Drag racing computers with FPGA design tools.
Lower runtime is better!
CPU | RAM | OS | Runtime (seconds) |
---|---|---|---|
Intel i9-13900K (5.8GHz) | DDR5 4800MT/s CL36 | Ubuntu 20.04 | 8,954 |
Intel i9-12900K (5.2GHz) | DDR5 4800MT/s CL36 | Ubuntu 20.04 | 9,833 |
AMD R9-5950X (4.9GHz) | DDR4 3600MT/s CL18 | Ubuntu 20.04 | 13,779 |
AMD EYPC 7313P (3.7GHz) | DDR4 3200MT/s CL22 | Ubuntu 20.04 | 16,501 |
Please use a OS supported by the Vivado version you are using. E.g. Ubuntu 20.04 for Vivado 2022.1.
Install perf tools:
sudo apt install linux-tools-`uname -r`
Set perf permissions (do this at your own risk):
sudo sh -c 'echo -1 >/proc/sys/kernel/perf_event_paranoid'
Install python dependencies:
pip3 install py-markdown-table
Install Vivado and Vitis HLS. You must select both of the following options during installation:
- "Install devices for Alveo and Xilinx edge acceleration platforms"
- "Virtex UltraScale+ HBM"
Make sure vivado and vitis_hls are in the path:
source /tools/Xilinx/Vivado/2022.1/settings64.sh
Run the benchmarks:
make benchmarks
Note: do not run the make command in parallel mode. Note: for hybrid CPUs such as Intel Alderlake, pin the process to the p-cores for correct performance counter metrics:
# For 8 p-cores (first 16 threads)
taskset -c 0-15 make benchmarks
Create the directory:
mkdir -p results/vivado_{MAJOR}_{MINOR}/{CPU_VENDOR}_{CPU_PART}_{CPU_MAX_FREQ}_{DDR_VERSION}_{DDR_SPEED}_{DDR_CAS_LATENCY}
# For example:
mkdir -p results/vivado_2022_1/intel_12900k_5_2ghz_ddr5_4800mt_cl36
Copy the output files to the above directory:
build/README.md
build/*.json
build/*.csv
Attach a link to your geekbench5 score in the README.md within your results directory.