Float convolution route predicates

The rules by which arm_convolve_f16(), arm_convolve_f16_acc16() and arm_convolve_f32() choose a route, in their dispatch order: 1x1, 1xN, conv1d k5, conv1d k3, small-C (MVE float builds), packed-patch GEMM (when ctx holds one patch row), direct. The routers, their buffer-size queries and the direct entries of each route share these, so that a caller selecting the entry per layer ahead of time takes the route the router takes.

Machine-readable model

• ARM_NN_CONV_NHWC_PATCH_GEMM_F32_MIN_OCmacroThe packed-patch GEMM route needs at least this many output channels (float32 and float16 alike).
• ARM_NN_CONV_NHWC_PATCH_GEMM_F32_MIN_POSmacroThe packed-patch GEMM route needs at least this many output positions per batch (float32).
• ARM_NN_CONV_NHWC_PATCH_GEMM_F16_MIN_OCmacroThe packed-patch GEMM route needs at least this many output channels (float16).
• ARM_NN_CONV_NHWC_PATCH_GEMM_F16_MIN_POSmacroThe packed-patch GEMM route needs at least this many output positions per batch (float16).
• ARM_NN_CONV_SMALL_C_F16_LANESmacroLanes of the float16 small-C kernel: the route takes fewer input channels than this.
• ARM_NN_CONV_SMALL_C_F32_LANESmacroLanes of the float32 small-C kernel; the general gate takes fewer input channels.
• arm_nn_conv_flt_emptyfunctionIdentify an empty output request, after dimension validation.
• arm_nn_conv_flt_axis_okfunctionBound the signed coordinate intermediates for one nonempty output axis.
• arm_nn_conv_flt_dims_okfunctionValidate common float convolution metadata and accessed arithmetic.
• arm_nn_conv_flt_1x1_window_okfunctionCheck window coverage for an unpadded 1x1 body that samples directly.
• arm_nn_conv_f32_small_c_coords_okfunctionBound f32 small-C coordinates, including inactive lanes and contiguous-load spans.
• arm_nn_conv_flt_is_1x1functionThe 1x1 route: a 1x1 filter with no padding.
• arm_nn_conv_flt_is_1d_kfunctionA conv1d with a 1 x k filter: batch 1, input and output height 1, unit stride and dilation, no padding.
• arm_nn_conv_flt_is_1xnfunctionThe shape part of the 1xN route: input, output and filter height 1, a filter wider than 1, unit stride and dilation in height, unit dilation in width, a positi…
• arm_nn_conv_f16_is_small_cfunctionThe float16 small-C route: fewer input channels than ARMNNCONVSMALLCF16LANES, a positive output depth and width, a positive input width, filter width, stride a…
• arm_nn_conv_f32_is_small_cfunctionThe float32 small-C route: fewer input channels than ARMNNCONVSMALLCF32LANES and a positive output depth and width.
• arm_nn_conv_f32_is_small_c_3x3functionAdditional float32 small-C region: 3x3 filters, four input channels and up to 8 output channels.
• arm_nn_conv_flt_is_patch_gemmfunctionThe shape part of the packed-patch GEMM route: at least ARMNNCONVNHWCPATCHGEMMF16MINOC output channels and ARMNNCONVNHWCPATCHGEMMF16MINPOS output positions per…
• arm_nn_dw_f16_is_1d_k3functionThe depthwise conv1d k3 route of armdepthwiseconvf16(): batch 1, channel multiplier 1, a 1x3 filter, input and output height 1, unit stride and dilation, no pa…
• arm_nn_dw_f16_is_2x5functionThe depthwise 2x5 route of armdepthwiseconvf16(): a 2x5 filter, input height 2, output height 1, unit stride and dilation, no padding; any batch and channel mu…
• arm_nn_dw_f16_is_cin1functionThe shape part of the depthwise one-input-channel route of armdepthwiseconvf16() on MVE float builds: one input channel and at least CONVERTDWCONVWITHONEINPUTC…
• arm_nn_dw_f16_cin1_min_ctx_sizefunctionThe least ctx->size, in bytes, of the depthwise one-input-channel route: the filter packed in blocks of 8 output channels, then one patch row of KH x KW elemen…

macro

ARM_NN_CONV_NHWC_PATCH_GEMM_F32_MIN_OC

C

The packed-patch GEMM route needs at least this many output channels (float32 and float16 alike).

