FFmpeg: compat/cuda/cuda_runtime.h File Reference

FFmpeg

Data Structures | Macros | Typedefs | Functions | Variables

cuda_runtime.h File Reference

Go to the source code of this file.

Data Structures

struct dim3

Macros

#define __global__ __attribute__((global))

#define __device__ __attribute__((device))

#define __device_builtin__ __attribute__((device_builtin))

#define __align__(N) __attribute__((aligned(N)))

#define __inline__ __inline__ __attribute__((always_inline))

#define max(a, b) ((a) > (b) ? (a) : (b))

#define min(a, b) ((a) < (b) ? (a) : (b))

#define abs(x) ((x) < 0 ? -(x) : (x))

#define atomicAdd(a, b) (__atomic_fetch_add(a, b, __ATOMIC_SEQ_CST))

#define GETCOMP(reg, comp)

#define GET(name, reg)

#define blockIdx (getBlockIdx())

#define blockDim (getBlockDim())

#define threadIdx (getThreadIdx())

#define make_int2(a, b) ((int2){.x = a, .y = b})

#define make_uchar2(a, b) ((uchar2){.x = a, .y = b})

#define make_ushort2(a, b) ((ushort2){.x = a, .y = b})

#define make_float2(a, b) ((float2){.x = a, .y = b})

#define make_int4(a, b, c, d) ((int4){.x = a, .y = b, .z = c, .w = d})

#define make_uchar4(a, b, c, d) ((uchar4){.x = a, .y = b, .z = c, .w = d})

#define make_ushort4(a, b, c, d) ((ushort4){.x = a, .y = b, .z = c, .w = d})

#define make_float4(a, b, c, d) ((float4){.x = a, .y = b, .z = c, .w = d})

#define TEX2D(type, ret) static inline __device__ void conv(type* out, unsigned a, unsigned b, unsigned c, unsigned d) {*out = (ret);}

Typedefs

typedef __device_builtin__ unsigned long long cudaTextureObject_t

Functions

struct __device_builtin__ __align__ (2) uchar2

struct __device_builtin__ __align__ (4) ushort2

struct __device_builtin__ __align__ (8) float2

struct __device_builtin__ __align__ (16) int4

template<typename T >

__device__ T tex2D (cudaTextureObject_t texObject, float x, float y)

template<>

__device__ float4 tex2D< float4 > (cudaTextureObject_t texObject, float x, float y)

template<>

__device__ float tex2D< float > (cudaTextureObject_t texObject, float x, float y)

template<>

__device__ float2 tex2D< float2 > (cudaTextureObject_t texObject, float x, float y)

static __device__ float floorf (float a)

static __device__ float floor (float a)

static __device__ double floor (double a)

static __device__ float ceilf (float a)

static __device__ float ceil (float a)

static __device__ double ceil (double a)

static __device__ float truncf (float a)

static __device__ float trunc (float a)

static __device__ double trunc (double a)

static __device__ float fabsf (float a)

static __device__ float fabs (float a)

static __device__ double fabs (double a)

static __device__ float sqrtf (float a)

static __device__ float __saturatef (float a)

static __device__ float __sinf (float a)

static __device__ float __cosf (float a)

static __device__ float __expf (float a)

static __device__ float __powf (float a, float b)

Variables

int2

int4

Macro Definition Documentation

◆ global

#define __global__ __attribute__((global))

Definition at line 27 of file cuda_runtime.h.

◆ device

#define __device__ __attribute__((device))

Definition at line 28 of file cuda_runtime.h.

◆ __device_builtin__

#define __device_builtin__ __attribute__((device_builtin))

Definition at line 29 of file cuda_runtime.h.

◆ align

#define __align__ ( N ) __attribute__((aligned(N)))

Definition at line 30 of file cuda_runtime.h.

◆ inline

#define __inline__ __inline__ __attribute__((always_inline))

Definition at line 31 of file cuda_runtime.h.

◆ max

#define max ( a,

b

) ((a) > (b) ? (a) : (b))

Definition at line 33 of file cuda_runtime.h.

◆ min

#define min ( a,

b

) ((a) < (b) ? (a) : (b))

Definition at line 34 of file cuda_runtime.h.

◆ abs

#define abs ( x ) ((x) < 0 ? -(x) : (x))

Examples: encode_audio.c.

Definition at line 35 of file cuda_runtime.h.

