There are a lot of variables that affect a full page refresh...caching, etc. I could not get those ms numbers in any scenario, not sure what I should be looking at.
Would be more useful to stress-test the lib by implementing dbmonster [1] and speedtest* [2]. These perf tests have the benefit of being good visual comparisons and also stressing the browser's GC to evaluate how efficient a lib is at minimizing overall mem usage and reducing jank.
I've been working on a DOM view layer lib [3] that does exceptionally well in both of those. (more demos and docs coming soon).
EDIT: added domvm dbmonster demo [4]
*speedtest is ultra-slow in FF for all libs, the winner is surprisingly innerHTML. still need to narrow down the cause to open an FF bug.
[1] http://joelrich.github.io/fastest-dbmonster/
[2] http://rawgit.com/crysalead-js/dom-layer/master/examples/spe...