SPIRAL

Introduction

This is a Python implementation of SPIRAL, a signal-power integrity co-analysis framework for high-speed inter-chiplet serial links validation. (https://ieeexplore.ieee.org/document/10473908)

The framework first builds equivalent models for the links with a machine learning based transmitter model and the impulse response of the channel and receiver. Then, the signal-power integrity is co-analyzed with a pulse response based method using the equivalent models. The framework calculates the output signals corresponding to the given input data and generates eye diagrams.

Dependencies

• Python = 3.8
• PyTorch >= 1.2.0
• Python packages:
  • numpy
  • scipy
  • skrf

Usage

First, prepare the trained TX model and S-parameters of the channel-RX. A few examples are provided, please refer to ./checkpoint to find TX models and ./link_rx to find channel-RX models.

Then, build the info file as b1_l3.txt and b2_s3.txt, which defines the models, UIs, S-parameters information, $Z_0$, $V_p$, dc resistance of channel, $C_L$, rising/falling time, input amplitude, input data, step and whether to consider the PSN and de-emphasis.

Run spiral_sipi.py to obtain the output signals and eye diagrams.