Ok, did a test run today. The setup was 100 Magmatic Dynamos in ten vertical rows of ten (100 x 80rf/t for 8000rf/t) powering 15 Redstone Furnaces (15rf/t per) and 15 Pulverisers (40rf/t per) processing cobble into stone, then the stone into cobble, etc. I made two of this setup, each in it's own chunk with no differences except one would only use Heavy Aluminium Wire (HAW) and the other would use Cryo-stabilised Fluxduct (CSF). On the HAW setup, I used two sealable HAW blocks (one where all power would go to from the Dynamos, the other where all power would go into for the machines) to simulate typical use of two sealed modules on a space station level, assuming seperate modules for power generation and resource processiong. All of this would take place inside a Void dimension (The Last Millennium dimension from the Extra Utilities mod) as to remove any impact any other TE or entities would have in the results.
Picture of setup:
http://i.imgur.com/bl2eTWc.jpg
This was done on a server running Forge 1566, Fastcraft 1.22 test 18 and Better FPS 1.1.1. The only other entity when results with Opis were taken was the player, which was in its own chunk next to the two setups. As TD ducts don't show as Tile Entities, I decided to go off chunk timing instead. I was actually surprised by the results. I admit, I did go in expecting the TD fluxducts to perform better than the GC wiring.
TD chunk with machines active: 365 microseconds/tick
GC chunk with machines active 169 microseconds/tick
TD chunk with machines not active 195 microseconds/tick
GC chunk with machines not active 95 microseconds/tick
Looking at those results, there's an obviously large relative difference in performance going on. On chunk timing basis, the GC wire is clearly better in terms of performance. The only weird anomaly I found is that one of the dynamos in the GC cluster would have a major timing difference compared to all the others, sitting at anything from 150-170 microseconds in entity timing. On server reloads, it would be a different dynamo too. Not sure as to the cause of that.
I'll be doing some more testing, this time with more realistic in game scenarios as no one would bother to make 100 RF generating blocks which only make 8000RF combined (IIRC a fully augmented dynamo can generate about 665RF/t). Also I think having a capacitor/storage/buffer might increase performance, but I could well be wrong. At the least it looks like I now have a new go to wire/duct/conduit for RF.
EDIT: Putting in a Resonant Energy Cell (from Thermal Expansion) between the Dynamos and Machines really improved things for the Fluxducts.
TD chunk with machines active with REC: 220 microseconds/tick
GC chunk with machines active with REC: 153 microseconds/tick
Not a lot of improvement for the GC chunk, but damn that brought the TD chunk down to levels near what was seen with the machines being inactive. Having 30 machines demand power from one source, and one source demanding power from 100 instead of 30 machines pulling from 100 separate really improved their performance.