Include/arm_nnsupportfunctions_flt.h:1562

#define ARM_NN_CONV_NHWC_PATCH_GEMM_F32_MIN_OC (8)

The packed-patch GEMM route needs at least this many output channels (float32 and float16 alike).

macro

ARM_NN_CONV_NHWC_PATCH_GEMM_F32_MIN_POS

C

The packed-patch GEMM route needs at least this many output positions per batch (float32).

Include/arm_nnsupportfunctions_flt.h:1564

#define ARM_NN_CONV_NHWC_PATCH_GEMM_F32_MIN_POS (8)

The packed-patch GEMM route needs at least this many output positions per batch (float32).

macro

ARM_NN_CONV_NHWC_PATCH_GEMM_F16_MIN_OC

C

The packed-patch GEMM route needs at least this many output channels (float16).

Include/arm_nnsupportfunctions_flt.h:1566

#define ARM_NN_CONV_NHWC_PATCH_GEMM_F16_MIN_OC (8)

The packed-patch GEMM route needs at least this many output channels (float16).

macro

ARM_NN_CONV_NHWC_PATCH_GEMM_F16_MIN_POS

C

The packed-patch GEMM route needs at least this many output positions per batch (float16).

Include/arm_nnsupportfunctions_flt.h:1568

#define ARM_NN_CONV_NHWC_PATCH_GEMM_F16_MIN_POS (8)

The packed-patch GEMM route needs at least this many output positions per batch (float16).

macro

ARM_NN_CONV_SMALL_C_F16_LANES

C

Lanes of the float16 small-C kernel: the route takes fewer input channels than this.

Include/arm_nnsupportfunctions_flt.h:1575

#define ARM_NN_CONV_SMALL_C_F16_LANES (8)

Lanes of the float16 small-C kernel: the route takes fewer input channels than this.

macro

ARM_NN_CONV_SMALL_C_F32_LANES

C

Lanes of the float32 small-C kernel; the general gate takes fewer input channels.

Include/arm_nnsupportfunctions_flt.h:1577

#define ARM_NN_CONV_SMALL_C_F32_LANES (4)

Lanes of the float32 small-C kernel; the general gate takes fewer input channels.

function

arm_nn_conv_flt_empty

C

Identify an empty output request, after dimension validation.

Include/arm_nnsupportfunctions_flt.h:1584

static bool arm_nn_conv_flt_empty(const cmsis_nn_dims *output_dims)

Identify an empty output request, after dimension validation.

Parameters

Parameters of arm_nn_conv_flt_empty

Name	Type	Direction	Description
output_dims	const cmsis_nn_dims *	in	Output tensor dimensions.

Returns

Returns of arm_nn_conv_flt_empty

Description
true for zero batch or spatial extent.

function

arm_nn_conv_flt_axis_ok

C

Bound the signed coordinate intermediates for one nonempty output axis.

Include/arm_nnsupportfunctions_flt.h:1598

static bool arm_nn_conv_flt_axis_ok(int32_t out, int32_t kernel, int32_t stride, int32_t pad, int32_t dilation)

Bound the signed coordinate intermediates for one nonempty output axis.

Parameters

Parameters of arm_nn_conv_flt_axis_ok

Name	Type	Direction	Description
out	int32_t	in	Number of output positions.
kernel	int32_t	in	Number of filter taps.
stride	int32_t	in	Positive output step.
pad	int32_t	in	Nonnegative leading padding.
dilation	int32_t	in	Positive filter step.

Returns

Returns of arm_nn_conv_flt_axis_ok

Description
true when all coordinate intermediates fit int32.

function

arm_nn_conv_flt_dims_ok

C

Validate common float convolution metadata and accessed arithmetic.

Include/arm_nnsupportfunctions_flt.h:1619

static bool arm_nn_conv_flt_dims_ok(
    const cmsis_nn_tile *stride,
    const cmsis_nn_tile *padding,
    const cmsis_nn_tile *dilation,
    const cmsis_nn_dims *input_dims,
    const cmsis_nn_dims *filter_dims,
    const cmsis_nn_dims *output_dims,
    size_t element_size,
    int32_t packed_lanes,
    bool packed
)

Validate common float convolution metadata and accessed arithmetic. Output windows are explicit; generic kernels clip their out-of-input taps.

