According to razvan (the author of this patch) "The final distribution across buckets seems to be at least as good as that of the jenkins hash. I'll keep looking into making other improvements to fq_codel. I wanted to let you know I haven't given up on this. Thank you for your support." Another kvetch of the current hashing process I have is the ipaddr[0] ^ ipaddr[1] ^ ipaddr[2] ^ ipaddr[3] in prehash of ipv6. But really, measuring, and profiling, makes more sense before proceeding far down this path. ---------- Forwarded message ---------- From: Dave Taht Date: Tue, Mar 24, 2015 at 9:53 AM Subject: dear mr monty carlo To: Paul McKenney so how would you go about trying out a different hash function in the kernel? Got this teeny "xxhash" patch from a contributor, who says it´s faster than jhash, and the benchmark says itś gooder than jhash - and is too shy to submit it for mainline kernel devs. https://code.google.com/p/xxhash/ I'll slam it into my tree, but I imagine there are more comprehensive ways to abuse and evaluate it on your side of the house.... -- Dave Täht Let's make wifi fast, less jittery and reliable again! https://plus.google.com/u/0/107942175615993706558/posts/TVX3o84jjmb -- Dave Täht Open Networking needs **Open Source Hardware** https://plus.google.com/u/0/+EricRaymond/posts/JqxCe2pFr67