Public Functions

Files
file	noodle.h
	CNN/ML primitives for tiny MCUs with pluggable filesystem backends.

Classes
struct	Conv
	File-backed convolution parameters. More...
struct	ConvMem
	Memory-backed convolution parameters. More...
struct	Pool
	2D pooling parameters. Use M = 1 and T = 1 for identity (no pooling). More...
struct	FCN
struct	FCNFile
struct	FCNMem

Typedefs
typedef void(*	CBFPtr) (float progress)
	Progress callback type used by long-running routines.

Common parameter semantics
Standard meanings shared by many functions. Parameters W Input spatial width (2D) or length (1D). K Kernel size (2D: KxK, 1D: K). S Stride. P Zero-padding (per side). 2D uses top/left padding of size P. M Pool kernel size (2D: MxM). T Pool stride. n_inputs Number of input channels/features. n_outputs Number of output channels/features. in_fn Base input filename template (see File naming convention). out_fn Base output filename template. weight_fn Weight filename template; receives both I and O indices. bias_fn Bias filename (one bias per output channel, scalar per line). with_relu If true, apply ReLU after bias.
void	noodle_setup_temp_buffers (void *b1, void *b2)
	Provide two reusable temporary buffers used internally by file-streaming operations.Must be called before conv/FCN variants that read from files. Two temp buffers are needed for operations that read from a file. For C*W*W tensor, the buffer should be W*W.
void	noodle_setup_temp_buffers (void *b2)
	Provide a single reusable temporary buffer used internally by file-streaming ops.Must be called before conv/FCN variants that read from files. One temp buffer is needed for operations that reads from a variable. Hence, only output accumulator buffe is needed. For C*W*W tensor, the buffer should be W*W.

File and File-System Utilities
bool	noodle_fs_init (uint8_t clk_pin, uint8_t cmd_pin, uint8_t d0_pin)
	Initialize SD/FS backend (pins variant is meaningful only for SD_MMC).
bool	noodle_fs_init (uint8_t clk_pin, uint8_t cmd_pin, uint8_t d0_pin, uint8_t d1_pin, uint8_t d2_pin, uint8_t d3_pin)
	Initialize SD/FS backend with default pins/settings.
bool	noodle_fs_init ()
bool	noodle_fs_init (uint8_t cs_pin)
	Initialize SD/FS backend with a specific CS_PIN.
void	noodle_read_top_line (const char *fn, char *line, size_t maxlen)
	Read the first line of a given text file.
size_t	noodle_read_bytes_until (NDL_File &file, char terminator, char *buffer, size_t length)
	Read bytes from a file until a terminator or length-1 (NULL terminated).
void	noodle_delete_file (const char *fn)
	Delete a file if it exists.

Memory utilities
float *	noodle_create_buffer (uint16_t size)
	Allocate a raw float buffer of size bytes.
void	noodle_delete_buffer (float *buffer)
	Free a buffer allocated by noodle_create_buffer.
void	noodle_array_to_file (float *array, const char *fn, uint16_t n)
	Write an array of n floats to fn, one value per line. File will be opened and closed.
void	noodle_grid_to_file (byte *grid, const char *fn, uint16_t n)
	Write an n×n byte grid to fn as bytes, row-major. File will be opened and closed.
void	noodle_grid_to_file (float *grid, const char *fn, uint16_t n)
	Write an n×n float grid to fn, row-major.
void	noodle_array_from_file (const char *fn, float *buffer, uint16_t K)
	Read a float array of length K from fn (one value per line).
void	noodle_array_from_file (NDL_File &fi, float *buffer, uint16_t K)
	Read a float array of length K from an opened file handler fi (one value per line).
void	noodle_grid_from_file (const char *fn, byte *buffer, uint16_t K)
void	noodle_grid_from_file (NDL_File &fi, byte *buffer, uint16_t K)
	Read an K × K grid (stored as byte) from fi (opened file handler) into buffer.
void	noodle_grid_from_file (const char *fn, int8_t *buffer, uint16_t K)
	Read an K × K grid (stored as float) from fn into buffer.
void	noodle_grid_from_file (NDL_File &fi, int8_t *buffer, uint16_t K)
	Read an K × K grid (stored as int8) from fi (opened file handler) into buffer.
void	noodle_grid_from_file (const char *fn, float *buffer, uint16_t K)
	Read an K × K grid (stored as float) from fn into buffer.
void	noodle_reset_buffer (float *buffer, uint16_t n)
	Fill buffer with zeros (n floats).
float *	noodle_slice (float *flat, size_t W, size_t z)
void	noodle_array_to_file (float *array, NDL_File &fo, uint16_t n)
	Write an array of n floats to fo (an opened file handler), one value per line. No file open and close operations.
void	noodle_grid_to_file (byte *grid, NDL_File &fo, uint16_t n)
	Write an n×n byte grid to fo (opened file handler) as bytes, row-major. No file open and close operations.
void	noodle_grid_to_file (float *grid, NDL_File &fo, uint16_t n)
	Write an n×n float grid to fo (an opened file handler), row-major.
void	noodle_grid_from_file (NDL_File &fi, float *buffer, uint16_t K)