Parameters

Parameters of arm_nn_conv_flt_dims_ok

Name	Type	Direction	Description
stride	const cmsis_nn_tile *	in	Spatial output steps.
padding	const cmsis_nn_tile *	in	Leading spatial padding.
dilation	const cmsis_nn_tile *	in	Spatial filter steps.
input_dims	const cmsis_nn_dims *	in	Input dimensions [N, H, W, C_IN].
filter_dims	const cmsis_nn_dims *	in	Filter dimensions [C_OUT, KH, KW, C_IN].
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT].
element_size	size_t	in	Bytes per tensor element.
packed_lanes	int32_t	in	Channel block size and vector loop width for the dtype.
packed	bool	in	Whether the filter uses NT_N_PACKED storage.

Returns

Returns of arm_nn_conv_flt_dims_ok

Description
true for consistent metadata and representable nonempty work, or a valid empty request.

function

arm_nn_conv_flt_1x1_window_ok

C

Check window coverage for an unpadded 1x1 body that samples directly.

Include/arm_nnsupportfunctions_flt.h:1676

static bool arm_nn_conv_flt_1x1_window_ok(
    const cmsis_nn_tile *stride,
    const cmsis_nn_dims *input_dims,
    const cmsis_nn_dims *output_dims
)

Check window coverage for an unpadded 1x1 body that samples directly.

Parameters

Parameters of arm_nn_conv_flt_1x1_window_ok

Name	Type	Direction	Description
stride	const cmsis_nn_tile *	in	Spatial output steps.
input_dims	const cmsis_nn_dims *	in	Input dimensions.
output_dims	const cmsis_nn_dims *	in	Output dimensions.

Returns

Returns of arm_nn_conv_flt_1x1_window_ok

Description
true when every requested sample lies inside the input.

function

arm_nn_conv_f32_small_c_coords_ok

C

Bound f32 small-C coordinates, including inactive lanes and contiguous-load spans.

Include/arm_nnsupportfunctions_flt.h:1695

static bool arm_nn_conv_f32_small_c_coords_ok(
    const cmsis_nn_tile *stride,
    const cmsis_nn_tile *padding,
    const cmsis_nn_tile *dilation,
    const cmsis_nn_dims *input_dims,
    const cmsis_nn_dims *filter_dims,
    const cmsis_nn_dims *output_dims
)

Bound f32 small-C coordinates, including inactive lanes and contiguous-load spans.

Parameters

Parameters of arm_nn_conv_f32_small_c_coords_ok

Name	Type	Direction	Description
stride	const cmsis_nn_tile *	in	Spatial output steps.
padding	const cmsis_nn_tile *	in	Leading spatial padding.
dilation	const cmsis_nn_tile *	in	Spatial filter steps.
input_dims	const cmsis_nn_dims *	in	Validated input dimensions.
filter_dims	const cmsis_nn_dims *	in	Validated filter dimensions.
output_dims	const cmsis_nn_dims *	in	Validated nonempty output dimensions.

Returns

Returns of arm_nn_conv_f32_small_c_coords_ok

Description
true when signed lane coordinates and products remain representable.

function

arm_nn_conv_flt_is_1x1

C

The 1x1 route: a 1x1 filter with no padding.

Include/arm_nnsupportfunctions_flt.h:1716

static bool arm_nn_conv_flt_is_1x1(const cmsis_nn_tile *padding, const cmsis_nn_dims *filter_dims)

The 1x1 route: a 1x1 filter with no padding.

Parameters

Parameters of arm_nn_conv_flt_is_1x1

Name	Type	Direction	Description
padding	const cmsis_nn_tile *	in	Spatial zero-padding
filter_dims	const cmsis_nn_dims *	in	Filter dimensions [C_OUT, KH, KW, C_IN]

Returns

Returns of arm_nn_conv_flt_is_1x1

Description
true if the router takes the 1x1 route

function

arm_nn_conv_flt_is_1d_k

C

A conv1d with a 1 x k filter: batch 1, input and output height 1, unit stride and dilation, no padding.

