sqdlab.github.io

Processor

This is data processing framework (python files/modules), built as semi-independent part of the SQDToolz stack. Internally it uses Numpy and Scipy for processing data on CPU and Cupy based on NVIDIA-CUDA for gpu processing. The input data follows a particular format as defined in Data format guide.

Data processing follows pipeline processing fashion and overall processor is to used as follows:

1. Create Processor Object using package object.
2. Reset/Clear processor pipeline (Processing that needs to be done).
3. Add processing stages in the order of how data should be processed.
4. Explicity add the last stage.
5. Pass the processor object to ACQ hal.

Remainder of this Document has the following outline:

• Detailed guide on creating, customising and binding processor objects.
• List of Processing stages for CPU and GPU.
• Customisation for each stage

1. Create Processor object:

Quick Start Guide also contains steps to creating processor object and binding it to lab.

General syntax:

<proc_obj> = <pkg_obj>.ProcessorCPU('<proc_name>', <lab_obj>)

Example:

myProc = stz.ProcessorCPU('ddcInteg', lab)

---

2. Reset the pipline:

It is a good idea to reset the processor object by using this function so it does not retain any previous processing stage from previous runs.

General Syntax

<proc_obj>.reset_pipeline()

Example:

myProc.reset_pipeline()

---

3. Add Processing Stages:

These stage are python modules located at : sqdtoolz/sqdtoolz/HAL/Processors/Processor-X, where X is CPU or GPU depending on the device you would like to use. WARNING(!): These stage CANNOT be inter-mixed. CPU Processor, requires adding ONLY CPU processing stage, using it with a GPU processor might cause the notebook or system CRASH.

A. Import the required stages:

General Syntax

from sqdtoolz.HAL.Processors.<proc_type>.<proc_type>_<stage_name> import*

Example:

from sqdtoolz.HAL.Processors.CPU.CPU_DDC import*
from sqdtoolz.HAL.Processors.CPU.CPU_FIR import*
from sqdtoolz.HAL.Processors.CPU.CPU_Mean import*

B. Adding the stages:

General Syntax:

<proc_obj>.add_stage( <pkg_obj>.<proc_type>_<stage_name>( <stage_params> ) )

Example:

myProc.add_stage(stz.CPU_DDC([25e6]))
myProc.add_stage(stz.CPU_FIR([{'Type' : 'low', 'Taps' : 128, 'fc' : 5e6, 'Win' : 'hamming'}]*2))
myProc.add_stage(stz.CPU_Mean('sample'))
myProc.add_stage(stz.CPU_Mean('segment'))

4. Add end stage:

This is nothing special ,but demarkates the end of processing pipeline, hence is added via a different function as follows:

General Syntax:

<proc_obj>.add_stage_end( <pkg_obj>.<proc_type>_<stage_name>( <stage_params> ) )

Example:

myProc.add_stage_end(stz.CPU_Mean('repetition'))

5. Bind processor to ACQ HAL:

The acqusition hal object, needs this processor object for processing the raw data coming from acqusition device. This object is passed as follows:

General syntax

<lab_obj>.HAL('<acq_hal_name>')..set_data_processor(<proc_obj>)

Example:

lab.HAL("DigiC").set_data_processor(myProc)

---

List of Processing stages

CPU processing stages:

• DDC
• ESD (or PSD Power Spectral Density) )
• FFT
• FIR
• Integrate
• Max
• Mean

GPU processing stages:

• DDC
• FIR
• Integrate
• Max
• Mean

---

Arguments for processing stages:

DDC

List of frequencies, corresponding to each channel of the acqusition device.

stz.CPU_DDC([25e6])

ESD

Tuple of IQ indices.

stz.CPU_ESD((0,1))

FFT

Tuple of IQ indices.

stz.CPU_FFT((0,1))

#### FIR Following are the optionals that are to passed as dictionary per channel (might have to create 2 list each belonging one sepeart channel:

• Type: low or high pass filter.
• Taps: number of taps to use.
• fc: Frequency cutoff.
• Win: Filter window.

stz.CPU_FIR([{'Type' : 'low', 'Taps' : 128, 'fc' : 5e6, 'Win' : 'hamming'}]*2)

Integrate

Name of parameter accross which it sums:

stz.CPU_Integrate("segments")

Max

Name of parameter accross which it find the max element:

stz.CPU_Max("segments")

Mean

Name of parameter accross which to calculate the mean:

CPU_Mean('segment')

---

This completed the Processor object documentation.