Coarsely speaking, if you're just looking at the feed required to grow a certain biomass of fish, in salmon aquaculture net-pen systems it's around 1.3 kilograms. Sometimes they do better. Sometimes, when things go badly, it's higher.
My recollection on beef productions is like yours. It's much higher. There are different reasons for that, but I would caution you against using a feed-to-growth ratio as a hard measure of overall ecological performance, because the diets are very different in these animals. For example, the moisture content--colleagues who may be speaking to you later might have a better sense of this--in a concentrated pellet feed might only be in the single digits, whereas the roughage and silage you feed to cattle have a much higher moisture content. You're trading off a lot of water in one case for a concentrated, high-nutrition, high-density feed. It is a poor measure of efficiency. These animals are also completely different, in that warm-blooded animals have to maintain their temperatures above background; fish don't.
Let me get to your core question on whether farmed salmon systems are higher-performing systems in terms of efficiency. You always have to think about what you are measuring when you talk about efficiency. If it's in terms of industrial energy inputs, then yes, they are a higher-performing system than terrestrial livestock, cattle in particular, but they are not so much better than chicken. Chicken is actually a very high-performance system that is very comparable to farmed salmon. If you are talking about greenhouse gas emissions, farmed salmon production is again much higher-performing than beef. Roughly speaking, let's put it in this way: if you produce a tonne of farmed salmon, you release, from a life-cycle perspective, about two tonnes of carbon dioxide equivalent. It's about 10 to 14 tonnes of carbon dioxide equivalent for beef.