Home
last modified time | relevance | path

Searched refs:__device__ (Results 1 – 25 of 56) sorted by relevance

123

/third_party/ffmpeg/libavfilter/cuda/
Dvector_helpers.cuh43 template<typename T, typename V> inline __device__ V to_floatN(const T &a) { return (V)a; } in to_floatN()
44 template<typename T, typename V> inline __device__ T from_floatN(const V &a) { return (T)a; } in from_floatN()
47 …template<typename V> inline __device__ T operator+(const T &a, const V &b) { return make_ ## T (a.…
48 …template<typename V> inline __device__ T operator-(const T &a, const V &b) { return make_ ## T (a.…
49 …template<typename V> inline __device__ T operator*(const T &a, V b) { return make_ ## T (a.x * b, …
50 …template<typename V> inline __device__ T operator/(const T &a, V b) { return make_ ## T (a.x / b, …
51 …template<typename V> inline __device__ T operator>>(const T &a, V b) { return make_ ## T (a.x >> b…
52 …template<typename V> inline __device__ T operator<<(const T &a, V b) { return make_ ## T (a.x << b…
53 …template<typename V> inline __device__ T &operator+=(T &a, const V &b) { a.x += b.x; a.y += b.y; r…
54 … template<typename V> inline __device__ void vec_set(T &a, const V &b) { a.x = b.x; a.y = b.y; } \
[all …]
/third_party/ffmpeg/compat/cuda/
Dcuda_runtime.h28 #define __device__ __attribute__((device)) macro
94 #define GET(name, reg) static inline __device__ uint3 name() {\
123 #define TEX2D(type, ret) static inline __device__ void conv(type* out, unsigned a, unsigned b, unsi…
137 inline __device__ T tex2D(cudaTextureObject_t texObject, float x, float y)
149 inline __device__ float4 tex2D<float4>(cudaTextureObject_t texObject, float x, float y)
159 inline __device__ float tex2D<float>(cudaTextureObject_t texObject, float x, float y)
165 inline __device__ float2 tex2D<float2>(cudaTextureObject_t texObject, float x, float y)
172 static inline __device__ float floorf(float a) { return __builtin_floorf(a); } in floorf()
173 static inline __device__ float floor(float a) { return __builtin_floorf(a); } in floor()
174 static inline __device__ double floor(double a) { return __builtin_floor(a); } in floor()
[all …]
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/
Dutil.cuh31 __device__ static inline double MsAtomicAdd(double *address, const double val) { in MsAtomicAdd()
42 __device__ static inline float MsAtomicAdd(float *address, const float val) { return atomicAdd(addr… in MsAtomicAdd()
44 __device__ static inline int MsAtomicAdd(int *address, int val) { return atomicAdd(address, val); } in MsAtomicAdd()
46 __device__ static inline unsigned int MsAtomicAdd(unsigned int *address, unsigned int val) { in MsAtomicAdd()
50 __device__ static inline int8_t MsAtomicAdd(int8_t *address, int8_t val) { in MsAtomicAdd()
69 __device__ static inline int64_t MsAtomicAdd(int64_t *address, int64_t val) { in MsAtomicAdd()
83 __device__ static inline bool MsAtomicAdd(bool *address, bool val) { in MsAtomicAdd()
88 __device__ static inline unsigned char MsAtomicAdd(short *address, short val) { // NOLINT in MsAtomicAdd()
114 __device__ static inline half MsAtomicAdd(half *address, half val) { in MsAtomicAdd()
134 __device__ static inline unsigned char MsAtomicAdd(unsigned char *address, unsigned char val) { in MsAtomicAdd()
[all …]
Dsparse_apply_proximal_adagrad_impl.cu20 __device__ __forceinline__ bool CompareFunc(T x, T y) { in CompareFunc()
25 __device__ __forceinline__ bool CompareFunc(half x, half y) { in CompareFunc()
