On 5/17/05, Hubert Chan <[EMAIL PROTECTED]> wrote: > Thanks, Ryan, for your quick reply. I suspected that the correct thing > to do would be to just accept the performance hit (although 15% seems > pretty significant). I'll wait to see if anyone else has suggestions. > Otherwise, I'll just compile using normal libgc, and talk with my > upstream. > > By the way, have you heard of any other comments about multi-threaded > libgc being slower than single-threaded?
I'd be surprised if it's that bad under NPTL, and if it is, I'd be surprised if it can't be substantially improved (at least on x86) with a little bit of oprofile work. When was that performance comparison done? Cheers, - Michael