◆ atomicAdd

#define atomicAdd ( a,

b

) (__atomic_fetch_add(a, b, __ATOMIC_SEQ_CST))

Definition at line 37 of file cuda_runtime.h.

◆ GETCOMP

#define GETCOMP ( reg,

comp

)

Value:

asm("mov.u32 %0, %%" #reg "." #comp ";" : "=r"(tmp)); \

ret.comp = tmp;

Definition at line 90 of file cuda_runtime.h.

◆ GET

#define GET ( name,

reg

)

Value:

static inline __device__ uint3 name() {\

uint3 ret; \

unsigned tmp; \

GETCOMP(reg, x) \

GETCOMP(reg, y) \

GETCOMP(reg, z) \

return ret; \

}

Definition at line 94 of file cuda_runtime.h.

◆ blockIdx

#define blockIdx (getBlockIdx())

Definition at line 108 of file cuda_runtime.h.

◆ blockDim

#define blockDim (getBlockDim())

Definition at line 109 of file cuda_runtime.h.

◆ threadIdx

#define threadIdx (getThreadIdx())

Definition at line 110 of file cuda_runtime.h.

◆ make_int2

#define make_int2 ( a,

b

) ((int2){.x = a, .y = b})

Definition at line 113 of file cuda_runtime.h.

◆ make_uchar2

#define make_uchar2 ( a,

b

) ((uchar2){.x = a, .y = b})

Definition at line 114 of file cuda_runtime.h.

◆ make_ushort2

#define make_ushort2 ( a,

b

) ((ushort2){.x = a, .y = b})

Definition at line 115 of file cuda_runtime.h.

◆ make_float2

#define make_float2 ( a,

b

) ((float2){.x = a, .y = b})

Definition at line 116 of file cuda_runtime.h.

◆ make_int4

#define make_int4 ( a,

b,

c,

d

) ((int4){.x = a, .y = b, .z = c, .w = d})

Definition at line 117 of file cuda_runtime.h.

◆ make_uchar4

#define make_uchar4 ( a,

b,

c,

d

) ((uchar4){.x = a, .y = b, .z = c, .w = d})

Definition at line 118 of file cuda_runtime.h.

◆ make_ushort4

#define make_ushort4 ( a,

b,

c,

d

) ((ushort4){.x = a, .y = b, .z = c, .w = d})

Definition at line 119 of file cuda_runtime.h.

◆ make_float4

#define make_float4 ( a,

b,

c,

d

) ((float4){.x = a, .y = b, .z = c, .w = d})

Definition at line 120 of file cuda_runtime.h.

◆ TEX2D

#define TEX2D ( type,

ret

) static inline __device__ void conv(type* out, unsigned a, unsigned b, unsigned c, unsigned d) {*out = (ret);}

Definition at line 123 of file cuda_runtime.h.

Typedef Documentation

◆ cudaTextureObject_t

typedef __device_builtin__ unsigned long long cudaTextureObject_t

Definition at line 40 of file cuda_runtime.h.

Function Documentation

◆ align() [1/4]

struct __device_builtin__ __align__ ( 2 )

Definition at line 42 of file cuda_runtime.h.

◆ align() [2/4]

struct __device_builtin__ __align__ ( 4 )

Definition at line 47 of file cuda_runtime.h.

◆ align() [3/4]

struct __device_builtin__ __align__ ( 8 )

Definition at line 52 of file cuda_runtime.h.

◆ align() [4/4]

struct __device_builtin__ __align__ ( 16 )

Definition at line 79 of file cuda_runtime.h.

◆ tex2D()

template<typename T >

__device__ T tex2D ( cudaTextureObject_t texObject,

float x,

float y

)

inline

Definition at line 137 of file cuda_runtime.h.

◆ tex2D< float4 >()

template<>

__device__ float4 tex2D< float4 > ( cudaTextureObject_t texObject,

float x,

float y

)

inline

Definition at line 149 of file cuda_runtime.h.

Referenced by tex2D< float >(), and tex2D< float2 >().

◆ tex2D< float >()

template<>

__device__ float tex2D< float > ( cudaTextureObject_t texObject,

float x,

float y

)

inline

Definition at line 159 of file cuda_runtime.h.

◆ tex2D< float2 >()

template<>

__device__ float2 tex2D< float2 > ( cudaTextureObject_t texObject,

float x,

float y

)

inline

Definition at line 165 of file cuda_runtime.h.

