/pps23_final_project

Primary LanguageShell

pps23_final_project

git repo for final project of parallel programming.

project proposal/presentation/report

  1. ./group_22_proposal.pdf
  2. ./group_22_slide.pdf
  3. ./group_22_report.pdf

To make the documents made for the course from tex source:

cd docs 
make

Please be sure to install xelatex and tex packages used by the documents.

External projects, libs

  1. Klayout: The project we would like to explore parallelism opportunity, the focus is on DRC(Design rule check).
  2. openMP: Our plan A to parallel some DRC operation like "and". openMP is shipped with gcc/g++, just #include <omp.h> in the cpp source and -fopenmp when linking the binary.
  3. taskflow: Our plan B to parallelize DRC operation on task level.

Test caravel design from skywater 130nm openPDK

  1. test_case/decred_controller.gds.gz is a cell extracted from caravel.gds.gz of my forked repo originates from a real chip tape out, see efabless link

steps to build klayout and view the gds

  1. install QT5, python, ruby lib and other dependacies, see https://www.klayout.de/build.html
cd klayout; ./build.sh -option "-j8" -without-qtbinding  -build build-release
export LD_LIBRARY_PATH="$PWD/build-release"
./build-release/klayout ../test_case/caravel.gds.gz

Test with original and operation benchmark input

  1. design input: test_case/caravel.gds.gz
  2. ruby runset of "and" operation: test.drc
  3. to execute it: ./drc.sh, at my test machine

TaskFlow experiment

  1. ./tf.sh

test host

  1. CPU: 12th Gen Intel(R) Core(TM) i7-12700K
  2. MEM: 32 GB DDR4 MEM
  3. GPU: NVIDIA GeForce RTX 3050 NVIDIA-SMI 530.41.03 Driver Version: 530.41.03 CUDA Version: 12.1
  4. OS: proxmox-ve: 7.3-1 (running kernel: 5.15.35-1-pve)