Include/arm_nnsupportfunctions_flt.h:1732

static bool arm_nn_conv_flt_is_1d_k(
    const cmsis_nn_tile *stride,
    const cmsis_nn_tile *padding,
    const cmsis_nn_tile *dilation,
    const cmsis_nn_dims *input_dims,
    const cmsis_nn_dims *filter_dims,
    const cmsis_nn_dims *output_dims,
    const int32_t k
)

A conv1d with a 1 x k filter: batch 1, input and output height 1, unit stride and dilation, no padding.

Parameters

Parameters of arm_nn_conv_flt_is_1d_k

Name	Type	Direction	Description
stride	const cmsis_nn_tile *	in	Spatial stride
padding	const cmsis_nn_tile *	in	Spatial zero-padding
dilation	const cmsis_nn_tile *	in	Spatial dilation
input_dims	const cmsis_nn_dims *	in	Input dimensions [N, H, W, C_IN]
filter_dims	const cmsis_nn_dims *	in	Filter dimensions [C_OUT, KH, KW, C_IN]
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]
k	const int32_t	in	Filter width

Returns

Returns of arm_nn_conv_flt_is_1d_k

Description
true for the conv1d k5 route with k = 5 and the conv1d k3 route with k = 3

function

arm_nn_conv_flt_is_1xn

C

The shape part of the 1xN route: input, output and filter height 1, a filter wider than 1, unit stride and dilation in height, unit dilation in width, a positi…

Include/arm_nnsupportfunctions_flt.h:1758

static bool arm_nn_conv_flt_is_1xn(
    const cmsis_nn_tile *stride,
    const cmsis_nn_tile *padding,
    const cmsis_nn_tile *dilation,
    const cmsis_nn_dims *input_dims,
    const cmsis_nn_dims *filter_dims,
    const cmsis_nn_dims *output_dims
)

The shape part of the 1xN route: input, output and filter height 1, a filter wider than 1, unit stride and dilation in height, unit dilation in width, a positive width stride and no height padding. Unless NN_DISABLE_SPECIALIZATION is defined, the routers take the conv1d routes for their shapes first. The router also needs ctx to hold arm_convolve_1_x_n_f16_get_buffer_size() (or _f32).

Parameters

Parameters of arm_nn_conv_flt_is_1xn

Name	Type	Direction	Description
stride	const cmsis_nn_tile *	in	Spatial stride
padding	const cmsis_nn_tile *	in	Spatial zero-padding
dilation	const cmsis_nn_tile *	in	Spatial dilation
input_dims	const cmsis_nn_dims *	in	Input dimensions [N, H, W, C_IN]
filter_dims	const cmsis_nn_dims *	in	Filter dimensions [C_OUT, KH, KW, C_IN]
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]

Returns

Returns of arm_nn_conv_flt_is_1xn

Description
true if the shape takes the 1xN route when ctx is large enough and no conv1d route claims it

function

arm_nn_conv_f16_is_small_c

C

The float16 small-C route: fewer input channels than ARMNNCONVSMALLCF16LANES, a positive output depth and width, a positive input width, filter width, stride a…

Include/arm_nnsupportfunctions_flt.h:1783

static bool arm_nn_conv_f16_is_small_c(
    const cmsis_nn_tile *stride,
    const cmsis_nn_tile *padding,
    const cmsis_nn_tile *dilation,
    const cmsis_nn_dims *input_dims,
    const cmsis_nn_dims *filter_dims,
    const cmsis_nn_dims *output_dims
)

The float16 small-C route: fewer input channels than ARM_NN_CONV_SMALL_C_F16_LANES, a positive output depth and width, a positive input width, filter width, stride and dilation, a non-negative padding, and every gather and scatter offset within 16 bits.