2D Convolution
Packed-file conventions: Input file: packed CHW planes (each plane W×W). Output file: packed CHW planes (each plane V_out×V_out) in output-channel order. Padding: Noodle uses symmetric, stride-independent padding. If P >= 0, that value is used as the symmetric padding on all sides. If P == 65535, padding is computed automatically as: P = floor((K - 1) / 2) which preserves spatial size when S = 1 and K is odd. Requires temporary buffers set via noodle_setup_temp_buffers. Buffer sizes are typically W×W floats, used as per-channel scratch space.
uint16_t	noodle_conv_byte (const char *in_fn, uint16_t n_inputs, uint16_t n_outputs, const char *out_fn, uint16_t W, const Conv &conv, const Pool &pool, CBFPtr progress_cb=NULL)
	File→File 2D conv with BYTE input feature maps.
uint16_t	noodle_conv_float (const char *in_fn, uint16_t n_inputs, uint16_t n_outputs, const char *out_fn, uint16_t W, const Conv &conv, const Pool &pool, CBFPtr progress_cb=NULL)
	File→File 2D conv with FLOAT input feature maps.
uint16_t	noodle_conv_float (const char *in_fn, uint16_t n_inputs, uint16_t n_outputs, float *output, uint16_t W, const Conv &conv, const Pool &pool, CBFPtr progress_cb=NULL)
	File→Memory 2D conv with FLOAT inputs; writes [O, Wo, Wo] tensor to output.
uint16_t	noodle_conv_float (float *input, uint16_t n_inputs, uint16_t n_outputs, const char *out_fn, uint16_t W, const Conv &conv, const Pool &pool, CBFPtr progress_cb=NULL)
	Memory→File 2D conv with FLOAT inputs and in-file conv parameters.
uint16_t	noodle_conv_float (float *input, uint16_t n_inputs, uint16_t n_outputs, const char *out_fn, uint16_t W, const ConvMem &conv, const Pool &pool, CBFPtr progress_cb=NULL)
	Memory→File 2D conv with FLOAT inputs and in-varibale conv parameters.
uint16_t	noodle_conv_float (float *input, uint16_t n_inputs, uint16_t n_outputs, float *output, uint16_t W, const Conv &conv, const Pool &pool, CBFPtr progress_cb=NULL)
	Memory→Memory 2D conv with FLOAT inputs and in-file conv parameters.
uint16_t	noodle_conv_float (float *input, uint16_t n_inputs, uint16_t n_outputs, float *output, uint16_t W, const ConvMem &conv, const Pool &pool, CBFPtr progress_cb=NULL)

1D Convolution
Conv.K used as kernel length W used as input length
uint16_t	noodle_conv1d (const char *in_fn, uint16_t n_inputs, const char *out_fn, uint16_t n_outputs, uint16_t W, const Conv &conv, const Pool &pool, CBFPtr progress_cb=NULL)
uint16_t	noodle_conv1d (const char *in_fn, uint16_t n_inputs, const char *out_fn, uint16_t n_outputs, uint16_t W, const Conv &conv, CBFPtr progress_cb=NULL)
uint16_t	noodle_conv1d (float *in, uint16_t n_inputs, float *out, uint16_t n_outputs, uint16_t W, const ConvMem &conv, CBFPtr progress_cb=NULL)
uint16_t	noodle_conv1d (const char *in_fn, uint16_t n_inputs, const char *out_fn, uint16_t n_outputs, uint16_t W, const ConvMem &conv, CBFPtr progress_cb)
uint16_t	noodle_conv1d (float *in, uint16_t n_inputs, const char *out_fn, uint16_t n_outputs, uint16_t W, const ConvMem &conv, CBFPtr progress_cb)
uint16_t	noodle_conv1d (const char *in_fn, uint16_t n_inputs, float *out, uint16_t n_outputs, uint16_t W, const ConvMem &conv, CBFPtr progress_cb)

Activations
uint16_t	noodle_soft_max (float *input_output, uint16_t n)
uint16_t	noodle_sigmoid (float *input_output, uint16_t n)
uint16_t	noodle_relu (float *input_output, uint16_t n)

