%3

.. DO NOT EDIT. .. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY. .. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE: .. "examples/scheduling_pipelining_retiming.py" .. LINE NUMBERS ARE GIVEN BELOW. .. only:: html .. note:: :class: sphx-glr-download-link-note :ref:`Go to the end ` to download the full example code. .. rst-class:: sphx-glr-example-title .. _sphx_glr_examples_scheduling_pipelining_retiming.py: ========================================= Scheduling and Pipelining/Retiming ========================================= When scheduling cyclically (modulo) there is implicit pipelining/retiming taking place. B-ASIC can easily be used to showcase this. .. GENERATED FROM PYTHON SOURCE LINES 9-21 .. code-block:: Python import matplotlib.pyplot as plt import numpy as np from scipy import signal from b_asic.core_operations import Addition, ConstantMultiplication from b_asic.schedule import Schedule from b_asic.scheduler import ALAPScheduler from b_asic.sfg_generators import direct_form_1_iir from b_asic.signal_generator import Impulse from b_asic.simulation import Simulation .. GENERATED FROM PYTHON SOURCE LINES 22-23 Design a simple direct form IIR low-pass filter. .. GENERATED FROM PYTHON SOURCE LINES 23-27 .. code-block:: Python N = 3 Wc = 0.2 b, a = signal.butter(N, Wc, btype="lowpass", output="ba") .. GENERATED FROM PYTHON SOURCE LINES 28-29 Generate the corresponding signal-flow-graph (SFG). .. GENERATED FROM PYTHON SOURCE LINES 29-32 .. code-block:: Python sfg = direct_form_1_iir(b, a) sfg .. raw:: html

.. GENERATED FROM PYTHON SOURCE LINES 33-34 Set latencies and execution times of the operations. .. GENERATED FROM PYTHON SOURCE LINES 34-39 .. code-block:: Python sfg.set_latency_of_type(Addition, 1) sfg.set_latency_of_type(ConstantMultiplication, 3) sfg.set_execution_time_of_type(Addition, 1) sfg.set_execution_time_of_type(ConstantMultiplication, 1) .. GENERATED FROM PYTHON SOURCE LINES 40-41 Print the critical path Tcp and the iteration period bound Tmin. .. GENERATED FROM PYTHON SOURCE LINES 41-46 .. code-block:: Python T_cp = sfg.critical_path_time() print("Tcp:", T_cp) T_min = sfg.iteration_period_bound() print("Tmin:", T_min) .. rst-class:: sphx-glr-script-out .. code-block:: none Tcp: 7 Tmin: 5 .. GENERATED FROM PYTHON SOURCE LINES 47-48 Create an ALAP schedule .. GENERATED FROM PYTHON SOURCE LINES 48-51 .. code-block:: Python schedule = Schedule(sfg, scheduler=ALAPScheduler(), cyclic=True) schedule .. raw:: html

.. GENERATED FROM PYTHON SOURCE LINES 52-53 Move some operations "over the edge" in order to reach Tcp = Tmin. .. GENERATED FROM PYTHON SOURCE LINES 53-60 .. code-block:: Python schedule.move_operation('out0', 2) schedule.move_operation('add2', 2) schedule.move_operation('add0', 2) schedule.move_operation('add3', 2) schedule.set_schedule_time(5) schedule .. raw:: html

.. GENERATED FROM PYTHON SOURCE LINES 61-62 Print the new critical path Tcp that is now equal to Tmin. .. GENERATED FROM PYTHON SOURCE LINES 62-67 .. code-block:: Python T_cp = schedule.sfg.critical_path_time() print("Tcp:", T_cp) T_min = schedule.sfg.iteration_period_bound() print("Tmin:", T_min) .. rst-class:: sphx-glr-script-out .. code-block:: none Tcp: 5 Tmin: 5 .. GENERATED FROM PYTHON SOURCE LINES 68-69 Show the reconstructed SFG that is now pipelined/retimed compared to the original. .. GENERATED FROM PYTHON SOURCE LINES 69-71 .. code-block:: Python schedule.sfg .. raw:: html

.. GENERATED FROM PYTHON SOURCE LINES 72-75 Simulate the impulse response of the original and reconstructed SFGs. Plot the frequency responses of the original filter, the original SFG and the reconstructed SFG to verify that the schedule is valid. .. GENERATED FROM PYTHON SOURCE LINES 75-112 .. code-block:: Python sim1 = Simulation(sfg, [Impulse()]) sim1.run_for(1000) sim2 = Simulation(schedule.sfg, [Impulse()]) sim2.run_for(1000) w, h = signal.freqz(b, a) # Plot 1: Original filter spectrum_0 = 20 * np.log10(np.abs(h)) plt.figure() plt.plot(w / np.pi, spectrum_0) plt.title("Original filter") plt.xlabel("Normalized frequency (x pi rad/sample)") plt.ylabel("Magnitude (dB)") plt.grid(True) plt.show() # Plot 2: Simulated SFG spectrum_1 = 20 * np.log10(np.abs(signal.freqz(sim1.results["out0"])[1])) plt.figure() plt.plot(w / np.pi, spectrum_1) plt.title("Simulated SFG") plt.xlabel("Normalized frequency (x pi rad/sample)") plt.ylabel("Magnitude (dB)") plt.grid(True) plt.show() # Plot 3: Recreated SFG spectrum_2 = 20 * np.log10(np.abs(signal.freqz(sim2.results["out0"])[1])) plt.figure() plt.plot(w / np.pi, spectrum_2) plt.title("Pipelined/retimed SFG") plt.xlabel("Normalized frequency (x pi rad/sample)") plt.ylabel("Magnitude (dB)") plt.grid(True) plt.show() .. rst-class:: sphx-glr-horizontal * .. image-sg:: /examples/images/sphx_glr_scheduling_pipelining_retiming_001.png :alt: Original filter :srcset: /examples/images/sphx_glr_scheduling_pipelining_retiming_001.png :class: sphx-glr-multi-img * .. image-sg:: /examples/images/sphx_glr_scheduling_pipelining_retiming_002.png :alt: Simulated SFG :srcset: /examples/images/sphx_glr_scheduling_pipelining_retiming_002.png :class: sphx-glr-multi-img * .. image-sg:: /examples/images/sphx_glr_scheduling_pipelining_retiming_003.png :alt: Pipelined/retimed SFG :srcset: /examples/images/sphx_glr_scheduling_pipelining_retiming_003.png :class: sphx-glr-multi-img .. rst-class:: sphx-glr-timing **Total running time of the script:** (0 minutes 0.817 seconds) .. _sphx_glr_download_examples_scheduling_pipelining_retiming.py: .. only:: html .. container:: sphx-glr-footer sphx-glr-footer-example .. container:: sphx-glr-download sphx-glr-download-jupyter :download:`Download Jupyter notebook: scheduling_pipelining_retiming.ipynb ` .. container:: sphx-glr-download sphx-glr-download-python :download:`Download Python source code: scheduling_pipelining_retiming.py ` .. container:: sphx-glr-download sphx-glr-download-zip :download:`Download zipped: scheduling_pipelining_retiming.zip ` .. only:: html .. rst-class:: sphx-glr-signature `Gallery generated by Sphinx-Gallery `_