Parameters

Parameters of arm_nn_conv_f16_is_small_c

Name	Type	Direction	Description
stride	const cmsis_nn_tile *	in	Spatial stride
padding	const cmsis_nn_tile *	in	Spatial zero-padding
dilation	const cmsis_nn_tile *	in	Spatial dilation
input_dims	const cmsis_nn_dims *	in	Input dimensions [N, H, W, C_IN]
filter_dims	const cmsis_nn_dims *	in	Filter dimensions [C_OUT, KH, KW, C_IN]
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]

Returns

Returns of arm_nn_conv_f16_is_small_c

Description
true if the float16 routers take the small-C route on MVE float builds

function

arm_nn_conv_f32_is_small_c

C

The float32 small-C route: fewer input channels than ARMNNCONVSMALLCF32LANES and a positive output depth and width.

Include/arm_nnsupportfunctions_flt.h:1814

static bool arm_nn_conv_f32_is_small_c(const cmsis_nn_dims *input_dims, const cmsis_nn_dims *output_dims)

The float32 small-C route: fewer input channels than ARM_NN_CONV_SMALL_C_F32_LANES and a positive output depth and width.

Parameters

Parameters of arm_nn_conv_f32_is_small_c

Name	Type	Direction	Description
input_dims	const cmsis_nn_dims *	in	Input dimensions [N, H, W, C_IN]
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]

Returns

Returns of arm_nn_conv_f32_is_small_c

Description
true if `arm_convolve_f32()` takes the small-C route on MVE float builds

function

arm_nn_conv_f32_is_small_c_3x3

C

Additional float32 small-C region: 3x3 filters, four input channels and up to 8 output channels.

Include/arm_nnsupportfunctions_flt.h:1828

static bool arm_nn_conv_f32_is_small_c_3x3(
    const cmsis_nn_dims *input_dims,
    const cmsis_nn_dims *filter_dims,
    const cmsis_nn_dims *output_dims
)

Additional float32 small-C region: 3x3 filters, four input channels and up to 8 output channels.

Note

Requires at least one full four-column output tile.

Parameters

Parameters of arm_nn_conv_f32_is_small_c_3x3

Name	Type	Direction	Description
input_dims	const cmsis_nn_dims *	in	Input dimensions [N, H, W, C_IN]
filter_dims	const cmsis_nn_dims *	in	Filter dimensions [C_OUT, HK, WK, C_IN]
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]

Returns

Returns of arm_nn_conv_f32_is_small_c_3x3

Description
true if the shape uses the small-C route, subject to its coordinate bounds

function

arm_nn_conv_flt_is_patch_gemm

C

The shape part of the packed-patch GEMM route: at least ARMNNCONVNHWCPATCHGEMMF16MINOC output channels and ARMNNCONVNHWCPATCHGEMMF16MINPOS output positions per…

Include/arm_nnsupportfunctions_flt.h:1843

static bool arm_nn_conv_flt_is_patch_gemm(const cmsis_nn_dims *output_dims)

The shape part of the packed-patch GEMM route: at least ARM_NN_CONV_NHWC_PATCH_GEMM_F16_MIN_OC output channels and ARM_NN_CONV_NHWC_PATCH_GEMM_F16_MIN_POS output positions per batch (the float32 thresholds are equal). The router also needs ctx to hold one patch row (KH x KW x C_IN elements).

Parameters

Parameters of arm_nn_conv_flt_is_patch_gemm

Name	Type	Direction	Description
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]

Returns

Returns of arm_nn_conv_flt_is_patch_gemm

Description
true if the shape takes the packed-patch GEMM route when ctx is large enough

function

arm_nn_dw_f16_is_1d_k3

C

The depthwise conv1d k3 route of armdepthwiseconvf16(): batch 1, channel multiplier 1, a 1x3 filter, input and output height 1, unit stride and dilation, no pa…

Include/arm_nnsupportfunctions_flt.h:1859

static bool arm_nn_dw_f16_is_1d_k3(
    const cmsis_nn_dw_conv_params_f16 *dw_conv_params,
    const cmsis_nn_dims *input_dims,
    const cmsis_nn_dims *filter_dims,
    const cmsis_nn_dims *output_dims
)

The depthwise conv1d k3 route of arm_depthwise_conv_f16(): batch 1, channel multiplier 1, a 1x3 filter, input and output height 1, unit stride and dilation, no padding.