Fully Connected Network
uint16_t	noodle_fcn (const int8_t *input, uint16_t n_inputs, uint16_t n_outputs, const char *out_fn, const FCNFile &fcn, CBFPtr progress_cb=NULL)
uint16_t	noodle_fcn (const char *in_fn, uint16_t n_inputs, uint16_t n_outputs, const char *out_fn, const FCNFile &fcn, CBFPtr progress_cb=NULL)
uint16_t	noodle_fcn (const float *input, uint16_t n_inputs, uint16_t n_outputs, float *output, const FCNMem &fcn, CBFPtr progress_cb=NULL)
uint16_t	noodle_fcn (const byte *input, uint16_t n_inputs, uint16_t n_outputs, float *output, const FCNFile &fcn, CBFPtr progress_cb=NULL)
uint16_t	noodle_fcn (const int8_t *input, uint16_t n_inputs, uint16_t n_outputs, float *output, const FCNFile &fcn, CBFPtr progress_cb=NULL)
uint16_t	noodle_fcn (const char *in_fn, uint16_t n_inputs, uint16_t n_outputs, float *output, const FCNFile &fcn, CBFPtr progress_cb=NULL)
uint16_t	noodle_fcn (const float *input, uint16_t n_inputs, uint16_t n_outputs, float *output, const FCNFile &fcn, CBFPtr progress_cb)
uint16_t	noodle_fcn (const float *input, uint16_t n_inputs, uint16_t n_outputs, const char *out_fn, const FCNFile &fcn, CBFPtr progress_cb)

Tensor Reshaping
uint16_t	noodle_flat (const char *in_fn, float *output, uint16_t V, uint16_t n_filters)
uint16_t	noodle_flat (float *input, float *output, uint16_t V, uint16_t n_filters)
uint16_t	noodle_gap (float *inout, uint16_t C, uint16_t W)
void	noodle_find_max (float *input, uint16_t n, float &max_val, uint16_t &max_idx)

2D Depth-wise Convolution
uint16_t	noodle_dwconv_float (const char *in_fn, uint16_t n_channels, const char *out_fn, uint16_t W, const Conv &conv, const Pool &pool, CBFPtr progress_cb)
uint16_t	noodle_dwconv_float (float *input, uint16_t n_channels, float *output, uint16_t W, const Conv &conv, const Pool &pool, CBFPtr progress_cb)
uint16_t	noodle_dwconv_float (float *input, uint16_t n_channels, float *output, uint16_t W, const ConvMem &conv, const Pool &pool, CBFPtr progress_cb)
void	noodle_unpack_bn_params (const float *bn_params, uint16_t C, const float **gamma, const float **beta, const float **mean, const float **var)

Batch Normalization
uint16_t	noodle_bn (float *x, uint16_t C, uint16_t W, const float *gamma, const float *beta, const float *mean, const float *var, float eps=1e-3)
uint16_t	noodle_bn (float *x, uint16_t C, uint16_t W, const float *bn_params, float eps=1e-3)
uint16_t	noodle_bn_relu (float *x, uint16_t C, uint16_t W, const float *gamma, const float *beta, const float *mean, const float *var, float eps=1e-3)
uint16_t	noodle_bn_relu (float *x, uint16_t C, uint16_t W, const float *bn_params, float eps=1e-3)

Detailed Description

Public functions, types, and configuration intended for application use.

Typedef Documentation

CBFPtr

typedef void(* CBFPtr) (float progress)

Progress callback type used by long-running routines.

Parameters

progress

A normalized progress in [0,1], monotonically nondecreasing.

Function Documentation

noodle_array_from_file() [1/2]

void noodle_array_from_file	(	const char *	fn,
		float *	buffer,
		uint16_t	K
	)

Read a float array of length K from fn (one value per line).

noodle_array_from_file() [2/2]

void noodle_array_from_file	(	NDL_File &	fi,
		float *	buffer,
		uint16_t	K
	)

Read a float array of length K from an opened file handler fi (one value per line).

noodle_array_to_file() [1/2]

void noodle_array_to_file	(	float *	array,
		const char *	fn,
		uint16_t	n
	)

Write an array of n floats to fn, one value per line. File will be opened and closed.

noodle_array_to_file() [2/2]

void noodle_array_to_file	(	float *	array,
		NDL_File &	fo,
		uint16_t	n
	)

Write an array of n floats to fo (an opened file handler), one value per line. No file open and close operations.

noodle_bn() [1/2]

uint16_t noodle_bn	(	float *	x,
		uint16_t	C,
		uint16_t	W,
		const float *	bn_params,
		float	eps = 1e-3
	)

Batch Normalization for a channel-first tensor in memory.

