On Sunday, September 9, 2018 at 4:50:38 AM UTC+2, Bekket McClane wrote: > > Hi, > > I'm currently studying some graph reduction passes. > It seems that LoopPeeling is only used when certain flag is turn on. I'm > curious about the status and usage of this pass >
The pass is on by default. > > Also, though the algorithm does the eligibility check, it doesn't do any > suitability check - it just peels the first iteration anyway. To me, this > doesn't make much sense, since force peeling the first iteration not always > give you performance improvement. Is there any specific reasons for doing > this? > Right, we just have not done the work of figuring out a suitable heuristic to enable loop peeling. We had some rough ideas, but we have never tried them. Do you know of any good heuristics for loop peeling (or literature where it is discussed)? Cheers, Jaro > > Thank you. > > Bekket > -- -- v8-dev mailing list v8-dev@googlegroups.com http://groups.google.com/group/v8-dev --- You received this message because you are subscribed to the Google Groups "v8-dev" group. To unsubscribe from this group and stop receiving emails from it, send an email to v8-dev+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