30 __device__ __forceinline__ T RsqrtFunc(T x) { in RsqrtFunc()
35 __device__ __forceinline__ half RsqrtFunc(half x) { in RsqrtFunc()
40 __device__ __forceinline__ T AbsFunc(T x) { in AbsFunc()
45 __device__ __forceinline__ half AbsFunc(half x) { in AbsFunc()
50 __device__ __forceinline__ T Sgn(T x) { in Sgn()
55 __device__ __forceinline__ half Sgn(half x) { in Sgn()
Dbroadcast_impl.cu26 …__device__ __host__ __forceinline__ bool operator()(const T &lhs, const T &rhs) { return lhs > rhs… in operator ()()
31 …__device__ __host__ __forceinline__ bool operator()(const T &lhs, const T &rhs) { return lhs < rhs… in operator ()()
36 …__device__ __host__ __forceinline__ bool operator()(const T &lhs, const T &rhs) { return lhs == rh… in operator ()()
41 __device__ __host__ __forceinline__ bool operator()(const half &lhs, const half &rhs) { in operator ()()
48 __device__ __host__ __forceinline__ bool operator()(const float &lhs, const float &rhs) { in operator ()()
55 …__device__ __host__ __forceinline__ bool operator()(const T &lhs, const T &rhs) { return lhs >= rh… in operator ()()
60 __device__ __host__ __forceinline__ bool operator()(const half &lhs, const half &rhs) { in operator ()()
69 __device__ __host__ __forceinline__ bool operator()(const float &lhs, const float &rhs) { in operator ()()
76 …__device__ __host__ __forceinline__ bool operator()(const T &lhs, const T &rhs) { return lhs <= rh… in operator ()()
81 __device__ __host__ __forceinline__ bool operator()(const half &lhs, const half &rhs) { in operator ()()
[all …]
Dtopk_lib.cuh20 constexpr __host__ __device__ int Log2(int n, int p = 0) { return (n <= 1) ? p : Log2(n / 2, p + 1)… in Log2()
21 constexpr __host__ __device__ bool IsPow2(int v) { return (v && !(v & (v - 1))); } in IsPow2()
22 constexpr __host__ __device__ int NextPow2(int v) { return (IsPow2(v) ? 2 * v : (1 << static_cast<i… in NextPow2()
24 __device__ __forceinline__ int GetLaneId() { in GetLaneId()
32 …__device__ static inline bool gt(T k1, S v1, T k2, S v2) { return k1 > k2 || (k1 == k2 && v1 < v2)… in gt()
33 …__device__ static inline bool lt(T k1, S v1, T k2, S v2) { return k1 < k2 || (k1 == k2 && v1 > v2)… in lt()
38 __device__ static inline bool lt(T a, T b) { return a < b; } in lt()
39 __device__ static inline bool gt(T a, T b) { return a > b; } in gt()
43 inline __device__ T shfl_xor(const T val, int laneMask, int width = kWarpSize) { in shfl_xor()
48 inline __device__ void L2CompareAndSwap(T *a, S *b, int i_1, int i_2) { in L2CompareAndSwap()
[all …]
Dftrl_impl.cu20 __device__ __forceinline__ T PowFunc(T x, T y) { in PowFunc()
25 __device__ __forceinline__ half PowFunc(half x, half y) { in PowFunc()
30 __device__ __forceinline__ bool CompareFunc(T x, T y) { in CompareFunc()
35 __device__ __forceinline__ bool CompareFunc(half x, half y) { in CompareFunc()
40 __device__ __forceinline__ T Sgn(T x) { in Sgn()
45 __device__ __forceinline__ half Sgn(half x) { in Sgn()
Dsparse_ftrl_impl.cu22 __device__ __forceinline__ T PowFunc(T x, T y) { in PowFunc()
27 __device__ __forceinline__ half PowFunc(half x, half y) { in PowFunc()
32 __device__ __forceinline__ bool CompareFunc(T x, T y) { in CompareFunc()
37 __device__ __forceinline__ bool CompareFunc(half x, half y) { in CompareFunc()
42 __device__ __forceinline__ T Sgn(T x) { in Sgn()
47 __device__ __forceinline__ half Sgn(half x) { in Sgn()
Dlayer_norm_grad_impl.cu27 inline __device__ T my_pow(T a, double b) { in my_pow()
