Significant speedup - string calculations

[tmeinlschmidt]

The issue with 'abc/%/def' % string - this is pretty costly. Did a test on my M1 and this approach is 2.5x slower than that new one. With tens of repeated processing per second, this could significantly improve processing.

test is pretty simple

import timeit

def current():
    for dt in ('Grid', 'Genset'):
        for ph in ('L1', 'L2', 'L3'):
            '/Ac/PvOn%s/%s/Power' % (dt, ph)
            '/Ac/PvOn%s/%s/Current' % (dt, ph)
            '/Ac/%s/%s/Power' % (dt, ph)
            '/Ac/%s/%s/Current' % (dt, ph)

PATHS = {}
for dt in ('Grid', 'Genset'):
    for ph in ('L1', 'L2', 'L3'):
        PATHS[(dt, ph, 'pvp')] = '/Ac/PvOn%s/%s/Power' % (dt, ph)
        PATHS[(dt, ph, 'pvc')] = '/Ac/PvOn%s/%s/Current' % (dt, ph)
        PATHS[(dt, ph, 'p')] = '/Ac/%s/%s/Power' % (dt, ph)
        PATHS[(dt, ph, 'c')] = '/Ac/%s/%s/Current' % (dt, ph)

def precomputed():
    for dt in ('Grid', 'Genset'):
        for ph in ('L1', 'L2', 'L3'):
            PATHS[(dt, ph, 'pvp')]
            PATHS[(dt, ph, 'pvc')]
            PATHS[(dt, ph, 'p')]
            PATHS[(dt, ph, 'c')]

print('current:     ', timeit.timeit(current, number=100000))
print('precomputed: ', timeit.timeit(precomputed, number=100000))

For few compued values (like active_input+1, phase) it was kept as it is.

results (mac m1)

current:      0.30169125
precomputed:  0.12832187499999997