A gram of antimatter reacting with a gram of matter releases Hiroshima-like energy, but is that true in practice? My understanding is that a particle must meet its antiparticle. So if you a one-gram cube of, say, anti-gold and shot it into a one-gram cube of normal gold, how would you get the particles to come in contact with the corresponding antiparticles? Would you perhaps try to vaporize both cubes using heat and then mix the hot gas together?
If you actually tried compressing the matter into the antimatter like in a Big Boy implosion device, is it not possible that the exterior layers of the cube would react and destroy the assembly before the antimatter inside the the cube found their antiparticle?
 A: Aside from the issues of handling antimatter (it's hard).

*

*Little Boy released around 63 TJ

*1g of matter has an energy of around 90 TJ

*1g of antimatter reacting with 1g of standard matter would be two grams, 180TJ

*0.35g of antimatter should therefor be enough

Getting masses that react quickly with each other to still get as close as possiblem to each other is a problem every engineer constructing a nuclear bomb faces. You don't want it to start reacting too early (that's why the masses were quite "far apart" in the gun type little boy bomb). Even the microseconds it takes to get the rod into the ball are a problem because once the chain reaction start it's getting really messy really quickly... and it want's to push itself appart quite violently (that's what you want in the end but not too early!)
But with antimatter you're "better off". Because you don't need gold to react with anti-gold. Any neutron will react with any anti neutron. So if after inital contact your gold cubes race away from each other, the anti-gold will react with anything its meeting, also air molecules. So your gold cube will at most travel a few meters until it is completely anihilated and you're left with a rapidly expanding sphere of superheated plasma.