32 inline __device__ half my_pow(half a, double b) { in my_pow()
37 inline __device__ void GammaAndBetaThreadReduce(const int &col, const int &row_dim, const int &col_… in GammaAndBetaThreadReduce()
57 inline __device__ void GammaAndBetaWarpReduce(T *dg, T *db) { in GammaAndBetaWarpReduce()
65 inline __device__ void GammaAndBetaBlockReduce(const int &col, const int &row_dim, T *dg, T *db, T … in GammaAndBetaBlockReduce()
110 inline __device__ void InputThreadReduce(const int &row, const int &col_dim, const int &param_dim, … in InputThreadReduce()
134 inline __device__ void InputThreadReduce(const int &row, const int &col_dim, const int &param_dim, … in InputThreadReduce()
158 inline __device__ void InputWarpReduce(T *sum1, T *sum2, T *sum3) { in InputWarpReduce()
167 inline __device__ void InputBlockReduce(const int &col_dim, T *sum1, T *sum2, T *sum3, T *share_mem… in InputBlockReduce()
190 inline __device__ void InputProp(const int &row, const int &col_dim, const int &param_dim, const T … in InputProp()
[all …]
Dlayer_norm_impl.cu26 inline __device__ void MeanAndVarAccumulation(T *mean, T *var, T *num, const T &val) { in MeanAndVarAccumulation()
37 inline __device__ void MeanAndVarMerge(T *m1, T *v1, T *n1, const T &m2, const T &v2, const T &n2) { in MeanAndVarMerge()
50 inline __device__ void ThreadReduce(const int &col_dim, const T *block_addr, T *mean, T *var, T *nu… in ThreadReduce()
64 inline __device__ void WarpReduce(T *mean, T *var, T *num) { in WarpReduce()
74 inline __device__ void BlockReduce(const int &col_dim, T *mean, T *var, T *num, T *mean_addr, T *va… in BlockReduce()
103 inline __device__ void LayerNorm(const int &row, const int &col_dim, const int &param_dim, const T … in LayerNorm()
113 inline __device__ void LayerNorm(const int &row, const int &col_dim, const int &param_dim, const ha… in LayerNorm()
Dlayer_norm_grad_grad_impl.cu30 inline __device__ T my_pow(T a, double b) { in my_pow()
36 inline __device__ half my_pow(half a, double b) { in my_pow()
42 inline __device__ void GammaAndBetaThreadReduce(const int &col, const int &row_dim, const int &col_… in GammaAndBetaThreadReduce()
69 inline __device__ void GammaAndBetaWarpReduce(T *part1, T *part2, T *part3) { in GammaAndBetaWarpReduce()
79 inline __device__ void GammaAndBetaBlockReduce(const int &col, const int &row_dim, T *part1, T *par… in GammaAndBetaBlockReduce()
125 inline __device__ void InputThreadReduceInnerMean(const int &row, const int &col_dim, const int &pa… in InputThreadReduceInnerMean()
154 inline __device__ void InputWarpReduceInnerMean(T *sum1, T *sum2, T *sum3, T *sum4) { in InputWarpReduceInnerMean()
165 inline __device__ void InputBlockReduceInnerMean(const int &col_dim, T *sum1, T *sum2, T *sum3, T *… in InputBlockReduceInnerMean()
192 inline __device__ void InputThreadReduceOuterMean(const int &row, const int &col_dim, const int &pa… in InputThreadReduceOuterMean()
230 inline __device__ void InputThreadReduceOuterMean(const int &row, const int &col_dim, const int &pa… in InputThreadReduceOuterMean()
[all …]
Dcast_impl.cu25 __device__ __forceinline__ void CastBase(const S *input_addr, T *output_addr) { in CastBase()
30 __device__ __forceinline__ void CastBase(const half *input_addr, uint64_t *output_addr) { in CastBase()
34 __device__ __forceinline__ void CastBase(const half *input_addr, int64_t *output_addr) { in CastBase()
