RISC Processor called EV21 implemented it in Quartus for the Altera Cyclone IV FPGA. It runs over a 87.5 MHz base clock, has branch prediction algorithms and a 5 stage pipeline.
In the world of processors, there is two main types of implementations that are currently running in modern day applications:
-
CISC Processors (Complex Instruction Set Computer) That in short, have many intructions in its instruction set
-
RISC Processors (Reduced Instruction Set Computer) That in the other hand, they are characterized for having a low amount of instructions to use.
In this work, we focused on implementing a RISC Processor that was capable of implementing a 5 stage pipeline. The main focus was to make the branch prediction algorithm for jumps in code, the most optimized possible. In the next section we will get into a little more detail on how this was implemented.
To implement this processor we used the Altera Cyclone IV in the DE0 Nano FPGA Evaluation Board. To program this design we used Quartus as the main compilation and debugging program, and the code was written in Verilog. A more detailed documentation of the processor can be found at EV21 Documentation.
The main structure of the Processor can be seen as follows:
Also, the Register Bank is implemented as:
As you can see, it is separeted in different modules, that working together they implement the 5 stage pipeline, it cosists of:
Fetch --> Decode --> Operand --> Execute --> Retire
Each of this stages will be explained in the following sections.
The Fetch Stage consist of four modules:
PC(Program Counter)Program MemoryInstruction Register
Where the Program counter must increment its counter so that it reads one instruction for every clock cylcle. It is important to note that the PC must be driven by the Processor clock and by the branch prediction Block (called Bloque 1 in the diagram). The program Memory is an internal ROM memory containing the instructions uploaded to the processor. Finally the instruction Register must take the data from the program memory and output its Instruction code.
The decode Stage is composed by the Micro Instruction ROM, and its purpose is to translate the instruction received from the Instruction Register to a corresponding Microinstruction Set by the programmers. It is important to know that in this design, every instruction is composed by one and only one microinstruction.
The operand Stage consist of the output of the Microinstruction 1 Block, this block is the block positionated on top of the UC-1 block. This stage main purpose is to load the operand in the correspondign positions for the ALU (Aritmetic Logic Unit) to perform the correspondign operation over those operands.
The execute Stage consist of the Microinstruction 2 Block located under the UC-1 block. The purpose of this stage is to tell the ALU to perform the operation to the operands previously loaded.
Finally the Retire stage consist of the Microinstruction 3 Block located under the Microinstruction 2. The main purpose of this stage is to save the results of the ALU in the corresponding register of the Register Bank.
The intruction set used for this Processor is as follows:
You can see, that as many other RISC Processors, there are only two instructions for memory management, MOM Y,W and MOM W,Y.
To test the implementation of this processor a compiler was made from mnemonic notation in the instruction set to machine language instructions. The compiler can be found in ./Compilador and the usage is as follows:
python evc.py <Name of the .ev Program to compile>
A Program_memory.mif file will be created if no error in the code has been found, to load this program in the Processor you must move this file to the ./EV21/ folder and compile the Quartus Project.
Some program examples can be found in the compilator folder.
The first program we have tested is the following:
MOK W,#10 //W=10
MOV 0,W //R0 = W
MOK W,#19 //W=19
ADW 1,0 //R1=W+R0+CY
MOV W,1 //W=R1 -> W=29
MOM 0,W //M(0) = W
MOK W,#7 //W=7
ADW 1,0 //R1 = R0+W+CY ->R1 = 17
MOM W,0 //W=M(0) ->W = 29
ADW 2,1 //R2 = R1+W+CY ->R2 = 46
MOV W,2 //W=R2 ->W=46
NOP
NOP
ADK W,#-1
JMP 1
And the waveform read by an oscilloscope can be seen as Follows:
The second Test program used was a simple counter program shown as:
NOP
NOP
MOK W,#1
MOV 0,W
MOK W,#0
NOP
NOP
NOP
NOP
ADR W,0
NOP
NOP
JMP 7
NOP
And the waveform read by an oscilloscope can be seen as Follows:
The third Test program implemented was a program to check if the input and output system was working propperly, for this, we inseted values on the input bus, and made the Processor copy thosw values on the Output bus.
NOP
NOP
NOP
NOP
NOP
MOV 30,28
NOP
NOP
NOP
NOP
JMP 3
And the waveform read by an oscilloscope can be seen as Follows:
Please do not hesitate to reach out to me if you find any issue with the code or if you have any questions.
-
Personal email: idiaz@itba.edu.ar
-
LinkedIn Profile: https://www.linkedin.com/in/iancraz/
MIT License
Copyright (c) 2021 Ian Cruz Diaz
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.