◆ floorf()

static __device__ float floorf ( float a )

inlinestatic

Definition at line 172 of file cuda_runtime.h.

Referenced by barrelsplit_to_xyz(), compute_sat_hue_metrics16(), compute_sat_hue_metrics8(), cube1x6_to_xyz(), cube3x2_to_xyz(), cube6x1_to_xyz(), eac_to_xyz(), fract(), frand(), lerp_colors(), lerp_colors16(), lerp_colors32(), plot_channel_log(), xyz_to_ball(), xyz_to_barrel(), xyz_to_barrelsplit(), xyz_to_cube1x6(), xyz_to_cube3x2(), xyz_to_cube6x1(), xyz_to_cylindrical(), xyz_to_cylindricalea(), xyz_to_dfisheye(), xyz_to_eac(), xyz_to_equirect(), xyz_to_equisolid(), xyz_to_fisheye(), xyz_to_flat(), xyz_to_hammer(), xyz_to_hequirect(), xyz_to_mercator(), xyz_to_octahedron(), xyz_to_orthographic(), xyz_to_pannini(), xyz_to_sinusoidal(), xyz_to_stereographic(), xyz_to_tetrahedron(), and xyz_to_tspyramid().

◆ floor() [1/2]

static __device__ float floor ( float a )

inlinestatic

Definition at line 173 of file cuda_runtime.h.

Referenced by ac3_bit_alloc_calc_bap_c(), ambisonics_config(), analyze_mono(), analyze_stereo(), av_channel_layout_ambisonic_order(), av_d2q(), av_ts_make_time_string2(), celt_quant_fine(), cng_encode_frame(), config_props(), create_map(), draw_mandelbrot(), eval_expr(), ff_ref_fdct(), ff_ref_idct(), ff_rint(), filter_frame(), floor_encode(), get_rotation(), init_cqt(), libdav1d_init(), MPA_encode_init(), qdmc_decode_init(), rint(), round(), roundf(), sox_read_header(), spectral_flatness(), tansig_approx(), trunc(), truncf(), update_residue_parameter(), vorbis_parse_audio_packet(), wavesynth_init(), and X264_frame().

◆ floor() [2/2]

static __device__ double floor ( double a )

inlinestatic

Definition at line 174 of file cuda_runtime.h.

◆ ceilf()

static __device__ float ceilf ( float a )

inlinestatic

Definition at line 175 of file cuda_runtime.h.

Referenced by config_input(), cube1x6_to_xyz(), cube3x2_to_xyz(), cube6x1_to_xyz(), ff_opus_psy_init(), kaiser_params(), opus_encode_init(), select_frame(), set_special_band_scalefactors(), unwrap(), xyz_to_cube1x6(), xyz_to_cube3x2(), xyz_to_cube6x1(), and xyz_to_dfisheye().

◆ ceil() [1/2]

static __device__ float ceil ( float a )

inlinestatic

Definition at line 176 of file cuda_runtime.h.

Referenced by avi_write_header(), calculate_blur(), config_output(), draw_response(), eval_expr(), ff_rint(), ff_vbv_update(), init_cqt(), initFilter(), libdav1d_init(), makelayers(), on2avc_decode_init(), request_frame(), resample_init(), rint(), round(), roundf(), segment_list_open(), set_params(), trunc(), truncf(), unsharp_opencl_make_filter_params(), update_residue_parameter(), and vvenc_set_framerate().

◆ ceil() [2/2]

static __device__ double ceil ( double a )

inlinestatic

Definition at line 177 of file cuda_runtime.h.

◆ truncf()

static __device__ float truncf ( float a )

inlinestatic

Definition at line 178 of file cuda_runtime.h.

Referenced by amrnb_decode_frame(), and amrwb_decode_frame().

◆ trunc() [1/2]

static __device__ float trunc ( float a )

inlinestatic

Definition at line 179 of file cuda_runtime.h.

Referenced by eval_expr(), and fwht_1d().

◆ trunc() [2/2]

static __device__ double trunc ( double a )

inlinestatic

Definition at line 180 of file cuda_runtime.h.

◆ fabsf()

static __device__ float fabsf ( float a )

inlinestatic

Definition at line 181 of file cuda_runtime.h.

