Skip to content

Latest commit

 

History

History
134 lines (108 loc) · 4.16 KB

toolchain_arty.md

File metadata and controls

134 lines (108 loc) · 4.16 KB

Installing open-source tools for ARTIX-7 FPGAs

There are two possible toolchains for ARTIX-7, either symbiflow (all-in-one, integrated, easier to install) or yosys + xray + nextpnr-xilinx (more complicated to install, but more up to date, and uses nextpnr instead of the heavier vpr).

Yosys/NextPNR/ProjectXRay

Step 1: yosys

Follow setup instructions from yosys website

TL;DR

Install prerequisites:

$ sudo apt-get install build-essential clang bison flex \
  libreadline-dev gawk tcl-dev libffi-dev git \
  graphviz xdot pkg-config python3 libboost-system-dev \
  libboost-python-dev libboost-filesystem-dev zlib1g-dev

If you want the GUI, you will need also to install:

$ sudo apt-get install libboost-python1.67.0-py38

Get the sources:

$ git clone https://github.com/YosysHQ/yosys.git

Compile and install it:

$ cd yosys
$ make
$ sudo make install

Step 2: prjxray

Either generate the database (but you need Vivado) or use a pre-generated one. OK, let's use the pre-generated one.

Follow the README from prjxray.

TL;DR

$ git clone https://github.com/SymbiFlow/prjxray.git
$ cd prjxray
$ git submodule update --init --recursive
$ make build
$ sudo make install
$ ./download-latest-db.sh
$ sudo mkdir -p /usr/share/nextpnr/
$ sudo cp -r database /usr/share/nextpnr/prjxray-db
$ sudo apt-get install python3 python3-pip python3-yaml
# NOT NEEDED ? $ sudo -H pip3 install -r requirements.txt # Rem: Arch Linux and Fedora uses *need* to read the README before.

Step 3: nextpnr-xilinx

Did not manage to build it with the gui, because on my box it gives a missing symbol error in boost-python, so I deactivated it.

$ git clone https://github.com/gatecat/nextpnr-xilinx.git
$ cd nextpnr-xilinx
$ git submodule init
$ git submodule update
$ mkdir build
$ cd build
$ cmake ../ -DARCH=xilinx -DBUILD_GUI=OFF -DBUILD_PYTHON=OFF 
$ make
$ sudo make install

One more thing, you will need to generate the chipdb (whatever that means, I do not understand everything here...). Learnt that from the 'building the Arty example' of this webpage). Needs to be adapted if you have a 100t instead of 35t.

$ cd nextpnr-xilinx
$ python3 xilinx/python/bbaexport.py --device xc7a35tcsg324-1 --bba xilinx/xc7a35t.bba
$ build/bbasm --l xilinx/xc7a35t.bba xilinx/xc7a35t.bin
$ sudo mkdir -p /usr/share/nextpnr/xilinx-chipdb
$ sudo cp xilinx/xc7a35t.bin /usr/share/nextpnr/xilinx-chipdb/

Here we go ! Now time to

ALTERNATIVE: SYMBIFLOW

Symbiflow documentation is here.

Step 1: Install symbiflow

To do that, a (custom) version of symbiflow-magic is bundled with FemtoRV. Symbiflow-magic is a makefile that downloads and configures a pre-compiled version of symbiflow for ARTIX-7. The version bundled with FemtoRV fixes a couple of issues.

$ cd learn-fpga/FemtoRV
$ sh -f TOOLS/get_symbifow.sh

It will download and install several packages (takes a while...)

Step 2: post-install step

Add anaconda initialization to your shell startup file as follows:

$ $HOME/opt/symbiflow/xc7/conda/bin/conda init $SHELL

Add the following line to your shell startup file (.bashrc if you use bash).

export PATH=/home/blevy/opt/symbiflow/xc7/install/bin/:$PATH
conda activate xc7

Start a new terminal window. If everything went well, the prompt should start with (xc7).

Edit learn_fpga/FemtoRV/Makefile, comment the line with arty_yosys_nextpnr.mk and uncomment the one with arty35_symbiflow.mk.