cordic went from 10 ms to 13 ms on arewefastyet. fannkuch went from 46 ms to 51 ms. It got even slower after that period.
Is this the ropes stuff maybe?
Seems unlikely, given the two affected tests, and the fact that these regressions are only on TM; ropes should affect TM/JM about equally.
There are some huge losses on V8 as well. Sayrer, can you investigate this? I am happy to work on fixes if you can point me to the changesets that did this.
where is the data?
I'm going to go out on a limb and say that this bug isn't likely to go anywhere at this point.