Here is a pair of F18T8 lamps running in my Lithonia 2' Shoplight. At one time it had a trigger start ballast, but it now contains a pair of preheat ballasts. These being krypton lamps also results in them running dim with striations until they have a chance to warm up. By comparison, F17T8 lamps light bright immediately, but draw a couple watts less, and should be a bit dimmer in the end. The difference isn't substantial, however. A potential issue with F17 is that the lower resistance cathodes might not be preheated adequately, and this could result in premature failure.
I also tested out the F18T8s on programmed start electronic. On the electronic HPF ballast, the F18 will use less power. It was also low ballast factor, and only used about 8W. I'm not sure if that's low enough for it to sputter, but it was quite dim. Another issue was that these ballasts may not preheat F18s sufficiently, and the ends would not glow orange during the heating phase like the F17s. It seems counter intuitive that both ballasts cannot properly heat the cathodes of the other. It makes sense though if you think about it. The the programmed start ballast uses low voltage for cathode heating, which dissipates enough heat when paired with the low resistance F17T8 but not the higher resistance F18T8. The preheat ballast on the other hand applies higher line voltage, which would normally also heat the the lower resistance filament to a greater degree, but doesn't due to the cathodes being in series with the ballast. As a result of the ballast limiting current, the higher resistance cathodes actually get hotter, much like a lower wattage incandescent bulb would glow brighter than a higher wattage bulb in the same series circuit.
In addition I tested the F18T8s on an HPF magnetic ballast. I'd expect this ballast to drive the lamps properly as long as it can start them, which is no guarantee. It should be acceptable considering it is rated to run shorter lamps. It turns out it did successfully fire up the lamps every time. Even so, the power factor was lower with the 18W lamps, initially around 70% and eventually reaching 80%, only after the heat in the room clicked on which helped further warm up the tubes. I'm not sure if the lower power factor would stress the ballast in the long term.