/pulpino

Primary LanguageCOtherNOASSERTION

Introduction

PULPino is an open-source single-core microcontroller system, based on 32-bit RISC-V cores developed at ETH Zurich. PULPino is configurable to use either the RISCY or the zero-riscy core.

RISCY is an in-order, single-issue core with 4 pipeline stages and it has an IPC close to 1, full support for the base integer instruction set (RV32I), compressed instructions (RV32C) and multiplication instruction set extension (RV32M). It can be configured to have single-precision floating-point instruction set extension (RV32F). It implements several ISA extensions such as: hardware loops, post-incrementing load and store instructions, bit-manipulation instructions, MAC operations, support fixed-point operations, packed-SIMD instructions and the dot product. It has been designed to increase the energy efficiency of in ultra-low-power signal processing applications. RISCY implementes a subset of the 1.9 privileged specification. Further informations can be found in http://ieeexplore.ieee.org/abstract/document/7864441/.

zero-riscy is an in-order, single-issue core with 2 pipeline stages and it has full support for the base integer instruction set (RV32I) and compressed instructions (RV32C). It can be configured to have multiplication instruction set extension (RV32M) and the reduced number of registers extension (RV32E). It has been designed to target ultra-low-power and ultra-low-area constraints. zero-riscy implementes a subset of the 1.9 privileged specification.

When the core is idle, the platform can be put into a low power mode, where only a simple event unit is active and everything else is clock-gated and consumes minimal power (leakage). A specialized event unit wakes up the core in case an event/interrupt arrives.

For communication with the outside world, PULPino contains a broad set of peripherals, including I2S, I2C, SPI and UART. The platform internal devices can be accessed from outside via JTAG and SPI which allows pre-loading RAMs with executable code. In standalone mode, the platform boots from an internal boot ROM and loads its program from an external SPI flash.

The PULPino platform is available for RTL simulation as well FPGA. PULPino has been taped-out as an ASIC in UMC 65nm in January 2016. It has full debug support on all targets. In addition we support extended profiling with source code annotated execution times through KCacheGrind in RTL simulations.

Requirements

PULPino has the following requirements

  • ModelSim in reasonably recent version (we tested it with versions >= 10.2c)
  • CMake >= 2.8.0, versions greater than 3.1.0 recommended due to support for ninja
  • riscv-toolchain, specifically you need riscv32-unknown-elf-gcc compiler and friends. There are two choices for this toolchain: Either using the official RISC-V toolchain supported by Berkeley or the custom RISC-V toolchain from ETH. The ETH versions supports all the ISA extensions that were incorporated into the RI5CY core as well as the reduced base instruction set for zero-riscy. Please make sure you are using the newlib version of the toolchain.
  • python2 >= 2.6
  • verilator 3.884 only necessary if you want to use Verilator to evaluate PULPino.

ISA Support

PULPino can run either with RISCY or zero-riscy. The software included in this repository is compatible with both the cores and automatically targets the correct ISA based on the flags used. The simulator (modelsim) must be explicitely told which edition you want to build. Use the environment variable USE_ZERO_RISCY and set it to either 1 for zero-riscy or 0 for RISCY.

Version Control

PULPino uses multiple git subrepositories

To clone those subrepositores and update them, use

./update-ips.py

This script will read the ips_lists.txt file and update to the versions specified in there. You can choose specific commits, tags or branches.

Documentation

There is a preliminary datasheet available that includes a block diagram and a memory map of PULPino. See docs/datasheet/ in this repository.

It is written in LaTeX and there is no pdf included in the repository. Simply type

make all

inside the folder to generate the pdf. Note that you need a working version of latex for this step.

Running simulations

The software is built using CMake. Create a build folder somewhere, e.g. in the sw folder

mkdir build

Copy the cmake-configure.{*}.gcc.sh bash script to the build folder. This script can be found in the sw subfolder of the git repository.

Modify the cmake-configure script to your needs and execute it inside the build folder. This will setup everything to perform simulations using ModelSim.

Four cmake-configure bash scripts have been already configured:

  1. cmake_configure.riscv.gcc.sh

It automatically selects the RISCY cores and compiles SW with all the PULP-extensions and the RV32IM support. The GCC ETH compiler is needed and the GCC march flag set to "IMXpulpv2".

  1. cmake_configure.riscvfloat.gcc.sh

It automatically selects the RISCY cores and compiles SW with all the PULP-extensions and the RV32IMF support. The GCC ETH compiler is needed and he GCC march flag set to "IMFXpulpv2".

  1. cmake_configure.zeroriscy.gcc.sh

It automatically selects the zero-riscy cores and compiles SW with the RV32IM support (march flag set to RV32IM).

  1. cmake_configure.microriscy.gcc.sh

It automatically selects the zero-riscy cores and compiles SW with the RV32E support. The slim GCC ETH compiler is needed and he GCC march flag set to "RV32I" and the "-m16r" is passed to the compiler to use only the RV32E ISA support.

Activate the RVC flag in the cmake file if compressed instructions are desired.

Inside the build folder, execute

make vcompile

to compile the RTL libraries using ModelSim.

To run a simulation in the modelsim GUI use

make helloworld.vsim

To run simulations in the modelsim console use

make helloworld.vsimc

This will output a summary at the end of the simulation. This is intended for batch processing of a large number of tests.

Replace helloworld with the test/application you want to run.

Using ninja instead of make

You can use ninja instead make to build software for PULPino, just replace all occurrences of make with ninja. The same targets are supported on both make and ninja.

Interactive debug

To interactively debug software via gdb, you need the jtag bridge as well as a working version of gdb for the ISA you want to debug. The debug bridge depends on the jtag_dpi package that emulates a JTAG port and provides a TCP socket to which the jtag bridge can connect to.

Utilities

We additionally provide some utilitiy targets that are supposed to make development for PULPino easier.

For disassembling a program call

make helloworld.read

To regenerate the bootcode and copy it to the rtl folder use

make boot_code.install

FPGA

PULPino can be synthesized and run on a ZedBoard. Take a look at the fpga subfolder for more information.

Creating a tarball of the PULPino sources

If for some reason you don't want to use the git sub-repository approach, you can create a tarball of the whole design by executing ./create-tarball.py. This will download the latest PULPino sources, including all IPS, remove the git internal folders and create a tar gz.

Arduino compatible libraries

Most of official Arduino libraries are supported by PULPino software, they can be compiled, simulated and uploded the same way as traditional software programs using the available PULPino utilities. You only need to include main.cpp at the beginning of the program:

#include "main.cpp"

Take a look at the sw/libs/Arduino_libs subfolder for more information about the status of the currently supported libraries.