So I fired up the old CRJ at UK2000's Manchester Extreme and hit Ctrl-Z to have a look at the framerate.
It was awful. Single figures awful, and that was just in a cold & dark cockpit staring at a terminal wall. Panning the view from side to side was mega jerky.
Now I admit I have the vast majority of sliders pushed to the right, and I know FSX's reputation. But by god, I'm running a 4.5GHz quad-core here, with a mid-range graphics card so a six year old game ought to be running a hell of a lot smoother than that.
Then I had an idea.
When I first built the flightsim PC, I was pleasantly surprised to discover that adding the "proper" graphics card didn't disable the onboard Intel graphics adapter on the motherboard. Since I'm a big fan of multi-monitor (for all things not just flightsimming) I used one of its outputs to give me a third 17" monitor display. I generally only used it for very graphics-lite functions like the ATC window or the GPS window. Surely that wouldn't interfere with the main graphics display.
On a hunch I disabled the on-board graphics adapter in the BIOS and fired up FSX back at Manchester Xtreme. Framerates were now steadily in the 19-20 range and panning was smooth as silk.
So as I understand it, the problem is right down at the hardware level. Having a slower performing graphics card working alongside a higher performing one will drag the performance of the faster card down to the level of the slower one. Even though my cockpit display was running on the faster NVidia card, it was effectively limited to the performance of the onboard Intel graphics. Hence very disappointing frame rates.
I really should have known better.
However that doesn't change the fact that I'm getting framerates that are really no better than OK on a bang up to date overclocked box running a six year old game. Just how inefficient must the FSX code be for this sort of performance?