In the same week, I read two comments about accuracy of userspace timing and was struck by the juxtaposition.
0. Jake's coverage about CPU Isolation: "Accounting for CPU time without the presence of the timer tick is one of the areas that needs work. Users still want to see things like load averages reflect the time being spent in user-space processing on an isolated CPU, but that information is normally updated in the timer tick interrupt. In order to isolate the CPU, though, the timer tick needs to be turned off."
1. Engadget's review of the new LG Nexus 4 handset at http://www.engadget.com/2012/11/02/nexus-4-review/: "we fully expected the phone to be even faster than the Optimus G, mainly because LG and Google have had the opportunity to make sure Android 4.2 is fully optimized with the manufacturer's hardware, and the lack of custom skin should theoretically keep everything running efficiently. The second concern is in the benchmarks we ran. . . . Since the Nexus 4 and the Optimus G are so similar in their chipsets and other components, the two's metrics should be easily comparable -- or at least in the same neighborhood as each other. But as you can see in the table above, some of the numbers are the complete opposite of what we expected. In fact, some of these results (most notably, Quadrant and Vellamo) are even lower than what we typically get out of dual-core Snapdragon S4 processors."
Hmmmm! Coincidence or no? Might some clever young Android engineer have optimized 4.2 for the flagship Nexus 4 a bit prematurely with CPU Isolation? I'm curious to hear the verdict of you real-time experts.