Parameters

x	Pointer to the input tensor in [C][W][W] layout.
C	Number of channels.
W	Width/height of each channel plane.
bn_params	Pointer to the packed batch normalization parameters.
eps	Small constant to avoid division by zero.

noodle_bn() [2/2]

uint16_t noodle_bn	(	float *	x,
		uint16_t	C,
		uint16_t	W,
		const float *	gamma,
		const float *	beta,
		const float *	mean,
		const float *	var,
		float	eps = 1e-3
	)

Batch Normalization for a channel-first tensor in memory.

Parameters

x	Pointer to the input tensor in [C][W][W] layout.
C	Number of channels.
W	Width/height of each channel plane.
gamma	Pointer to the per-channel scale parameters.
beta	Pointer to the per-channel shift parameters.
mean	Pointer to the per-channel mean parameters.
var	Pointer to the per-channel variance parameters.
eps	Small constant to avoid division by zero.

noodle_bn_relu() [1/2]

uint16_t noodle_bn_relu	(	float *	x,
		uint16_t	C,
		uint16_t	W,
		const float *	bn_params,
		float	eps = 1e-3
	)

noodle_bn_relu() [2/2]

uint16_t noodle_bn_relu	(	float *	x,
		uint16_t	C,
		uint16_t	W,
		const float *	gamma,
		const float *	beta,
		const float *	mean,
		const float *	var,
		float	eps = 1e-3
	)

Batch Normalization followed by ReLU for a channel-first tensor in memory.

Parameters

x	Pointer to the input tensor in [C][W][W] layout.
C	Number of channels.
W	Width/height of each channel plane.
gamma	Pointer to the per-channel scale parameters.
beta	Pointer to the per-channel shift parameters.
mean	Pointer to the per-channel mean parameters.
var	Pointer to the per-channel variance parameters.
eps	Small constant to avoid division by zero.

noodle_conv1d() [1/6]

uint16_t noodle_conv1d	(	const char *	in_fn,
		uint16_t	n_inputs,
		const char *	out_fn,
		uint16_t	n_outputs,
		uint16_t	W,
		const Conv &	conv,
		CBFPtr	progress_cb = NULL
	)

File CHW→File CHW 1D convolution with bias+activation and NO pooling stage.

Semantics as above but appends raw conv+bias(+ReLU) sequences for each output channel to out_fn.

Parameters

in_fn	Packed input filename (CHW).
n_inputs	Number of input channels I.
out_fn	Packed output filename (CHW).
n_outputs	Number of output channels O.
W	Input length.
conv	Convolution parameters (K, P, S, weight_fn, bias_fn, act).
progress_cb	Optional progress callback in [0,1].

Returns

V (pre-pooling output length).

noodle_conv1d() [2/6]

uint16_t noodle_conv1d	(	const char *	in_fn,
		uint16_t	n_inputs,
		const char *	out_fn,
		uint16_t	n_outputs,
		uint16_t	W,
		const Conv &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb = NULL
	)

File CHW→File CHW 1D convolution with optional bias+activation and a pooling stage.

This follows the same I/O convention as noodle_conv_float():

• in_fn is a single packed input file containing all input channels in CHW order (for 1D: C then W samples, one channel after another).
• out_fn is a single packed output file; for each output channel O we append either the pooled sequence or the raw sequence (depending on overload).
• Weights are read sequentially from conv.weight_fn in the order: for O in [0..n_outputs) and I in [0..n_inputs), read K floats (kernel taps).
• Biases are read sequentially from conv.bias_fn (one float per output channel).

Parameters

in_fn	Packed input filename (CHW).
n_inputs	Number of input channels I.
out_fn	Packed output filename (CHW).
n_outputs	Number of output channels O.
W	Input length.
conv	Convolution parameters (K, P, S, weight_fn, bias_fn, act).
pool	Pool parameters (kernel M, stride T).
progress_cb	Optional progress callback in [0,1].

Returns

V_out after pooling.

noodle_conv1d() [3/6]

uint16_t noodle_conv1d	(	const char *	in_fn,
		uint16_t	n_inputs,
		const char *	out_fn,
		uint16_t	n_outputs,
		uint16_t	W,
		const ConvMem &	conv,
		CBFPtr	progress_cb
	)

noodle_conv1d() [4/6]

uint16_t noodle_conv1d	(	const char *	in_fn,
		uint16_t	n_inputs,
		float *	out,
		uint16_t	n_outputs,
		uint16_t	W,
		const ConvMem &	conv,
		CBFPtr	progress_cb
	)

noodle_conv1d() [5/6]

uint16_t noodle_conv1d	(	float *	in,
		uint16_t	n_inputs,
		const char *	out_fn,
		uint16_t	n_outputs,
		uint16_t	W,
		const ConvMem &	conv,
		CBFPtr	progress_cb
	)

noodle_conv1d() [6/6]

uint16_t noodle_conv1d	(	float *	in,
		uint16_t	n_inputs,
		float *	out,
		uint16_t	n_outputs,
		uint16_t	W,
		const ConvMem &	conv,
		CBFPtr	progress_cb = NULL
	)

Memory→Memory 1D convolution with optional bias+activation and NO pooling stage. This operation does NOT need temp buffers!

