throughput_cpu_ref: sequential reference implementation
throughput_cpu_omp: OpenMP implementation provided by Gomez et al.
throughput_pim: PIM implementation provided by Gomez et al., excluding data transfer and host CPU calculations
throughput: PIM implementation provided by Gomez et al, entire calculation

n_elements=196608 → data/loc-gowalla_edges.txt
n_elements=2760448 → data/roadNet-CA.txt

baseline/cpu:
timer0 → sequential reference implementation
timer1 → OpenMP implementation

PIM:
timer0 → transfer graph CPU → DPU
timer1 → transfer config CPU → DPU
timer2 → frontier calculations on DPU
timer3 → retrieve frontier from DPUs, calculate the next one, and send it back
timer4 → retrieve results from DPUs