A constant 1 g acceleration for decades is beyond any feasibility with the universe as we know it, so that problem is purely theoretical.
A 100% antimatter/matter annihilation rocket with an I
sp of 3*10
7 seconds (5 orders of magnitude higher than chemical rockets, 10 orders of magnitude in terms of energy density) would use about 2/3 of the ship mass as propellant per year to maintain this acceleration (as seen by the ship). Over 30 years, you need 10
14 times the ship mass as propellant (likely more as you need multiple stages). A single kilogram as final ship mass would need 10
14 kg of antimatter/matter, with an energy content of 10
31 J, several billion times the global yearly energy consumption.
There is no known way to get 100% efficiency, however. In particular, some energy will be lost to neutrinos (assuming we have to use baryons). If just 50% of the energy contributes to thrust, we need 8/9 of the ship mass per year. Over 30 years, a single kilogram of ship mass will need 10
28 kg of matter/antimatter, or 10
45 J. That exceeds the energy released in a typical supernova by an order of magnitude and violates the rule that
supernovae are always stronger.
You can scoop up matter on the way and use that as propellant - but that won't work beyond some critical (engineering-dependent) speed as scooping up the particles will slow the ship more than expelling it can accelerate it.