Referenced by aac_encode_frame(), abs_pow34_v(), adaptive_gain_control(), angle_transform(), build_map(), butter_filter_stereo_samples(), calc_channel_uphases(), calculate_cubic_bc_coeffs(), calculate_factors(), checkasm_check_synth_filter(), decode_spectrum_and_dequant(), do_hsvkey_pixel(), elliott(), envelope(), eq_interp(), equal(), ff_frame_to_dnn_classify(), ff_frame_to_dnn_detect(), ff_pnm_decode_header(), ff_proc_from_dnn_to_frame(), ff_proc_from_frame_to_dnn(), filter_2_1(), filter_5_0_side(), filter_5_1_back(), filter_5_1_side(), filter_frame(), filter_stereo(), filter_surround(), find_form_factor(), find_peak(), fir_to_phase(), float_near_abs_eps(), get_lin_bin_abs(), get_log_bin_abs(), get_log_bin_sign(), headphone_convolute(), headphone_fast_convolute(), init_model_ov(), IsAlmostEqual(), main(), octahedron_to_xyz(), project(), quantize_and_encode_band_cost_template(), sanitize_values(), search_for_quantizers_anmr(), search_for_quantizers_twoloop(), set_params(), smooth(), sofalizer_convolute(), sofalizer_fast_convolute(), spectral_slope(), synthesis(), tetrahedron_to_xyz(), thresholding(), to_linear(), unwrap(), update_stat(), upmix_7_1_5_0_side(), upmix_7_1_5_1(), xyz_to_dfisheye(), xyz_to_flat(), xyz_to_octahedron(), and yule_filter_stereo_samples().

◆ fabs() [1/2]

static __device__ float fabs ( float a )

inlinestatic

Definition at line 182 of file cuda_runtime.h.

Referenced by aac_encode_frame(), av_bessel_i0(), av_d2q(), av_guess_frame_rate(), av_ts_make_time_string2(), avformat_find_stream_info(), bitreduction(), build_filter(), calc_noise_floor(), calc_thr_3gpp(), check_external_clock_speed(), check_values(), compand_delay(), compand_nodelay(), compare(), compressor(), compute_acc_den(), compute_acc_num(), compute_quant_matrix(), compute_target_delay(), config_filter(), config_input(), config_output(), config_props(), configure_input_video_filter(), configure_video_device(), configure_video_filters(), cubic_interpolate_func(), D(), dct_error(), decfloat(), decode_lspf(), decode_vector(), decompose_zp2biquads(), detect_clicks(), detect_clips(), detect_peak(), detect_scene_change(), double_near_abs_eps(), draw_response(), dump_fir(), encode_block(), evrc_decode_frame(), expand(), factor(), ff_icc_profile_detect_transfer(), ff_inlink_evaluate_timeline_at_frame(), ff_lpc_calc_coefs(), ff_pca(), ff_rfps_calculate(), filter_channels(), filter_frame(), filter_slice(), filter_slice_grey_edge(), find_peak_magnitude(), fir_to_phase(), fixed_gain_smooth(), floor_offset(), frame_erasure(), gate(), gauss_make_triangular(), generate_kernel(), generate_window_func(), get_coeff(), get_floor_average(), get_rotation(), get_scene_score(), get_video_frame(), hypot(), init(), init_muxer(), init_pass2(), init_video_param(), inv_cdf(), lfo_advance(), lup_decompose(), M(), main(), mcompand_channel(), measure_absmean(), mov_read_tkhd(), noise_filter(), optimize_colors(), parse_points(), parse_primary(), pchip_edge_case(), perform_compression(), points_not_collinear(), precalc_coefs(), print_stats(), procamp_vaapi_init(), process_frame(), quantize_lpc_coefs(), random_ts(), ready_residue(), residue_encode(), run_psnr(), set_constqp(), set_metadata(), set_vbr(), ssim_db(), swr_build_matrix2(), swr_next_pts(), sync_clock_to_slave(), synchronize_audio(), test_dcmul_add(), test_evaluate(), test_fcmul_add(), test_update(), test_vector_dmul(), test_vector_dmul_scalar(), test_vector_fmul(), test_vector_fmul_scalar(), true_peak_limiter(), twinvq_mulawinv(), update_qp(), update_stat(), vaapi_encode_av1_configure(), velvet_filter(), video_thread(), write_hls_media_playlist(), X264_init(), and XAVS_init().

◆ fabs() [2/2]

static __device__ double fabs ( double a )

inlinestatic

Definition at line 183 of file cuda_runtime.h.

