You’re right, and it was amazing to me when I went through the basic numbers and saw just how much more efficient it was. The Orion wikihas the equation for calculating Isp. Plugging in their number of 0.5 for collimation, and Dyson’s own number for debris velocity of 3000 km/s, yields an Isp of around 150,000 s^-1. (Admittedly, this is derived from the energy in a deuterium fusion reaction.) The Isp for large rocket engines like the retired Space Shuttle Main Engine is in the 350-450 s^-1 range. You still have to build the bombs for an Orion though, and I wonder what the cost of that is compared to liquefying hydrogen and oxygen?
There have been several plans for exotic modifications of the scheme since then. I think building a 2 km ring out of ceramic semiconductor is going to be a bit difficult.
The thing about the Orion nukes is that unlike a bomb that has to be delivered, for the Orion scheme the size and weight of the nuclear charges didn’t matter, at least within very broad limits, given that they were talking about ships with a gross weight of thousands or tens of thousands of tonnes. Nor was yield critical, given that the whole scheme revolved around coping with almost more yield than could be withstood. The main design focus of the Orion nukes was obtaining a useful explosion from the absolute minimum amount of fissionable material. All other components of the charges- depleted uranium, lithium deuteride, beryllium, etc.- would be orders of magnitude cheaper.
In general, your standard fission implosion design depends on two things: compressing the fissionable material to the smallest possible volume, and then keeping it from blowing itself apart before the chain reactions have had a chance to occur. The first is done by the explosive charges that compress the core- the more explosive you can pack on, the higher a density you can achieve. And then the core is surrounded by what’s called a “tamper”, a shell of dense material that by simple inertia gets in the way of the nascent explosion and holds it in for a few extra microseconds, allowing more chain reaction generations to occur.
For Orion, you could use both a charge of explosive and a tamper way too large and heavy to be practical for a bomb or warhead, but ideal to enable each charge to use the smallest possible amount of fissionable material. And the extra weight was actually useful, since the design goal was to convert as much of the bomb’s energy as possible into kinetic energy in the form of a hot fast moving plasma, and the extra mass facilitated this.