Parameters

in	Input array (CHW).
n_inputs	Number of input channels I.
out	Output array (CHW).
n_outputs	Number of output channels O.
W	Input length.
conv	Convolution parameters (K, P, S, weight_fn, bias_fn, act).
progress_cb	Optional progress callback in [0,1].

Returns

V (pre-pooling output length).

noodle_conv_byte()

uint16_t noodle_conv_byte	(	const char *	in_fn,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		const char *	out_fn,
		uint16_t	W,
		const Conv &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb = NULL
	)

File→File 2D conv with BYTE input feature maps.

noodle_conv_float() [1/6]

uint16_t noodle_conv_float	(	const char *	in_fn,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		const char *	out_fn,
		uint16_t	W,
		const Conv &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb = NULL
	)

File→File 2D conv with FLOAT input feature maps.

noodle_conv_float() [2/6]

uint16_t noodle_conv_float	(	const char *	in_fn,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		float *	output,
		uint16_t	W,
		const Conv &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb = NULL
	)

File→Memory 2D conv with FLOAT inputs; writes [O, Wo, Wo] tensor to output.

noodle_conv_float() [3/6]

uint16_t noodle_conv_float	(	float *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		const char *	out_fn,
		uint16_t	W,
		const Conv &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb = NULL
	)

Memory→File 2D conv with FLOAT inputs and in-file conv parameters.

noodle_conv_float() [4/6]

uint16_t noodle_conv_float	(	float *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		const char *	out_fn,
		uint16_t	W,
		const ConvMem &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb = NULL
	)

Memory→File 2D conv with FLOAT inputs and in-varibale conv parameters.

noodle_conv_float() [5/6]

uint16_t noodle_conv_float	(	float *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		float *	output,
		uint16_t	W,
		const Conv &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb = NULL
	)

Memory→Memory 2D conv with FLOAT inputs and in-file conv parameters.

noodle_conv_float() [6/6]

uint16_t noodle_conv_float	(	float *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		float *	output,
		uint16_t	W,
		const ConvMem &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb = NULL
	)

Memory→Memory 2D conv with FLOAT inputs and in-variable conv parameters.

noodle_create_buffer()

float * noodle_create_buffer

(

uint16_t

size

)

Allocate a raw float buffer of size bytes.

noodle_delete_buffer()

void noodle_delete_buffer

(

float *

buffer

)

Free a buffer allocated by noodle_create_buffer.

noodle_delete_file()

void noodle_delete_file

(

const char *

fn

)

Delete a file if it exists.

noodle_dwconv_float() [1/3]

uint16_t noodle_dwconv_float	(	const char *	in_fn,
		uint16_t	n_channels,
		const char *	out_fn,
		uint16_t	W,
		const Conv &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb
	)

Depthwise convolution (float input/output; params from files).

For each input channel I, reads the I-th input feature map from in_fn (tokenized by I), convolves it with the depthwise kernel read from conv.weight_fn (also tokenized by I), adds bias from conv.bias_fn (one bias per input channel), applies activation, and writes the output feature map to out_fn (tokenized by I). Requires temp buffers set via noodle_setup_temp_buffers.

Parameters

in_fn	Base input filename template (receives I).
n_channels	Number of input/output channels.
out_fn	Base output filename template (receives I).
W	Input width/height.
conv	Convolution parameters (K, P, S, weight_fn, bias_fn, act).
pool	Pool parameters (kernel M, stride T).
progress_cb	Optional progress callback in [0,1].

Returns

Output width after pooling.

noodle_dwconv_float() [2/3]

uint16_t noodle_dwconv_float	(	float *	input,
		uint16_t	n_channels,
		float *	output,
		uint16_t	W,
		const Conv &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb
	)

Memory → memory depthwise conv (float input).

Assumes:

• input layout: [C][W][W] flattened
• output layout: [C][Wo][Wo] flattened (Wo depends on pooling)

  Parameters

  input	Pointer to the input tensor in [C][W][W] layout.
  n_channels	Number of input/output channels.
  output	Pointer to the output tensor in [C][Wo][Wo] layout.
  W	Input width/height.
  conv	Convolution parameters (K, P, S, weight_fn, bias_fn, act).
  pool	Pool parameters (kernel M, stride T).
  progress_cb	Optional progress callback in [0,1].

