Interested in these recent Reaper/Sonar comparisons, I tested the latest Reaper against Sonar X3d.
I used Sonar's TL-64 Tube Leveler plugin to test, since it strikes me as a pretty plain vanilla processor without a lot of weird tricks going on.
I put 64 instances in Reaper and in Sonar - 8 tracks of audio each with 8 instances. I turned Reaper's "Anticipative FX processing" off, and Sonar's "Always stream Audio through FX" off (not that that should make any difference), with ThreadSchedulingModel in Sonar set to both 1 and 2. The computer is a i7 3770 with an RME interface at 1024 samples.
I watched the Windows Task Manager as I played back in each program. In Reaper the CPU usage hovered between 10 and 12 percent, while in Sonar it hovered between 10 and 16 percent, with a spike to 20 percent immediately on playback start.
It comes as no surprise to me that Sonar's CPU usage would be higher when performing the same task, because Sonar in general feels like a program with more overhead. For one small example,the Reaper fx window hearkens back to Windows 98, while Sonar uses more pretty graphics.
Sonar is like a Subaru station wagon. Reaper is a BMW diesel retrofitted to use biofuels. To worry that a Subaru gets worse gas mileage than the biofuel-BMW is to ignore that the driving experience of the two cars is entirely different.