RadarRangeFFT#

Functions#

NVCVStatus pvaRangeFFTCreate(NVCVOperatorHandle *handle, NVCVTensorRequirements const *const inTensorRequirements, NVCVTensorRequirements const *const winTensorRequirements, NVCVTensorRequirements const *const outTensorRequirements, PVARangeFFTParams const *const params)

Constructs an instance of the RangeFFT operator.

NVCVStatus pvaRangeFFTSubmit(NVCVOperatorHandle handle, cupvaStream_t stream, NVCVTensorHandle const in, NVCVTensorHandle const win, NVCVTensorHandle out)

Submits the RangeFFT operator to a cuPVA stream.

NVCVStatus pvaRangeFFTSubmit(NVCVOperatorHandle handle, cudaStream_t stream, NVCVTensorHandle const in, NVCVTensorHandle const win, NVCVTensorHandle out)

Submits the RangeFFT operator to a CUDA stream.

Functions#

NVCVStatus pvaRangeFFTCreate(
NVCVOperatorHandle *handle,
NVCVTensorRequirements const *const inTensorRequirements,
NVCVTensorRequirements const *const winTensorRequirements,
NVCVTensorRequirements const *const outTensorRequirements,
PVARangeFFTParams const *const params,
)#

Constructs an instance of the RangeFFT operator.

Limitations:

For real input (S32): the supported number of samples must be in the range [4, 1024], must be even and must be factorizable by only factors of 2, 3, 4, and 5 For complex input (2S32): the supported number of samples must be in the range [2, 512] and must be factorizable by only factors of 2, 3, 4, and 5. For real input (S32): the number of range bins can’t be greater than (the number of samples / 2 + 1). For complex input (2S32): the number of range bins can’t be greater than the number of samples. If inputLayout or outputLayout is PVA_FFT_LAYOUT_RX_DOPPLER_RANGE, or outputLayout is PVA_FFT_LAYOUT_RANGE_RX_DOPPLER: the number of chirps must be a multiple of 8. If inputLayout is PVA_FFT_LAYOUT_RX_DOPPLER_RANGE: (number of RX channels * number of chirps) must not exceed 2048. G below (1 to 8) follows from the arguments: it is the largest value, up to the maximum for the number of samples below, for which the number of chirps is a multiple of 8 * G and, for a PVA_FFT_LAYOUT_RX_DOPPLER_RANGE output with (number of RX channels * number of chirps) above 2048 * G, the conditions listed for that output hold. The maximum is:

  • real input: 8 for 4 to 108 samples, 7 for 120 to 144, 6 for 150 to 160, 5 for 162 (6 with a PVA_FFT_LAYOUT_RANGE_RX_DOPPLER output), 5 for 180 to 192, 4 for 200 to 240, 3 for 250 to 324, 2 for 360 to 486, and 1 for 500 to 1024;

  • complex input: 8 for 2 to 54 samples, 7 for 60 to 72, 6 for 75 to 81, 5 for 90 to 96, 4 for 100 to 120, 3 for 125 to 162, 2 for 180 to 243, and 1 for 250 to 512. If inputLayout or outputLayout is PVA_FFT_LAYOUT_DOPPLER_RX_RANGE: at most 256 RX channels, and the number of chirps must not exceed 2048 * G. If inputLayout is PVA_FFT_LAYOUT_DOPPLER_RX_RANGE: (number of RX channels * number of samples) must not exceed 65535 for real input and 32767 for complex input. If outputLayout is PVA_FFT_LAYOUT_DOPPLER_RX_RANGE: (number of RX channels * number of range bins) must not exceed 32767. If outputLayout is PVA_FFT_LAYOUT_RANGE_RX_DOPPLER: the input must be real (S32), and (number of RX channels * number of chirps) must not exceed 2048 * G. If outputLayout is PVA_FFT_LAYOUT_RX_DOPPLER_RANGE and (number of RX channels * number of chirps) exceeds 2048 * G:

  • the number of range bins must be the maximum: (the number of samples / 2 + 1) for real input, the number of samples for complex input;

  • for real input the number of samples must be a multiple of 16; for complex input it must be one more than a multiple of 8;

  • (number of RX channels * number of chirps / (8 * G)) must be the product of an even number and another number, both no greater than 256, with the even number times (64 * G * number of range bins) no greater than 8388607. The input, window and output tensors must be densely packed, with no padding between dimensions. inputLayout must be PVA_FFT_LAYOUT_DOPPLER_RX_RANGE or PVA_FFT_LAYOUT_RX_DOPPLER_RANGE. outputLayout must be PVA_FFT_LAYOUT_DOPPLER_RX_RANGE, PVA_FFT_LAYOUT_RX_DOPPLER_RANGE or PVA_FFT_LAYOUT_RANGE_RX_DOPPLER.

Input: Data Layout: [HCW, NHCW]

If inputLayout is PVA_FFT_LAYOUT_DOPPLER_RX_RANGE (the default): W: Number of samples. C: Number of RX channels. H: Number of chirps.

If inputLayout is PVA_FFT_LAYOUT_RX_DOPPLER_RANGE: W: Number of samples. C: Number of chirps. H: Number of RX channels.

Batches: [1]

Rank: 3 for [HCW] or 4 for [NHCW].

Data Type

Allowed

8bit Unsigned

No

8bit Signed

No

16bit Unsigned

No

16bit Signed

No

32bit Unsigned

No

32bit Signed

Yes (real input)

32bit Float

No

64bit Float

No

2x16bit Unsigned

No

2x16bit Signed

No

2x32bit Unsigned

No

2x32bit Signed

Yes (complex input)

Input: Data Layout: [W, NW] W: Number of coefficients. Must be the same as the number of samples in the input tensor.