◆ sqrtf()

static __device__ float sqrtf ( float a )

inlinestatic

Definition at line 184 of file cuda_runtime.h.

Referenced by aac_encode_frame(), abs_pow34_v(), ball_to_xyz(), basic_block_filtering(), bayes_threshold(), bessel_reinit(), calc_cpl_coord(), calc_input_response(), calc_reduced_thr_3gpp(), calc_thr_3gpp(), calculate_gamma(), cas_slice16(), cas_slice8(), celt_alg_quant(), celt_alg_unquant(), celt_calc_theta(), celt_frame_mdct(), celt_renormalize_vector(), celt_stereo_is_decouple(), celt_stereo_merge(), compute_frame_features(), compute_kernel(), compute_pitch_gain(), compute_rms(), config_input(), config_output(), config_params(), config_props(), dct(), decode_spectrum_and_dequant(), do_hsvkey_pixel(), euclidean_kernel(), ff_amr_set_fixed_gain(), ff_eac3_apply_spectral_extension(), filter16_prewitt(), filter16_roberts(), filter16_scharr(), filter16_sobel(), filter_channel(), filter_flt(), filter_frame(), filter_prewitt(), filter_roberts(), filter_scharr(), filter_sobel(), find_form_factor(), find_rms(), finish_block(), fir_to_phase(), frequency_band(), gather_input(), get_factors(), get_qm_factor(), get_scale(), get_zeropadded_input(), hammer_to_xyz(), hue_rotate_matrix(), init(), init_segment(), make_lpf(), mark_pns(), multicomb(), nb_decode(), normalize(), normalize_vector(), on2avc_scale(), pannini_to_xyz(), perspective_to_xyz(), pitch_filter(), pos_pow34(), print_stats(), process_anticollapse(), project(), ps_tableinit(), pvq_band_cost(), quant(), quant_band_template(), r_distance(), remap_log(), rgtc2_block_internal(), run_postproc(), sbr_gain_calc(), search_for_pns(), search_for_quantizers_anmr(), search_for_quantizers_twoloop(), set_params(), shue_rotate_matrix(), siren_init(), spectral_flux(), spectral_spread(), speex_decode_stereo(), spx_coordinates(), stddevf(), step_collect_psy_metrics(), weuclidean_kernel(), win(), xyz_to_ball(), and xyz_to_hammer().

◆ __saturatef()

static __device__ float __saturatef ( float a )

inlinestatic

Definition at line 186 of file cuda_runtime.h.

◆ __sinf()

static __device__ float __sinf ( float a )

inlinestatic

Definition at line 187 of file cuda_runtime.h.

◆ __cosf()

static __device__ float __cosf ( float a )

inlinestatic

Definition at line 188 of file cuda_runtime.h.

◆ __expf()

static __device__ float __expf ( float a )

inlinestatic

Definition at line 189 of file cuda_runtime.h.

◆ __powf()

static __device__ float __powf ( float a,

float b

)

inlinestatic

Definition at line 190 of file cuda_runtime.h.

Variable Documentation

◆ uchar2

uchar2

Definition at line 45 of file cuda_runtime.h.

◆ ushort2

ushort2

Definition at line 50 of file cuda_runtime.h.

◆ float2

float2

Definition at line 55 of file cuda_runtime.h.

◆ int2

int2

Definition at line 60 of file cuda_runtime.h.

◆ uchar4

uchar4

Definition at line 72 of file cuda_runtime.h.

◆ ushort4

ushort4

Definition at line 77 of file cuda_runtime.h.

◆ int4

int4

Definition at line 82 of file cuda_runtime.h.

◆ float4

float4

Definition at line 87 of file cuda_runtime.h.

Referenced by tex2D< float2 >(), and tex2D< float4 >().

it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf default minimum maximum flags name is the option name

Definition: writing_filters.txt:88

static void comp(unsigned char *dst, ptrdiff_t dst_stride, unsigned char *src, ptrdiff_t src_stride, int add)

Definition: eamad.c:81

static uint8_t tmp[11]

Definition: aes_ctr.c:28

#define __device__

Definition: cuda_runtime.h:28

ret

Definition: filter_design.txt:187

Definition: cuda_runtime.h:62

Generated on Fri Aug 22 2025 13:59:47 for FFmpeg by doxygen 1.8.17