  Returns

  Output width after pooling.
  

noodle_dwconv_float() [3/3]

uint16_t noodle_dwconv_float	(	float *	input,
		uint16_t	n_channels,
		float *	output,
		uint16_t	W,
		const ConvMem &	conv,
		const Pool &	pool,
		CBFPtr	progress_cb
	)

Memory → memory depthwise conv (float input) with in-variable weights/bias. Assumes:

• input layout: [C][W][W] flattened
• output layout: [C][Wo][Wo] flattened (Wo depends on pooling )

  Parameters

  input	Pointer to the input tensor in [C][W][W] layout.
  n_channels	Number of input/output channels.
  output	Pointer to the output tensor in [C][Wo][Wo] layout.
  W	Input width/height.
  conv	Convolution parameters with in-variable weights/bias.
  pool	Pool parameters (kernel M, stride T).
  progress_cb	Optional progress callback in [0,1].

  Returns

  Output width after pooling.
  

noodle_fcn() [1/8]

uint16_t noodle_fcn	(	const byte *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		float *	output,
		const FCNFile &	fcn,
		CBFPtr	progress_cb = NULL
	)

Memory→Memory fully-connected layer (byte inputs; params from files).

Parameters

input	Byte array of length n_inputs (0..255 interpreted as float).
n_inputs	Number of inputs.
n_outputs	Number of outputs.
output	Float array of length n_outputs (written).
fcn	Filenames for weights/bias and activation mode.
progress_cb	Optional progress callback.

Returns

n_outputs.

noodle_fcn() [2/8]

uint16_t noodle_fcn	(	const char *	in_fn,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		const char *	out_fn,
		const FCNFile &	fcn,
		CBFPtr	progress_cb = NULL
	)

File→File fully-connected layer (float text inputs; params from files).

For each output neuron O, rewinds in_fn, accumulates dot(W[O], x) + b[O], applies activation, and appends to out_fn.

Parameters

in_fn	Input filename containing n_inputs floats (one per line).
n_inputs	Number of inputs.
n_outputs	Number of outputs.
out_fn	Output filename (appends/overwrites as created).
fcn	Filenames for weights/bias.
progress_cb	Optional progress callback in [0,1].

Returns

n_outputs.

noodle_fcn() [3/8]

uint16_t noodle_fcn	(	const char *	in_fn,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		float *	output,
		const FCNFile &	fcn,
		CBFPtr	progress_cb = NULL
	)

File→Memory fully-connected layer (float output; params from files).

Reads inputs from in_fn for each output neuron O, computing y[O] = dot(W[O], x) + b[O], then applies activation.

Parameters

in_fn	Input filename with n_inputs floats per forward pass.
n_inputs	Number of inputs.
n_outputs	Number of outputs.
output	Float array of length n_outputs (written).
fcn	Filenames for weights/bias.
progress_cb	Optional progress callback.

Returns

n_outputs.

noodle_fcn() [4/8]

uint16_t noodle_fcn	(	const float *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		const char *	out_fn,
		const FCNFile &	fcn,
		CBFPtr	progress_cb
	)

Memory→File fully-connected layer (float inputs; params from files).

Computes y = W·x + b, optionally applies activation, and writes n_outputs lines to out_fn.

Parameters

input	Pointer to n_inputs float values.
n_inputs	Number of inputs.
n_outputs	Number of outputs.
out_fn	Output filename (one float per line).
fcn	Filenames for weights and bias; weights read row-major [O, I].
progress_cb	Optional progress callback in [0,1].

Returns

n_outputs.

noodle_fcn() [5/8]

uint16_t noodle_fcn	(	const float *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		float *	output,
		const FCNFile &	fcn,
		CBFPtr	progress_cb
	)

Memory→Memory fully-connected layer (float output; params from files).

Reads inputs from input for each output neuron O, computing y[O] = dot(W[O], x) + b[O], then applies activation.

Parameters

input	Float array of length n_inputs.
n_inputs	Number of inputs.
n_outputs	Number of outputs.
output	Float array of length n_outputs (written).
fcn	Filenames for weights/bias.
progress_cb	Optional progress callback.

Returns

n_outputs.

noodle_fcn() [6/8]

uint16_t noodle_fcn	(	const float *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		float *	output,
		const FCNMem &	fcn,
		CBFPtr	progress_cb = NULL
	)

Memory→Memory fully-connected layer (float inputs; explicit in-variable weights/bias).

Weights are row-major [n_outputs, n_inputs] and biases length n_outputs.

Parameters

input	Float array of length n_inputs.
n_inputs	Number of inputs.
n_outputs	Number of outputs.
output	Float array of length n_outputs (written).
fcn	in-variable weights/bias and activation.
progress_cb	Optional progress callback in [0,1].

