laforge has posted comments on this change. ( https://gerrit.osmocom.org/c/libosmo-abis/+/19070 )
Change subject: move flip_buf_bits() from DAHDI to a shared flip.c ...................................................................... Patch Set 5: I finally did some benchmarking: reversing the bit-order in a 1 Mbyte sized buffer. * on a core-i7, the flip (lookup table) always seems to win, no matter how many iterations of buffer re-ordering: 1 iterations: flip=15396248, revb=19484348; revb = 126 % 10 iterations: flip=69081928, revb=162441630; revb = 235 % 100 iterations: flip=311541169, revb=580125967; revb = 186 % 1000 iterations: flip=2430175640, revb=5823932997; revb = 239 % * on an atom, the flip (lookup table) seems to win at a lower margin for all workloads except a single flip: 1 iterations: flip=20162864, revb=18311024; revb = 90 % 10 iterations: flip=150938160, revb=217883200; revb = 144 % 100 iterations: flip=1509970264, revb=1818844456; revb = 120 % 1000 iterations: flip=14332313768, revb=17541548328; revb = 122 % Now I'm a bit puzzled. Will try to measure with more realistic buffer sizes (100...1000 bytes) -- To view, visit https://gerrit.osmocom.org/c/libosmo-abis/+/19070 To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings Gerrit-Project: libosmo-abis Gerrit-Branch: master Gerrit-Change-Id: I8c6dc3eea921a73ebe65540aad0f0c0584d47e5a Gerrit-Change-Number: 19070 Gerrit-PatchSet: 5 Gerrit-Owner: laforge <[email protected]> Gerrit-Reviewer: Jenkins Builder Gerrit-Reviewer: laforge <[email protected]> Gerrit-CC: tnt <[email protected]> Gerrit-Comment-Date: Sun, 02 Aug 2020 07:53:36 +0000 Gerrit-HasComments: No Gerrit-Has-Labels: No Gerrit-MessageType: comment
