Off thread full parsing enables the browser to perform full bytecode parsing of specific JS scripts off the main thread instead of parsing them on the main thread during execution.
The recent nightly experiment on Full Parsing (bug 1722551) revealed that off thread full parsing can reduce the amount of time spent executing JS by 10-15% which translates to a median improvement of about 2% in page load time across all segments.
An interest result from the experiment also showed that in the low cpu segment, enabling full parsing can attain an improvement in page load time of up to 10% and a reduction in JS execution time by up to 20%.
There is a trade off in memory for this improvement, however. The experiment showed a 1% regression in the median for total memory, and a 4% regression in the worst case above the 80th percentile.
In general, we believe the risk is low and the trade off in memory is acceptable given the significant improvements in page load time and JS execution time, especially for users with only 1-2 cores.