Returns

n_outputs.

noodle_fcn() [7/8]

uint16_t noodle_fcn	(	const int8_t *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		const char *	out_fn,
		const FCNFile &	fcn,
		CBFPtr	progress_cb = NULL
	)

Memory→File fully-connected layer (int8 inputs; weights/bias from files).

Computes y = W·x + b, optionally applies ReLU, and writes n_outputs lines to out_fn.

Parameters

input	Pointer to n_inputs int8 values.
n_inputs	Number of inputs.
n_outputs	Number of outputs.
out_fn	Output filename (one float per line).
fcn	Filenames for weights and bias; weights read row-major [O, I].
progress_cb	Optional progress callback in [0,1].

Returns

n_outputs.

noodle_fcn() [8/8]

uint16_t noodle_fcn	(	const int8_t *	input,
		uint16_t	n_inputs,
		uint16_t	n_outputs,
		float *	output,
		const FCNFile &	fcn,
		CBFPtr	progress_cb = NULL
	)

Memory→Memory fully-connected layer (int8 inputs; params from files).

Parameters

input	Int8 array of length n_inputs.
n_inputs	Number of inputs.
n_outputs	Number of outputs.
output	Float array of length n_outputs (written).
fcn	Filenames for weights/bias and activation mode.
progress_cb	Optional progress callback.

Returns

n_outputs.

noodle_find_max()

void noodle_find_max	(	float *	input,
		uint16_t	n,
		float &	max_val,
		uint16_t &	max_idx
	)

Find the maximum value and its index in a float array.

Parameters

input	Pointer to the input float array.
n	Length of the input array.
max_val	Reference to store the maximum value found.
max_idx	Reference to store the index of the maximum value.

noodle_flat() [1/2]

uint16_t noodle_flat	(	const char *	in_fn,
		float *	output,
		uint16_t	V,
		uint16_t	n_filters
	)

File→Memory flatten: reads n_filters feature maps from files named by in_fn

(tokenized by O via ::noodle_n2ll at positions 4/6 as appropriate) and writes a vector of length V×V×n_filters in row-major [i* n_filters + k].

Parameters

in_fn	Base filename of pooled feature maps (receives O).
output	Output buffer of length V×V×n_filters.
V	Spatial size (width=height).
n_filters	Number of channels (O).

Returns

V×V×n_filters.

noodle_flat() [2/2]

uint16_t noodle_flat	(	float *	input,
		float *	output,
		uint16_t	V,
		uint16_t	n_filters
	)

Memory→Memory flatten: flattens [O, V, V] into a vector of length V×V×n_filters.

Parameters

input	Base pointer to stacked feature maps [O, V, V].
output	Output buffer of length V×V×n_filters.
V	Spatial size.
n_filters	Number of channels O.

Returns

V×V×n_filters.

noodle_fs_init() [1/4]

bool noodle_fs_init

(

)

Initialize SD/FS backend with default pins/settings.

noodle_fs_init() [2/4]

bool noodle_fs_init	(	uint8_t	clk_pin,
		uint8_t	cmd_pin,
		uint8_t	d0_pin
	)

Initialize SD/FS backend (pins variant is meaningful only for SD_MMC).

noodle_fs_init() [3/4]

bool noodle_fs_init	(	uint8_t	clk_pin,
		uint8_t	cmd_pin,
		uint8_t	d0_pin,
		uint8_t	d1_pin,
		uint8_t	d2_pin,
		uint8_t	d3_pin
	)

Initialize SD/FS backend with default pins/settings.

noodle_fs_init() [4/4]

bool noodle_fs_init

(

uint8_t

cs_pin

)

Initialize SD/FS backend with a specific CS_PIN.

noodle_gap()

uint16_t noodle_gap	(	float *	inout,
		uint16_t	C,
		uint16_t	W
	)

Global Average Pooling for a channel-first tensor in memory.

Parameters

x_chw	Pointer to the input tensor in [C][W][W] layout.
C	Number of channels.
W	Width/height of each channel plane.

Returns

Number of channels C

noodle_grid_from_file() [1/6]

void noodle_grid_from_file	(	const char *	fn,
		byte *	buffer,
		uint16_t	K
	)

Read an K×K byte grid (stored as float) from fn into buffer.

noodle_grid_from_file() [2/6]

void noodle_grid_from_file	(	const char *	fn,
		float *	buffer,
		uint16_t	K
	)

Read an K × K grid (stored as float) from fn into buffer.

noodle_grid_from_file() [3/6]

void noodle_grid_from_file	(	const char *	fn,
		int8_t *	buffer,
		uint16_t	K
	)

Read an K × K grid (stored as float) from fn into buffer.

noodle_grid_from_file() [4/6]

void noodle_grid_from_file	(	NDL_File &	fi,
		byte *	buffer,
		uint16_t	K
	)

