introduction.org

About RISCV-FiveStage

The task in this exercise is to implement a 5-stage pipelined processor for the RISCV32I instruction set.

This exercise framework is used for the two graded exercises in the processor design course TDT4255, however you are more than welcome to use this project yourself, or to teach a class. Please reach out if you do!

If you are doing this as part of the TDT4255 course be sure to join our slack group. Slack links only last for a month, so the invite link will likely be expired. Here it is anyways, feel free to join even if you’re not taking the course at NTNU. https://join.slack.com/t/tdt4255-2020/shared_invite/zt-gcj17f7b-9fHN75cLmD1QaIH~n7csiw

In this exercise you will build a 5-stage RISCV32I processor that is able to run real RISC-V programs as long as they only use the 32I instruction subset. Since this is your first time building a processor, starting with a 5-stage design presents a very difficult challenge, which is why this exercise is split into two parts. In the first part the instructions will be interspersed with NOP instructions, four NOPs for every real. This means that you do not need to take into account dependencies and so forth, making things a lot easier for you.

For the second exercise the only difference is that NOP instructions will not be inserted. You can read about this in the ex2 guide, and will not be discussed further here.

In the project skeleton files (Found here) you can see that a lot of code has already been provided, which can make it difficult to get started. Hopefully this document can help clear up at least some of the confusion. The rest of this document gives an overview of the exercise framework and testing. If you want to jump straight to something practical you can start following the exercise guide, however at some point you should read through the rest of this document.

A tour of FiveStage

In order to orient yourself you first need a map, thus a high level overview of the processor you’re going to design is showed underneath: Keep in mind that this is just a high level sketch, omitting many details as well entire features (for instance branch logic)

Important When you are done, use the provided ./deliver.sh script to pack up the archive. If you’re unable to run bash scripts then please ensure that you deliver a zip archive. Not .rar or anything else, just use zip because my grading script knows how to handle that in addition to the one used by deliver.sh named after your username. Nothing more, nothing less, just your username. This archive should be runnable as is, thus you need to include all the necessary files. (I may or may not diff the tests to check if you’re screwing with them)

Now that you have an idea of what you’re building it is time to take inventory of the files included in the skeleton, and what, if anything should be added.

• ./src/main/scala/Tile.scala

  This is the top level module for the system as a whole. This is where the test harness accessses your design, providing the necessary IO.

  You should not modify this module for other purposes than debugging.

• ./src/main/scala/CPU.scala

  This is the top level module for your processor. In this module the various stages and barriers that make up your processor should be declared and wired together. Some of these modules have already been declared in order to wire up the debugging logic for your test harness. This file corresponds to the high-level overview in its entirety.

  This module is intended to be further fleshed out by you.

  As you work with this module, try keeping logic to a minimum to help readability. If you end up with a lot of signal select logic, consider moving that to a separate module.

• ./src/main/scala/IF.scala

  This is the instruction fetch stage. In this stage instruction fetching should happen, meaning you will have to add logic for handling branches, jumps, and for exercise 2, stalls. The reason this module is already included is that it contains the instruction memory, described next which is heavily coupled to the testing harness.

  This module is intended to be further fleshed out by you.

• ./src/main/scala/IMem.scala

  This module contains the instruction memory for your processor. Upon testing the test harness loads your program into the instruction memory, freeing you from the hassle.

  You should not modify this module for other purposes than maaaaybe debugging.

• ./src/main/scala/ID.scala

  The instruction decode stage. The reason this module is included is that the registers reside here, thus for the test harness to work it must be wired up to the register unit to record its state updates.

  This module is intended to be further fleshed out by you.

• ./src/main/scala/Registers.scala

  Contains the registers for your processor. Note that the zero register is alredy disabled, you do not need to do this yourself. The test harness ensures that all register updates are recorded.

  You should not modify this module for other purposes than maaaaybe debugging.

• ./src/main/scala/MEM.scala

  Like ID and IF, the MEM skeleton module is included so that the test harness can set up and monitor the data memory

  This module is intended to be further fleshed out by you.

• ./src/main/scala/DMem.scala

  Like the registers and Imem, the DMem is already implemented.

  You should not modify this module for other purposes than maaaaybe debugging.

• ./src/main/scala/Const.scala

  Contains helpful constants for decoding, used by the decoder which is provided.

  This module may be fleshed out further by you if you so choose.

• ./src/main/scala/Decoder.scala

  The decoder shows how to conveniently demux the instruction. In the provided ID.scala file a decoder module has already been instantiated. You should flesh it out further. You may find it useful to alter this module, especially in exercise 2.

  This module should be further fleshed out by you.

• ./src/main/scala/ToplevelSignals.scala

  Contains helpful constants. You should add your own constants here when you find the need for them. You are not required to use it at all, but it is very helpful.

  This module can be further fleshed out by you.

• ./src/main/scala/SetupSignals.scala

  You should obviously not modify this file. You may choose to create a similar file for debug signals, modeled on how the test harness is built.

  You should not modify this module at all.

Tests

In addition to the skeleton files it’s useful to take a look at how the tests work. You will not need to alter anything here other than the test manifest, but some of these settings can be quite useful to alter. The main attraction is the test options. By altering the verbosity settings you may change what is output. The settings are:

• printIfSuccessful Enables logging on tests that succeed. You typically want this turned off, at least for the full test runner.
• printErrors Enables logging of errors. You obviously want this one on, at least on the single test.
• printParsedProgram Prints the desugared program. Useful when the test asm contains instructions that needs to be expanded or altered. Unsure what “bnez” means? Turn this setting on and see!
• printVMtrace Enables printing of the VM trace, showing how the ideal machine executes a test
• printVMfinal Enables printing of the final VM state, showing how the registers look after completion. Useful if you want to see what a program returns.
• printMergedTrace Enables printing of a merged trace. With this option enabled you get to see how the VM and your processor executed the program side by side. This setting is extremely helpful to track down where your program goes wrong! This option attempts to synchronize the execution traces as best as it can, however once your processor design derails this becomes impossible, leading to rather nonsensical output. The output should look like this (picture is from exercise 2, without NOP padding)

  

  Instructions that were only executed by either VM or Your design is colored red or blue.

  IF YOU ARE COLOR BLIND YOU SHOULD ALTER THE DISPLAY COLORS!

  On some windows terminal emulators there exists a bug that causes colors to not display correctly, giving your terminal a very.. rastafarian look as shown below:

  

• nopPadded Set this to false when you’re ready to enter the big-boy league
• breakPoints Not implemented. It’s there as a teaser, urging you to implement it so I don’t have to.