These are chat archives for symengine/symengine
thread_pool) the compiler does something differently which causes the slowdown
malloc()implementation and recovers the lost performance
thread_poolis called, some compiler optimizations go off and we get a slowdown.
tcmallocetc were not consistent, ignore all my previous comments
thread_poolis the problem right?
thread_poolonly if needed?
thread_poolis not included but some other piece of threading functionality is included in symengine
poly_muland see if the slowdown is back even without
piranha.hppI see around a 5% more cost in
callgrindto see the functions being called on both cases
mp_integer.hppit's back to the good commit