staffantj

Apologies, I typo'd the subject, hopefully fixed now. It occurs at, for instance, in master: https://github.com/google/benchmark/blob/4528c76b718acc9b57956f63069c699ae21edcab/include/benchmark/benchmark.h#L337 There's no compiler error / warning message that I've seen, merely the documentation on...

This leaves an error for me. From the command line: ``` $ python -c 'import riptable as rt;tmp1 = rt.Dataset({'f': [1,1,2,2], 'g': list('abcd'), 'h': [5,6,7,8]});rt.putmask(tmp1['h'], tmp1.f == 1, tmp1.groupby(['f'])['h'].shift(-1))' Traceback...

It appears that the python layer calls in with the REDUCE_MAX function number (202). regardless of the data set in use. Thus the NaN detection logic is not in play,...

The failing AVX2 specializations might be better off as constexpr's - not sure if that will work out, but could be worth trying.

One attempt at this can be found at https://github.com/Quansight/pnumpy/commit/d13448f5203a7105e510d0fdfbba300013a8adb3 . This still has the fundamental problem of assuming that volatile variables (even when manipulated "atomically") participate in any kind of...