On 1/12/2013, at 5:51 AM, Hynek Vychodil wrote: > But dc computes precise > > $ dc <<<"20k2452.45 100*245245-p" > 0 > > $ dc <<<"20k28 85*312*5*0.85/p" > 4368000.00000000000000000000 That's hardly surprising. dc(1) implements *decimal* floating point, though not IEEE decimal floating point.