Ariane is a 6-stage, single issue, in-order CPU which implements the 64-bit RISC-V instruction set. It fully implements I, M and C extensions as specified in Volume I: User-Level ISA V 2.1 as well as the draft privilege extension 1.10. It implements three privilege levels M, S, U to fully support a Unix-like operating system.
It has configurable size, separate TLBs, a hardware PTW and branch-prediction (branch target buffer and branch history table). The primary design goal was on reducing critical path length.
Go and get the RISC-V tools.
Checkout the repository and initialize all submodules
git clone https://github.com/pulp-platform/ariane.git
git submodule update --init --recursive
The Verilator testbench relies on our forked version of riscv-fesvr
which can be found here. Follow the README there and make sure that your compiler and linker is aware of the library (e.g.: add it to your path if it is in a non-default directory).
Build the Verilator model of Ariane by using the Makefile:
make verilate
This will create a C++ model of the core including a SystemVerilog wrapper and link it against a C++ testbench (in the tb
subfolder). The binary can be found in the obj_dir
and accepts a RISC-V ELF binary as an argument, e.g.:
obj_dir/Variane_wrapped -p rv64um-v-divuw
The Verilator testbench makes use of the riscv-fesvr
. That means that bare riscv-tests
can be run on the simulator.
Due to the way the C++ testbench is constructed we need a slightly altered version of the
riscv-fesvr
which can be found here.
It is possible to cross compile and run your own C-code or benchmarks on Ariane. The following steps need to be followed to compile and run:
Compile the file using the following command (you need to have the riscv-tests repo checked-out):
riscv64-unknown-elf-gcc -I./riscv-tests/benchmarks/../env -I./riscv-tests/benchmarks/common \
-DPREALLOCATE=1 -mcmodel=medany -static -std=gnu99 -O2 -ffast-math -fno-common \
-fno-builtin-printf ./riscv-tests/benchmarks/common/syscalls.c -static -nostdlib \
./riscv-tests/benchmarks/common/crt.S -nostartfiles -lm -lgcc \
-T ./riscv-tests/benchmarks/common/test.ld -o hello.riscv hello.c
Use the generated ELF file as an input to the Verilator model:
obj_dir/Variane_wrapped -p hello.riscv
While developing Ariane it has become evident that, in order to support Linux, the atomic extension is going to be mandatory. While the core is currently booting Linux by emulating Atomics in BBL (in a single core environment this is trivially met by disabling interrupts) this is not the behavior which is intended. For that reason we are going to fully support all atomic extensions in the very near future.
Furthermore, we have major IPC improvements planned. Specifically this will resolve about the way branches and jumps are currently handled in the core.
The core has been developed with a full licensed version of QuestaSim. If you happen to have this simulator available yourself here is how you could run the core with it. You need to generate both an elf
file and a hex
file, most easily this can be done by calling:
elf2hex 8 2097152 elf_file.riscv 2147483648 > elf_file.riscv.hex
Start the simulation using Modelsim:
make build
make sim
To specify the test to run use (e.g.: you want to run rv64ui-p-sraw
inside the tmp/risc-tests/build/isa
folder:
make sim riscv-test=rv64ui-p-sraw
If you need to specify a different directory you can pass the optional riscv-test-dir
flag:
make sim riscv-test=elf_name riscv-test-dir=/path/to/elf/and/hex/file
If you call simc
instead of sim
it will run without the GUI.
Or start any of the unit tests by:
make alu
Ariane's core testbench is fully compatible with the randomized constrained testing framework called Torture. To start testing Ariane all you need is to step into the riscv-torture/
folder and issue:
make rgentest
Which will produce a single randomized program, runs it on Spike (see Getting Started) and on the RTL simulator (QuestaSim) by calling ariane-run-torture
.
Torture's overnight tests work the same way, just call
make rnight
C (a.k.a. Verilator) tests are currently not supported.
Check out the contribution guide