoh2quoh9xohishioz4oo4phofu3ogha', + 'pu4oorea0uh2tahB8aiZoonge1aophaes6ogaiK9ailaigeej4zoVou8ielotee', + 'cae2thei3Luphuqu0zeeG8leeZuchahxaicai4ui4Eedohte9uW6gae8Geeh0ea', + 'air7tuy7ohw5sho2Tahpai8aep4so5ria7eaShus5weaqu0Naquei2xaeyoo2ae', + 'vohge4aeCh7ahwoo7Jaex6sohl0Koong4Iejisei8Coir0iemeiz9uru9Iebaep', + 'aepeidie8aiw6waish9gie4Woolae2thuj5phae4phexux7gishaeph4Deu7ooS', + 'vahc5ia0xohHooViT0uyuxookiaquu2ogueth0ahquoudeefohshai8aeThahba', + 'mun3oagah2eequaenohfoo8DaigeghoozaV2eiveeQuee7kah0quaa6tiesheet', + 'ooSet4IdieC4ugow3za0die4ohGoh1oopoh6luaPhaeng4Eechea1hae0eimie5', + 'iedeimadaefu2NeiPaey2jooloov5iehiegeakoo4ueso7aeK9ahqu2Thahkaes', + 'nahquah9Quuu2uuf0aJah7eishi2siegh8ue5eiJa2EeVu8ebohkepoh4dahNgo', + 'io1bie7chioPiej5ae2oohe2fee6ooP2thaeJohjohb9Se8tang3eipaifeimai', + 'oungoqu6dieneejiechez1xeD2Zi9iox2Ahchaiy9ithah3ohVoolu2euQuuawo', + 'thaew0veigei4neishohd8mecaixuqu7eeshiex1chaigohmoThoghoitoTa0Eo', + 'ahroob2phohvaiz0Ohteik2ohtakie6Iu1vitho8IyiyeeleeShae9defaiw9ki', + 'DohHoothohzeaxolai3Toh5eJie7ahlah9reF0ohn1chaipoogain2aibahw4no', + 'aif8lo5she4aich5cho2rie8ieJaujeem2Joongeedae4vie3tah1Leequaix1O', + 'Aang0Shaih6chahthie1ahZ7aewei9thiethee7iuThah3yoongi8ahngiobaa5', + 'iephoBuayoothah0Ru6aichai4aiw8deg1umongauvaixai3ohy6oowohlee8ei', + 'ohn5shigoameer0aejohgoh8oChohlaecho9jie6shu0ahg9Bohngau6paevei9', + 'edahghaishak0paigh1eecuich3aad7yeB0ieD6akeeliem2beifufaekee6eat', + 'hiechahgheloh2zo7Ieghaiph0phahhu8aeyuiKie1xeipheech9zai4aeme0ee', + 'Cube' + ] + name = '/' + '/'.join(name_parts) + + # Now check the resulting Alembic file. + abcprop = self.abcprop(abc, '%s/.xform' % name) + self.assertEqual(abcprop['.vals'], [ + 1.0, 0.0, 0.0, 0.0, + 0.0, 1.0, 0.0, 0.0, + 0.0, 0.0, 1.0, 0.0, + 0.0, 3.0, 0.0, 1.0, + ]) + + abcprop = self.abcprop(abc, '%s/CubeShape/.geom' % name) + self.assertIn('.faceCounts', abcprop) + + if __name__ == '__main__': parser = argparse.ArgumentParser() parser.add_argument('--blender', required=True) diff --git a/tests/python/bl_alembic_import_test.py b/tests/python/bl_alembic_import_test.py index e036157e59c..e83d38aec63 100644 --- a/tests/python/bl_alembic_import_test.py +++ b/tests/python/bl_alembic_import_test.py @@ -122,6 +122,15 @@ class SimpleImportTest(unittest.TestCase): self.assertAlmostEqual(y, 0) self.assertAlmostEqual(z, 0) + def test_import_long_names(self): + # This file contains very long names. The longest name is 4047 chars. + bpy.ops.wm.alembic_import( + filepath=str(self.testdir / "long-names.abc"), + as_background_job=False) + + self.assertIn('Cube', bpy.data.objects) + self.assertEqual('CubeShape', bpy.data.objects['Cube'].data.name) + def main(): global args |