Batches: [1]

Rank: 1 for [W] or 2 for [NW].

Data Type

Allowed

8bit Unsigned

No

8bit Signed

No

16bit Unsigned

No

16bit Signed

No

32bit Unsigned

No

32bit Signed

Yes

32bit Float

No

64bit Float

No

2x16bit Unsigned

No

2x16bit Signed

No

2x32bit Unsigned

No

2x32bit Signed

No

Output: Data Layout: [HCW, NHCW]

If outputLayout is PVA_FFT_LAYOUT_DOPPLER_RX_RANGE (the default): W: Number of range bins. C: Number of RX channels. H: Number of chirps.

If outputLayout is PVA_FFT_LAYOUT_RX_DOPPLER_RANGE: W: Number of range bins. C: Number of chirps. H: Number of RX channels.

If outputLayout is PVA_FFT_LAYOUT_RANGE_RX_DOPPLER (real input only): W: Number of chirps. C: Number of RX channels. H: Number of range bins.

Batches: [1]

Rank: 3 for [HCW] or 4 for [NHCW].

Data Type

Allowed

8bit Unsigned

No

8bit Signed

No

16bit Unsigned

No

16bit Signed

No

32bit Unsigned

No

32bit Signed

No

32bit Float

No

64bit Float

No

2x16bit Unsigned

No

2x16bit Signed

No

2x32bit Unsigned

No

2x32bit Signed

Yes

Input/Output Dependency:

Property

Input == Output

Data Layout

Yes

Data Type

No

Number

Yes

Channels

Layout-dependent (see below)

Width

N/A

Height

Layout-dependent (see below)

The input and output always have the same number of RX channels and chirps, but the tensor dimension that carries each depends on the layout, so C and H are equal only when both ends put them in the same place. inputLayout and outputLayout are independent: a PVA_FFT_LAYOUT_RX_DOPPLER_RANGE input may produce a PVA_FFT_LAYOUT_DOPPLER_RX_RANGE output and vice versa.

Parameters:
  • handle – [out] Where the operator instance handle will be written to.

    • Must not be NULL.

  • inTensorRequirements – [in] Pointer to the NVCVTensorRequirements structure which contains input Tensor layout rank, shape and data type information.

  • winTensorRequirements – [in] Pointer to the NVCVTensorRequirements structure which contains window Tensor layout rank, shape and data type information.

  • outTensorRequirements – [in] Pointer to the NVCVTensorRequirements structure which contains output Tensor layout rank, shape and data type information.

  • params – [in] Pointer to the RangeFFT parameters. PVARangeFFTParams.

     windowType:   PVA_BATCH_FFT_WINDOW_HANNING or PVA_BATCH_FFT_WINDOW_USER_DEFINED.
     inputLayout:  PVA_FFT_LAYOUT_DOPPLER_RX_RANGE (default) or PVA_FFT_LAYOUT_RX_DOPPLER_RANGE.
     outputLayout: PVA_FFT_LAYOUT_DOPPLER_RX_RANGE (default), PVA_FFT_LAYOUT_RX_DOPPLER_RANGE or
                   PVA_FFT_LAYOUT_RANGE_RX_DOPPLER (real input only).
    
     A NULL params pointer selects the defaults, which are the first enumerator of each field.
     Layouts other than those listed above are not supported by this operator.
    

Return values:
  • NVCV_ERROR_INVALID_ARGUMENT – Handle is null or some parameter is outside valid range.

  • NVCV_ERROR_OUT_OF_MEMORY – Not enough memory to create the operator.

  • NVCV_ERROR_INVALID_IMAGE_FORMAT – Image format is invalid.

  • NVCV_SUCCESS – Operation executed successfully.

NVCVStatus pvaRangeFFTSubmit(
NVCVOperatorHandle handle,
cupvaStream_t stream,
NVCVTensorHandle const in,
NVCVTensorHandle const win,
NVCVTensorHandle out,
)#

Submits the RangeFFT operator to a cuPVA stream.

Parameters:
  • handle – [in] Handle to the operator.

    • Must not be NULL.

  • stream – [in] Handle to a valid cuPVA stream.

  • in – [in] Input tensor handle.

  • win – [in] Window tensor handle.

  • out – [out] Output tensor handle.

Return values:
  • NVCV_ERROR_INVALID_ARGUMENT – Handle is null.

  • NVCV_ERROR_INTERNAL – Internal error in the operator, invalid types passed in.

  • NVCV_SUCCESS – Operation executed successfully.

NVCVStatus pvaRangeFFTSubmit(
NVCVOperatorHandle handle,
cudaStream_t stream,
NVCVTensorHandle const in,
NVCVTensorHandle const win,
NVCVTensorHandle out,
)#

Submits the RangeFFT operator to a CUDA stream.

Note

CUDA stream support requirements:

  • PVA SDK 2.7.0 or later

  • Jetpack 7 or later

  • DriveOS 7 or later

  • x86 Emulator is not supported

Parameters:
  • handle – [in] Handle to the operator.

  • stream – [in] Handle to a valid CUDA stream.

  • in – [in] Input tensor handle.

  • win – [in] Window tensor handle.

  • out – [out] Output tensor handle.

Return values:
  • NVCV_ERROR_INVALID_ARGUMENT – Handle is null.

  • NVCV_ERROR_INTERNAL – Internal error in the operator, invalid types passed in.

  • NVCV_SUCCESS – Operation executed successfully.