Our group has succesfully implemented and verified the following for our RV32I processor (note that the majority of this testing took place using gtkwave simulator)
| Other Implementations |
|---|
| Single cycle simple cpu |
The following blocks and tasks were distributid amongs the group as such:
Note the key : M - Major contribution p - partial contribution
| Block / feature | Denzil Erza-Essien 02593040 | Morris Griffin 02616740 | Apshara Amiruzzaman 02383793 | Dzuldiniy Hussain bin Dzulkeflee 02574233 |
|---|---|---|---|---|
| Bpredictor.sv | M | |||
| Branch_taken.sv | M | M | ||
| alu.sv | p | M | ||
| control_unit.sv | M | M | ||
| data_mem_cache.sv | M | p | ||
| hazard_unit.sv | M | p | ||
| instructionmemory.sv | M | |||
| pc.sv | p | M | ||
| registerfile.sv | M | |||
| signextend.sv | p | M | ||
| top.sv | M | |||
| Pipelined design | M | p | ||
| Full instruction implementation | M | M | ||
| Verification | M | M |
The following commands (and descriptions) run the following script files to allow for testing:
| Command | Description |
|---|---|
source start.sh |
Runs the file program.S in gtkwave |
source run_f1.fsm |
Runs the f1 fsm in gtkwave |
When simulating, the five .mem files were used to check for correct implementation. This can be seen in the following
screenshots:
After this, a simple F1 program was also tested to ensure everything worked as expected:
The project was completed using git commands throughout. Specific implementations were made
in branches and then merged into the main. This also allowed us to have different variants of the processor as required.
Although the commits and table do roughly represent the contribution of each team member, the following must also be noted:
- In some implementations, mainly those in which partial involvement is present, people may have been working together (whilst one person pushes into the repo), hence showing commit history for only one person.
- Although some of the blocks were trivial in theory, when implementing to the top file, there was a multitude of debugs all of which took hours to resolve, hence masking the time and contribution of members.
- Even with some of the more trivial implementations, given the further extensions, these basic blocks had to be extended to ensure complete design was achieved. This required time which may not be shown and collaboration between group members for all the blocks present.
We believe that the personal statements combined with the commit history and contribution table accurately represent the contribution of each member in the team.