Read an K × K grid (stored as byte) from fi (opened file handler) into buffer.

noodle_grid_from_file() [5/6]

void noodle_grid_from_file	(	NDL_File &	fi,
		float *	buffer,
		uint16_t	K
	)

Read an K × K grid (stored as float) from an opened file handler fi into buffer.

noodle_grid_from_file() [6/6]

void noodle_grid_from_file	(	NDL_File &	fi,
		int8_t *	buffer,
		uint16_t	K
	)

Read an K × K grid (stored as int8) from fi (opened file handler) into buffer.

noodle_grid_to_file() [1/4]

void noodle_grid_to_file	(	byte *	grid,
		const char *	fn,
		uint16_t	n
	)

Write an n×n byte grid to fn as bytes, row-major. File will be opened and closed.

noodle_grid_to_file() [2/4]

void noodle_grid_to_file	(	byte *	grid,
		NDL_File &	fo,
		uint16_t	n
	)

Write an n×n byte grid to fo (opened file handler) as bytes, row-major. No file open and close operations.

noodle_grid_to_file() [3/4]

void noodle_grid_to_file	(	float *	grid,
		const char *	fn,
		uint16_t	n
	)

Write an n×n float grid to fn, row-major.

noodle_grid_to_file() [4/4]

void noodle_grid_to_file	(	float *	grid,
		NDL_File &	fo,
		uint16_t	n
	)

Write an n×n float grid to fo (an opened file handler), row-major.

noodle_read_bytes_until()

size_t noodle_read_bytes_until	(	NDL_File &	file,
		char	terminator,
		char *	buffer,
		size_t	length
	)

Read bytes from a file until a terminator or length-1 (NULL terminated).

Parameters

file	Open file handle.
terminator	Stop when this character is read (not stored).
buffer	Destination buffer (will always be NULL terminated).
length	Maximum bytes to write into buffer including the NULL.

Returns

Number of characters written (excluding NULL).

noodle_read_top_line()

void noodle_read_top_line	(	const char *	fn,
		char *	line,
		size_t	maxlen
	)

Read the first line of a given text file.

Parameters

fn	File name to read.
line	Reading result.
maxlen	Maximum character length to read.

noodle_relu()

uint16_t noodle_relu	(	float *	input_output,
		uint16_t	n
	)

In-place ReLU over a length-n vector. Returns n.

noodle_reset_buffer()

void noodle_reset_buffer	(	float *	buffer,
		uint16_t	n
	)

Fill buffer with zeros (n floats).

noodle_setup_temp_buffers() [1/2]

void noodle_setup_temp_buffers	(	void *	b1,
		void *	b2
	)

Provide two reusable temporary buffers used internally by file-streaming operations.Must be called before conv/FCN variants that read from files. Two temp buffers are needed for operations that read from a file. For C*W*W tensor, the buffer should be W*W.

Parameters

b1	Buffer #1 (input scratch). See size guidance above.
b2	Buffer #2 (float accumulator). See size guidance above.

noodle_setup_temp_buffers() [2/2]

void noodle_setup_temp_buffers

(

void *

b2

)

Provide a single reusable temporary buffer used internally by file-streaming ops.Must be called before conv/FCN variants that read from files. One temp buffer is needed for operations that reads from a variable. Hence, only output accumulator buffe is needed. For C*W*W tensor, the buffer should be W*W.

Parameters

b2	Buffer #2 (float accumulator). See size guidance above.

noodle_sigmoid()

uint16_t noodle_sigmoid	(	float *	input_output,
		uint16_t	n
	)

In-place sigmoid over a length-n vector. Returns n.

noodle_slice()

float * noodle_slice ( float *  flat, size_t  W, size_t  z  )

inline

Slice a stacked [Z, W, W] tensor laid out as contiguous planes.

Parameters

flat	Pointer to base of the contiguous array.
W	Width/height of each 2D plane.
z	Plane index to slice.

Returns

Pointer to the start of plane z (no bounds checks).

noodle_soft_max()

uint16_t noodle_soft_max	(	float *	input_output,
		uint16_t	n
	)

In-place softmax over a length-n vector. Returns n.

noodle_unpack_bn_params()

void noodle_unpack_bn_params	(	const float *	bn_params,
		uint16_t	C,
		const float **	gamma,
		const float **	beta,
		const float **	mean,
		const float **	var
	)

Unpack batch normalization parameters from a flat array.

Parameters

bn_params	Pointer to the packed batch normalization parameters.
C	Number of channels.
gamma	Output pointer to the per-channel scale parameters.
beta	Output pointer to the per-channel shift parameters.
mean	Output pointer to the per-channel mean parameters.
var	Output pointer to the per-channel variance parameters.