38 __device__ __forceinline__ void CastBase(const half *input_addr, uint32_t *output_addr) { in CastBase()
42 __device__ __forceinline__ void CastBase(const half *input_addr, int32_t *output_addr) { in CastBase()
46 __device__ __forceinline__ void CastBase(const half *input_addr, uint16_t *output_addr) { in CastBase()
50 __device__ __forceinline__ void CastBase(const half *input_addr, int16_t *output_addr) { in CastBase()
54 __device__ __forceinline__ void CastBase(const half *input_addr, uint8_t *output_addr) { in CastBase()
58 __device__ __forceinline__ void CastBase(const half *input_addr, int8_t *output_addr) { in CastBase()
63 __device__ __forceinline__ void CastBase(const uint64_t *input_addr, half *output_addr) { in CastBase()
[all …]
Dlayer_norm_impl.cuh26 __device__ float *addr() { in addr()
33 __device__ half *addr() { in addr()
Dbroadcast_grad_impl.cu23 …__device__ __forceinline__ void operator()(const T &x1, const T &x2, const bool &grad_x1, const bo… in operator ()()
35 …__device__ __forceinline__ void operator()(const T &x1, const T &x2, const bool &grad_x1, const bo… in operator ()()
45 __device__ __forceinline__ int Index(const int &index, const int &dim) { return dim == 1 ? 0 : inde… in Index()
48 __device__ __forceinline__ void BroadcastGradOperator(const int &l0, const int &l1, const int &l2, … in BroadcastGradOperator()
91 __device__ __forceinline__ void NoBroadcastOperator(const int &nums, const bool &grad_x1, const boo… in NoBroadcastOperator()
Droi_align_impl.cu21 inline __device__ int roi_cast_int(float x) { return __float2int_rd(x); } in roi_cast_int()
22 inline __device__ int roi_cast_int(half x) { return __half2int_rd(x); } in roi_cast_int()
23 inline __device__ int roi_round_int(float x) { return __float2int_rn(x + 0.00007); } in roi_round_int()
24 inline __device__ int roi_round_int(half x) { return __half2int_rn(x + static_cast<half>(0.00007));… in roi_round_int()
27 __device__ void bilinear_interpolate(const int height, const int width, T y, T x, int *x_low, int *… in bilinear_interpolate()
71 __device__ void bin_box(int thread_idx, const T *roi_boxes, int roi_cols, const T spatial_scale, co… in bin_box()
Dloss_with_reduction_impl.cu22 inline __device__ float logT(float x) { return logf(x); } in logT()
23 inline __device__ half logT(half x) { return hlog(x); } in logT()
24 inline __device__ float castT(float ref, int x) { return __int2float_rd(x); } in castT()
25 inline __device__ half castT(half ref, int x) { return __int2half_rd(x); } in castT()
26 inline __device__ float maxT(float a, float b) { return fmaxf(a, b); } in maxT()
27 inline __device__ half maxT(half a, half b) { return a > b ? a : b; } in maxT()
57 __device__ void MultiplyDevice(const S a, const T b, T *out) { in MultiplyDevice()
62 __device__ void MultiplyDevice(const half a, const float b, float *out) { in MultiplyDevice()
69 __device__ void MultiplyDevice(const float a, const half b, half *out) { in MultiplyDevice()
Diou_impl.cu19 __device__ float CoordinateMax(const float a, const float b) { in CoordinateMax()
23 __device__ float CoordinateMin(const float a, const float b) { in CoordinateMin()
Dunsorted_segment_min.cu21 __device__ __forceinline__ void max_val_init(T *init_val) { in max_val_init()
26 __device__ __forceinline__ void max_val_init(half *init_val) { in max_val_init()
Dsync_batch_norm_grad_impl.cu25 __inline__ __device__ float HalfFloatInputConvert(const half val) { return __half2float(val); } in HalfFloatInputConvert()