Parameters

Parameters of arm_nn_dw_f16_is_1d_k3

Name	Type	Direction	Description
dw_conv_params	const cmsis_nn_dw_conv_params_f16 *	in	Depthwise parameters
input_dims	const cmsis_nn_dims *	in	Input dimensions [N, H, W, C_IN]
filter_dims	const cmsis_nn_dims *	in	Filter dimensions [1, HK, WK, C_OUT]
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]

Returns

Returns of arm_nn_dw_f16_is_1d_k3

Description
true if the router takes the conv1d k3 route (unless NN_DISABLE_SPECIALIZATION is defined)

function

arm_nn_dw_f16_is_2x5

C

The depthwise 2x5 route of armdepthwiseconvf16(): a 2x5 filter, input height 2, output height 1, unit stride and dilation, no padding; any batch and channel mu…

Include/arm_nnsupportfunctions_flt.h:1879

static bool arm_nn_dw_f16_is_2x5(
    const cmsis_nn_dw_conv_params_f16 *dw_conv_params,
    const cmsis_nn_dims *input_dims,
    const cmsis_nn_dims *filter_dims,
    const cmsis_nn_dims *output_dims
)

The depthwise 2x5 route of arm_depthwise_conv_f16(): a 2x5 filter, input height 2, output height 1, unit stride and dilation, no padding; any batch and channel multiplier.

Parameters

Parameters of arm_nn_dw_f16_is_2x5

Name	Type	Direction	Description
dw_conv_params	const cmsis_nn_dw_conv_params_f16 *	in	Depthwise parameters
input_dims	const cmsis_nn_dims *	in	Input dimensions [N, H, W, C_IN]
filter_dims	const cmsis_nn_dims *	in	Filter dimensions [1, HK, WK, C_OUT]
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]

Returns

Returns of arm_nn_dw_f16_is_2x5

Description
true if the router takes the 2x5 route (unless NN_DISABLE_SPECIALIZATION is defined)

function

arm_nn_dw_f16_is_cin1

C

The shape part of the depthwise one-input-channel route of armdepthwiseconvf16() on MVE float builds: one input channel and at least CONVERTDWCONVWITHONEINPUTC…

Include/arm_nnsupportfunctions_flt.h:1899

static bool arm_nn_dw_f16_is_cin1(const cmsis_nn_dims *input_dims, const cmsis_nn_dims *output_dims)

The shape part of the depthwise one-input-channel route of arm_depthwise_conv_f16() on MVE float builds: one input channel and at least CONVERT_DW_CONV_WITH_ONE_INPUT_CH_AND_OUTPUT_CH_ABOVE_THRESHOLD output channels (the threshold depends on the compiler). The router takes the route after the channel-multiplier-1 route, and only when ctx->size is at least arm_nn_dw_f16_cin1_min_ctx_size; otherwise it takes the generic route.

Parameters

Parameters of arm_nn_dw_f16_is_cin1

Name	Type	Direction	Description
input_dims	const cmsis_nn_dims *	in	Input dimensions [N, H, W, C_IN]
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]

Returns

Returns of arm_nn_dw_f16_is_cin1

Description
true if the shape takes the one-input-channel route when ctx is large enough

function

arm_nn_dw_f16_cin1_min_ctx_size

C

The least ctx->size, in bytes, of the depthwise one-input-channel route: the filter packed in blocks of 8 output channels, then one patch row of KH x KW elemen…

Include/arm_nnsupportfunctions_flt.h:1912

static int64_t arm_nn_dw_f16_cin1_min_ctx_size(const cmsis_nn_dims *filter_dims, const cmsis_nn_dims *output_dims)

The least ctx->size, in bytes, of the depthwise one-input-channel route: the filter packed in blocks of 8 output channels, then one patch row of KH x KW elements. More scratch packs more patch rows per step.

Parameters

Parameters of arm_nn_dw_f16_cin1_min_ctx_size

Name	Type	Direction	Description
filter_dims	const cmsis_nn_dims *	in	Filter dimensions [1, HK, WK, C_OUT]
output_dims	const cmsis_nn_dims *	in	Output dimensions [N, H, W, C_OUT]

Returns

Returns of arm_nn_dw_f16_cin1_min_ctx_size

Description
the size in bytes (64 bits, so that the caller can bound it), or INT64_MAX for a negative dimension or more than INT32_MAX taps, which no ctx can hold