26 __inline__ __device__ float HalfFloatInputConvert(const float val) { return val; } in HalfFloatInputConvert()
27 __inline__ __device__ void HalfFloatOutputAssign(const float val, float *arr, int idx) { arr[idx] =… in HalfFloatOutputAssign()
28 __inline__ __device__ void HalfFloatOutputAssign(const float val, half *arr, int idx) { arr[idx] = … in HalfFloatOutputAssign()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sponge/
Dcommon_sponge.cuh125 __device__ __host__ static inline VECTOR operator-(const VECTOR &veca, const VECTOR &vecb) { in operator -()
133 __device__ __host__ static inline VECTOR Get_Periodic_Displacement(const UNSIGNED_INT_VECTOR uvec_a, in Get_Periodic_Displacement()
143 __device__ __host__ static inline VECTOR Get_Periodic_Displacement(const UINT_VECTOR_LJ_TYPE uvec_a, in Get_Periodic_Displacement()
153 __device__ __host__ static inline VECTOR Get_Periodic_Displacement(const VECTOR vec_a, const VECTOR… in Get_Periodic_Displacement()
163 __device__ __host__ static inline VECTOR Get_Periodic_Displacement(const VECTOR vec_a, const VECTOR… in Get_Periodic_Displacement()
174 __device__ __host__ static inline VECTOR operator+(const VECTOR &veca, const VECTOR &vecb) { in operator +()
182 __device__ __host__ static inline float operator*(const VECTOR &veca, const VECTOR &vecb) { in operator *()
185 __device__ __host__ static inline VECTOR operator*(const float &a, const VECTOR &vecb) { in operator *()
193 __device__ __host__ static inline VECTOR operator-(const VECTOR &vecb) { in operator -()
201 __device__ __host__ static inline VECTOR operator^(const VECTOR &veca, const VECTOR &vecb) { in operator ^()
[all …]
/third_party/boost/boost/numeric/odeint/external/thrust/
Dthrust_operations.hpp42 __host__ __device__
59 __host__ __device__
79 __host__ __device__
101 __host__ __device__
128 __host__ __device__
157 __host__ __device__
188 __host__ __device__
214 __host__ __device__
/third_party/ffmpeg/libavfilter/
Dvf_scale_cuda_bicubic.cu28 __device__ inline float4 lanczos_coeffs(float x, float param) in lanczos_coeffs()
50 __device__ inline float4 bicubic_coeffs(float x, float param) in bicubic_coeffs()
63 __device__ inline void derived_fast_coeffs(float4 coeffs, float x, float *h0, float *h1, float *s) in derived_fast_coeffs()
74 __device__ inline V apply_coeffs(float4 coeffs, V c0, V c1, V c2, V c3) in apply_coeffs()
85 __device__ inline void Subsample_Bicubic(coeffs_function_t coeffs_function, in Subsample_Bicubic()
128 __device__ inline void Subsample_FastBicubic(coeffs_function_t coeffs_function, in Subsample_FastBicubic()
Dvf_yadif_cuda.cu22 __inline__ __device__ T spatial_predictor(T a, T b, T c, T d, T e, T f, T g, in spatial_predictor()
51 __inline__ __device__ int max3(int a, int b, int c) in max3()
57 __inline__ __device__ int min3(int a, int b, int c) in min3()
64 __inline__ __device__ T temporal_predictor(T A, T B, T C, T D, T E, T F, in temporal_predictor()
97 __inline__ __device__ void yadif_single(T *dst, in yadif_single()
167 __inline__ __device__ void yadif_double(T *dst, in yadif_double()
Dvf_scale_cuda.cu26 __device__ inline void Subsample_Nearest(cudaTextureObject_t tex, in Subsample_Nearest()
47 __device__ inline void Subsample_Bilinear(cudaTextureObject_t tex, in Subsample_Bilinear()
/third_party/boost/libs/numeric/odeint/examples/thrust/
Dphase_oscillator_ensemble.cu60 __host__ __device__
69 __host__ __device__
114 __host__ __device__

123